Communication method, and apparatus

Abstract

A communication method and an apparatus. In the method, a terminal obtains QoE measurement configuration information, and sends the QoE measurement configuration information, the QoE measurement configuration information being associated with a first AI task; on the basis of the QoE measurement configuration information, an AI node obtains QoE measurement result information, wherein the QoE measurement result information can be used for determining a target AI node, and the target AI node is used for executing the first AI task; that is to say, the terminal triggers the AI node to perform QoE measurement for the first AI task, wherein the QoE measurement result information of the AI node can represent the QoE of the service of the AI node for providing the first AI task, and the target AI node determined on the basis of the QoE measurement result information can provide the terminal with an AI service having higher QoE.

Description

Communication method and apparatus

[0001] Cross-reference to Related Applications

[0002] This application claims priority to the Chinese Patent Application No. 202510213593.8, filed on February 25, 2025, and entitled "A Communication Method and Apparatus", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0003] Embodiments of the present application relate to the field of wireless communication, and in particular, to a communication method and apparatus. BACKGROUND

[0004] Artificial intelligence (AI) is a technology that performs complex calculations by simulating the human brain. With the improvement of data storage and computing power, AI technology has been increasingly used. The 3rd Generation Partnership Project (3GPP) proposes to apply AI technology to communication systems to improve network performance and user experience.

[0005] AI nodes are included in a wireless communication network, which can be, for example, a core network (CN) node deployed with an AI module, a radio access network (RAN) node deployed with an AI module, or other network elements (such as an AI server) for performing AI tasks. An AI node can provide AI services for a terminal, for example, a terminal submits an AI task to an AI node, the AI node performs the AI task to obtain a result, and then returns the result to the terminal. Further, the terminal can send a request for AI services to a RAN node, and the RAN node selects an AI node to provide AI services for the terminal based on the hardware conditions (such as computing resources and storage resources) of the AI node, and sends the request to the selected AI node.

[0006] Currently, the way of selecting an AI node cannot provide a terminal with AI services of high quality of experience (QoE). SUMMARY

[0007] The present application provides a communication method and apparatus, which enables the selected AI node to provide a terminal with AI services of high quality of experience (QoE).

[0008] In a first aspect, embodiments of this application provide a communication method that can be applied to the terminal side, such as a terminal or a communication module and / or computing module in the terminal, or a circuit or chip in the terminal responsible for communication functions (such as a modem chip, also known as a baseband chip, or a system-on-chip (SoC) chip containing a modem core or a system-in-package (SIP) chip), or a circuit or chip in the terminal responsible for communication and / or computing functions (such as a graphics processing unit (GPU), an AI processor, or an application-specific integrated circuit (ASIC)), or a logic node, logic module, or software that can implement all or part of the terminal functions.

[0009] Taking the application of this method to a terminal as an example, in this method, the terminal obtains QoE measurement configuration information, which is associated with the first AI task. The QoE measurement configuration information is used to obtain QoE measurement result information, which is used to determine the target AI node. The target AI node is used to execute the first AI task. The terminal sends the QoE measurement configuration information.

[0010] In this process, the terminal obtains QoE measurement configuration information, for example, the terminal generates QoE measurement configuration information.

[0011] For example, the terminal sends QoE measurement configuration information to the AI ​​node.

[0012] In the above technical solution, the terminal triggers the AI ​​node to perform QoE measurement to obtain QoE measurement result information. The QoE measurement result information of the AI ​​node can characterize the QoE of the service provided by the AI ​​node. The target AI node determined based on the QoE measurement result information can provide the terminal with AI services with higher QoE.

[0013] In one possible design, the terminal also receives QoE measurement results and determines the target AI node based on these results.

[0014] In the above technical solution, the terminal determines the target AI node based on the QoE measurement results, and the target AI node can provide the terminal with AI services with a higher QoE.

[0015] In one possible design, the terminal also receives switching information, indicating a switch to a network node used to manage the target AI node. This network node could be, for example, an access network node or a core network node.

[0016] In the above technical solution, when the network node accessed by the terminal (denoted as the first network node) is not the network node used to manage the target AI node (denoted as the second network node), the terminal can also receive switching information and switch to the second network node according to the switching information. This allows the terminal to send the traffic of the first AI task to the second network node, which then forwards the traffic of the first AI task to the target AI node. This helps reduce the number of times the traffic of the first AI task is forwarded and lowers transmission latency.

[0017] In one possible design, QoE measurement configuration information is used to indicate the model for the first AI task, and the QoE measurement configuration information is carried in a signaling radio bearer (SRB) or a data radio bearer (DRB).

[0018] In the above technical solution, the QoE measurement configuration information is used to indicate the model of the first AI task, so that the AI ​​node can obtain the QoE measurement result information based on the model of the first AI task, which facilitates the selection of subsequent target AI nodes.

[0019] For example, when the QoE measurement configuration information includes a model of the first AI task, the QoE measurement configuration information is carried in the DRB. When the QoE measurement configuration information includes a model description of the first AI task, the QoE measurement configuration information is carried in the SRB.

[0020] In the above technical solutions, using different methods to carry QoE measurement configuration information with different data volumes helps to improve transmission reliability and efficiency.

[0021] In one possible design, the QoE measurement results are the measurement results of a second AI task, which serves as a reference task for the first AI task. For example, the second AI task is an AI task that is being executed or has already been executed on an AI node.

[0022] In the above technical solution, QoE measurement result information is obtained by measuring the reference task of the first AI task (i.e., the second AI task) without having to start the measurement task of the first AI task, which helps to reduce the workload of AI nodes.

[0023] In one possible design, the first AI task and the second AI task are of the same type.

[0024] For example, both the first AI task and the second AI task belong to training tasks, both the first AI task and the second AI task belong to inference tasks, or both the first AI task and the second AI task belong to data acquisition tasks.

[0025] In the above technical solution, although the QoE measurement result is the result of the second AI task, it can accurately characterize the QoE of the AI ​​node when performing the first AI task. Therefore, the identified target AI node can provide the terminal with AI services that offer higher QoE.

[0026] Secondly, embodiments of this application provide a communication method that can be applied to the AI ​​node side, such as an AI node or a module (e.g., a circuit, chip, or chip system) in an AI node, or a logic node, logic module, or software that can implement all or part of the functions of an AI node, or a circuit or chip (e.g., a GPU, AI processor, or ASIC) in an AI node that is responsible for communication and / or computing functions.

[0027] Taking an AI node as an example, in this method, the AI ​​node receives QoE measurement configuration information, which is associated with the first AI task. Based on the QoE measurement configuration information, the AI ​​node obtains QoE measurement result information, which is used to determine the target AI node, and the target AI node is used to execute the first AI task. The AI ​​node then sends the QoE measurement result information.

[0028] For example, the AI ​​node sends QoE measurement result information to the RAN node.

[0029] In one possible design, the QoE measurement result information is the measurement result of the second AI task, which serves as a reference task for the first AI task. In another possible design, the first and second AI tasks belong to the same type of task.

[0030] Thirdly, embodiments of this application provide a communication method that can be applied to the network side, such as RAN nodes on the network side, modules (e.g., circuits, chips, or chip systems) in RAN nodes, or logic nodes, logic modules, or software that can implement all or part of the functions of RAN nodes, or circuits or chips (e.g., GPUs, AI processors, or ASICs) in RAN nodes that are responsible for communication and / or computing functions.

[0031] Taking the application of this method to a RAN node as an example, in this method, the RAN node receives QoE measurement result information. This QoE measurement result information is obtained by the AI ​​node based on the QoE measurement configuration information from the terminal, and the QoE measurement configuration information is associated with the first AI task. Based on the QoE measurement result information, the RAN node determines the target AI node, which is used to execute the first AI task.

[0032] In the above technical solution, the terminal triggers the AI ​​node to perform QoE measurement to obtain QoE measurement result information. The QoE measurement result information of the AI ​​node can characterize the QoE of the service provided by the AI ​​node. Based on the QoE measurement result information, the RAN node determines the target AI node that can provide the terminal with AI services with a higher QoE.

[0033] In one possible design, the RAN node receives the traffic for the terminal's first AI task. The RAN node then sends the traffic for the first AI task to the target AI node.

[0034] In the above technical solution, the RAN node determines the target AI node on its own. Therefore, when the RAN node forwards the traffic of the first AI task to the target AI node, it does not need to parse the messages from the terminal, which helps to better adapt to the current communication network.

[0035] In one possible design, when the RAN node determines the target AI node based on QoE measurement results, specifically, the RAN node determines multiple target AI nodes and their priority order based on the QoE measurement results. When the RAN node sends the traffic for the first AI task to the target AI nodes, specifically, the RAN node sends the traffic for the first AI task to the multiple target AI nodes based on the multiple target AI nodes, their priority order, and their load.

[0036] In the above technical solution, the RAN node determines multiple target AI nodes and their priority order. When forwarding the traffic of the first AI task, it forwards the traffic based on the priority order and load of the multiple target AI nodes. In this way, the target AI node that receives the traffic of the first AI task can provide the terminal with AI services with a higher QoE.

[0037] In one possible design, the RAN node also sends handover information to the terminal, instructing it to switch to the network node used to manage the target AI node. For example, when the RAN node determines that it is not the RAN node used to manage the target AI node, it can also send handover information to the terminal, instructing the terminal to switch to the RAN node used to manage the target AI node.

[0038] The above technical solution helps to reduce the number of traffic forwardings for the first AI task and reduce transmission latency.

[0039] Fourthly, this application provides a communication device that has the functions of the first aspect described above. For example, the communication device includes modules, units, or means that perform the operations involved in the first aspect. These modules, units, or means can be implemented by software, hardware, or a combination of software and hardware.

[0040] Fifthly, this application provides a communication device that has the functions of the second aspect above. For example, the communication device includes a module, unit, or means for performing the operations involved in the second aspect above. The module, unit, or means can be implemented by software, hardware, or a combination of software and hardware.

[0041] Sixthly, this application provides a communication device that has the functions of the third aspect above. For example, the communication device includes a module, unit, or means for performing the operations involved in the third aspect above. The module, unit, or means can be implemented by software, hardware, or a combination of software and hardware.

[0042] In a seventh aspect, this application provides a communication device including an interface circuit and one or more processors. The one or more processors are coupled to a memory. The memory stores part or all of the necessary computer program or instructions for implementing the functions described in the first aspect. The one or more processors are executable to carry out the computer program or instructions, causing the communication device to implement the methods in any possible design or implementation of the first aspect. The interface circuit is used to implement the communication functions within the communication device and / or the communication functions between the communication device and other devices or components.

[0043] In one possible design, the processor is used to communicate with other devices or components through the interface circuit.

[0044] In one possible design, the communication device may also include the memory.

[0045] The aforementioned communication device may be a terminal, or a communication and / or computing module in a terminal, or a chip in a terminal responsible for communication functions such as a modem chip (also known as a baseband chip) or a SoC or SIP chip containing a modem module, or a circuit or chip in a terminal responsible for communication and / or computing functions (such as a GPU, AI processor, or ASIC), or a logical node or logical module capable of implementing all or part of the terminal functions.

[0046] Eighthly, this application provides a communication device including an interface circuit and one or more processors. The one or more processors are coupled to a memory. The memory stores part or all of the necessary computer program or instructions for implementing the functions described in the second aspect above. The one or more processors are executable to carry out the computer program or instructions, causing the communication device to implement the methods in any possible design or implementation of the second aspect above. The interface circuit is used to implement the communication functions within the communication device and / or the communication functions between the communication device and other devices or components.

[0047] In one possible design, the processor is used to communicate with other devices or components through the interface circuit.

[0048] In one possible design, the communication device may also include the memory.

[0049] The aforementioned communication device may be an AI node, or a module (such as a circuit, chip, or chip system) within an AI node, or a circuit or chip (such as a GPU, AI processor, or ASIC) within an AI node responsible for communication and / or computing functions, or a logical node or logical module capable of implementing all or part of the functions of an AI node.

[0050] Ninthly, this application provides a communication device including an interface circuit and one or more processors. The one or more processors are coupled to a memory. The memory stores part or all of the computer program or instructions necessary to implement the functions described in the third aspect above. The one or more processors are executable to carry out the computer program or instructions, causing the communication device to implement the methods in any possible design or implementation of the third aspect above. The interface circuit is used to implement the communication functions within the communication device and / or the communication functions between the communication device and other devices or components.

[0051] In one possible design, the processor is used to communicate with other devices or components through the interface circuit.

[0052] In one possible design, the communication device may also include the memory.

[0053] The aforementioned communication device may be a RAN node, or a module (e.g., a circuit, chip, or chip system) in a RAN node, or a circuit or chip (e.g., a GPU, AI processor, or ASIC) in a RAN node responsible for communication and / or computing functions, or a logical node or logical module capable of implementing all or part of the functions of a RAN node.

[0054] In a tenth aspect, this application provides a communication system including the communication device of the fourth aspect and the communication device of the fifth aspect described above. Optionally, it also includes the communication device of the sixth aspect described above.

[0055] Eleventhly, this application provides a computer-readable storage medium storing computer-readable instructions, which, when read and executed by a computer, cause the computer to perform any of the possible designs in the first to third aspects described above.

[0056] In a twelfth aspect, this application provides a computer program product that, when read and executed by a computer, causes the computer to perform any of the possible designs in the first to third aspects described above.

[0057] The technical effects that can be achieved by any of the second to twelfth aspects mentioned above can be referred to the description of the beneficial effects in the other aspects. Attached Figure Description

[0058] Figure 1 is a schematic diagram of a communication system architecture;

[0059] Figures 2 and 3 are schematic diagrams of possible application frameworks;

[0060] Figure 4 is a schematic diagram of a QoE measurement process;

[0061] Figures 5–10 are schematic flowcharts of the communication method provided in this application;

[0062] Figure 11 is a schematic diagram of the structure of a communication device provided in this application;

[0063] Figure 12 is a schematic diagram of the structure of a terminal provided in this application. Detailed Implementation

[0064] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is only for explaining specific embodiments and is not intended to limit the application. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0065] Figure 1 illustrates a possible, non-limiting system diagram. As shown in Figure 1, the communication system 10 includes a RAN 100 and a core network 200. RAN 100 includes at least one RAN node (110a and 110b in Figure 1, collectively referred to as 110) and at least one terminal (120a-120j in Figure 1, collectively referred to as 120). RAN 100 may also include other RAN nodes, such as wireless relay devices and / or wireless backhaul devices (not shown in Figure 1). Terminal 120 is wirelessly connected to RAN node 110. RAN node 110 is wirelessly or wired connected to core network 200. The core network node in core network 200 and RAN node 110 in RAN 100 can be different physical devices, or they can be the same physical device integrating core network logical functions and wireless access network logical functions.

[0066] RAN 100 can be a 3GPP-related cellular system, such as a 4G or 5G mobile communication system, or a future-oriented evolution system. RAN 100 can also be an open RAN (O-RAN or ORAN), a cloud RAN (CRAN), a virtualized RAN (vRAN), an artificial intelligence radio access network (AI RAN), or a wireless fidelity (WiFi) system. RAN 100 can also be a communication system that integrates two or more of the above systems.

[0067] RAN node 110, sometimes also referred to as access network node, access network equipment, RAN entity, or access node, is part of the communication system and assists terminals in achieving wireless access. Multiple RAN nodes 110 in communication system 10 can be of the same type or different types. In some scenarios, the roles of RAN node 110 and terminal 120 are relative. For example, network element 120i in Figure 1 can be a helicopter or drone, which can be configured as a mobile base station. For terminals 120j accessing RAN 100 through network element 120i, network element 120i is a base station; but for base station 110a, network element 120i is a terminal. RAN node 110 and terminal 120 are sometimes both referred to as communication devices. For example, network elements 110a and 110b in Figure 1 can be understood as communication devices with base station functions, and network elements 120a-120j can be understood as communication devices with terminal functions.

[0068] In one possible scenario, a RAN node can be a base station, an evolved NodeB (eNodeB), an access point (AP), a transmission reception point (TRP), a next-generation NodeB (gNB), a base station in a future mobile communication system, or an access node in a WiFi system. A RAN node can be a macro base station (as shown in Figure 1, 110a), a micro base station or indoor station (as shown in Figure 1, 110b), a relay node or donor node, or a radio controller in a CRAN scenario. Optionally, a RAN node can also be a server, wearable device, vehicle, or in-vehicle equipment. For example, in vehicle-to-everything (V2X) technology, the RAN node can be a roadside unit (RSU).

[0069] All or part of the functions of the RAN node in this application can also be implemented through software functions executed on hardware, or through virtualization functions instantiated on a platform (e.g., a cloud platform). The RAN node may also include communication modules, circuits, or chips that perform corresponding communication functions. The RAN node may also be configured with program instructions for performing corresponding communication functions and corresponding program instructions. The RAN node in this application can also be a logical node, logical module, or software capable of implementing all or part of the RAN node functions, or a circuit or chip (such as a graphics processing unit (GPU), artificial intelligence (AI) processor, or application-specific integrated circuit (ASIC)) responsible for communication and / or computing functions in an access node.

[0070] In another possible scenario, multiple RAN nodes collaborate to assist terminals in achieving wireless access, with different RAN nodes implementing some of the base station's functions. For example, RAN nodes can be central units (CUs), distributed units (DUs), CU-control plane (CPs), CU-user plane (UPs), or radio units (RUs). CUs and DUs can be separate entities or included in the same network element, such as a baseband unit (BBU). RUs can be included in radio frequency equipment or radio frequency units, such as remote radio units (RRUs), active antenna units (AAUs), or remote radio heads (RRHs). Furthermore, RAN nodes can also be computing units, providing computational power for tasks such as model inference and / or model training, and can also be used to implement one or more of the following: task partitioning, scheduling, and orchestration. The functionality of a computing unit can be implemented by a separate module independent of other units (e.g., CU, DU, RU), or by one or more other units (e.g., one or more of CU, DU, RU).

[0071] In different systems, CU (or CU-CP and CU-UP), DU, computing unit, or RU may have different names, but those skilled in the art will understand their meaning. For example, in an ORAN system, CU can also be called O-CU (open CU), DU can also be called O-DU, CU-CP can also be called O-CU-CP, CU-UP can also be called O-CU-UP, and RU can also be called O-RU. For ease of description, this application uses CU, CU-CP, CU-UP, DU, computing unit, and RU as examples. Any of the units among CU (or CU-CP, CU-UP), DU, computing unit, and RU in this application can be implemented through software modules, hardware modules, or a combination of software modules and hardware modules.

[0072] A terminal can be a device or module that accesses the aforementioned communication system and has corresponding communication functions. A terminal can also be called a terminal device, user equipment (UE), mobile station, mobile terminal, etc. Terminals can be widely used in various scenarios, such as device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), Internet of Things (IoT), virtual reality, augmented reality, industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart transportation, smart cities, etc. A terminal can be a mobile phone, tablet computer, computer with wireless transceiver capabilities, wearable device, vehicle, drone, helicopter, airplane, ship, robot, robotic arm, smart home device, transportation vehicle with wireless communication capabilities, communication module, etc. The embodiments of this application do not limit the device form of the terminal. The terminal typically contains communication modules, circuits, or chips that perform corresponding communication functions. Furthermore, it may also contain modules, circuits, or chips (such as GPUs, AI processors, or ASICs) that perform corresponding communication and / or computing functions. The terminal can also be configured with program instructions for performing these communication and / or computing functions.

[0073] To support AI technology in wireless networks, AI nodes may also be introduced into the network.

[0074] AI nodes can be deployed in one or more of the following locations within the communication system: RAN nodes, terminals, or core network nodes, etc. Alternatively, AI nodes can be deployed independently, for example, in a location other than any of the above-mentioned devices, such as in the host or cloud server of an over-the-top (OTT) system. AI nodes can communicate with other devices in the communication system, which can be one or more of the following: access network nodes, terminals, or core network nodes, etc.

[0075] It is understood that this application does not limit the number of AI nodes. For example, when there are multiple AI nodes, they can be divided based on function, such as different AI nodes being responsible for different functions.

[0076] It can also be understood that AI nodes can be independent devices, or they can be integrated into the same device to achieve different functions. Alternatively, they can be network elements in hardware devices, software functions executed on dedicated hardware, or virtualization functions instantiated on a platform (e.g., a cloud platform). This application does not limit the specific form of the aforementioned AI nodes.

[0077] AI nodes can be AI network elements or AI modules.

[0078] Figure 2 illustrates a possible application framework in a communication system. As shown in Figure 2, network elements in the communication system are connected via interfaces (e.g., NG, Xn) or over-the-air interfaces. These network element nodes, such as core network nodes, RAN nodes, terminals, or Operations Administration and Maintenance (OAM) nodes, may each contain one or more AI modules (only one is shown in Figure 2 for clarity). Access network nodes may function as standalone RAN nodes or may include multiple RAN nodes, such as CUs and DUs. CUs and / or DUs may also contain one or more AI modules. CUs may also be split into CU-CPs and CU-UPs, with one or more AI modules configured in each CU-CP and / or CU-UP.

[0079] AI modules are used to implement corresponding AI functions. AI modules deployed in different network elements can be the same or different. The models of AI modules can achieve different functions depending on the parameter configurations. The models of AI modules can be configured based on one or more of the following parameters: structural parameters (e.g., at least one of the following: number of neural network layers, neural network width, inter-layer connections, neuron weights, neuron activation function, or biases in the activation function), input parameters (e.g., the type and / or dimension of the input parameters), or output parameters (e.g., the type and / or dimension of the output parameters). The biases in the activation function can also be referred to as the biases of the neural network.

[0080] In one example, the neural network mentioned above can be a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), or a generative adversarial network (GAN).

[0081] Deep Neural Networks (DNNs) are artificial neural network architectures with multiple layers of nonlinear transformation units stacked in a hierarchical structure to form deep computational models. Compared to shallow neural networks, deep neural networks have more hidden layers, allowing the network model to capture more complex data structures and higher-level abstract features.

[0082] A CNN is a deep neural network with a convolutional structure. A CNN contains a feature extractor consisting of convolutional layers and subsampling layers. This feature extractor can be viewed as a filter, and the convolution process can be seen as performing convolution between a trainable filter and an input image or a convolutional feature map.

[0083] RNN is a type of recursive neural network that takes sequence data as input, recursively moves along the direction of sequence evolution, and connects all nodes (recurrent units) in a chain-like manner.

[0084] GAN is a deep learning model. It consists of a generator and a discriminator, and is trained through adversarial learning. Its purpose is to estimate the potential distribution of data samples and generate new data samples.

[0085] An AI module can have one or more models. A model can infer an output, which includes one or more parameters. The learning, training, or inference processes of different models can be deployed on different nodes or devices, or they can be deployed on the same node or device.

[0086] Figure 3 illustrates a possible application framework in a communication system. As shown in Figure 3, the communication system includes a RAN intelligent controller (RIC). For example, the RIC can be the AI ​​module shown in Figure 2, which is used to implement AI-related functions. RICs include near-real-time RICs (near-RT RICs) and non-real-time RICs (non-RT RICs).

[0087] Non-real-time RICs primarily handle non-real-time information, such as data that is not sensitive to latency, with latency on the order of seconds. Real-time RICs primarily handle near-real-time information, such as data that is relatively sensitive to latency, with latency on the order of tens of milliseconds.

[0088] Near real-time (NRT) RICs can be used for model training and inference. For example, they can be used to train AI models and then use those models for inference. NRT RICs can obtain network-side and / or terminal-side information from RAN nodes (e.g., CUs, CU-CPs, CU-UPs, DUs, compute nodes, and / or RUs) and / or terminals. This information can be used as training data or inference data. NRT RICs can deliver inference results to RAN nodes and / or terminals. Inference results can be exchanged between CUs and DUs, and / or between DUs and RUs. For example, a NRT RIC might deliver an inference result to a DU, which then forwards it to an RU.

[0089] Non-real-time RICs can also be used for model training and inference. For example, they can be used to train AI models and then use those models for inference. Non-real-time RICs can obtain network-side and / or terminal-side information from RAN nodes (e.g., CUs, CU-CPs, CU-UPs, DUs, compute nodes, and / or RUs) and / or terminals. This information can be used as training data or inference data, and the inference results can be delivered to RAN nodes and / or terminals. Inference results can be exchanged between CUs and DUs, and / or between DUs and RUs; for example, a non-real-time RIC delivers inference results to a DU, which then forwards them to an RU.

[0090] Near real-time RICs and non-real-time RICs can also be configured as separate network elements. Near real-time RICs and non-real-time RICs can also be part of other devices. For example, near real-time RICs can be set in RAN nodes (e.g., CU, DU), while non-real-time RICs can be set in OAM, servers, core network nodes, or other network nodes.

[0091] With the construction and development of networks, the operation mode of communication networks is gradually shifting from a single voice service to a multi-service model with multiple services running in parallel. For some streaming or voice services, such as streaming services and Multimedia Telephony Service for IP Multimedia Subsystem (MTSI) services, signal quality alone cannot reflect the user experience when using these services. To understand the user experience, a process can be initiated by the RAN node to measure the quality of experience (QoE) at the terminal (also known as application layer measurement). This allows for a better understanding of the user experience when using these services, thereby optimizing the network to improve the user experience.

[0092] In the QoE measurement process, the configuration information for the QoE measurement task (also known as QoE measurement configuration information) can be sent from the CN device or OAM node to the RAN node, which then sends this configuration information to the terminal. The terminal executes the QoE measurement task based on this configuration information, obtains the measurement results, and sends these results to the RAN node. The RAN node ultimately reports the measurement results to the QoE measurement collection node, which analyzes and evaluates these results to understand the user experience.

[0093] QoE measurement can be divided into signaling-based QoE measurement and management-based QoE measurement.

[0094] Among them, signaling-based QoE measurement refers to a QoE measurement task that is targeted at a specific terminal. For example, the core network node sends the configuration information of the signaling-based QoE measurement task to the RAN node through terminal-level signaling.

[0095] Managed QoE measurement refers to a QoE measurement task that is not targeted at a specific terminal. For example, the AI ​​node management (EM) node or OAM node sends the configuration information for the managed QoE measurement to the RAN node, and the RAN node selects a subset of terminals for QoE measurement based on the capabilities of the terminals currently connected to the RAN node and other information.

[0096] Figure 4 is a schematic diagram of a QoE measurement process. The QoE measurement process includes:

[0097] Step 401a or Step 401b: The RAN node obtains QoE measurement configuration information. In Step 401a, the RAN node receives a management-based QoE measurement request from the OAM node. In Step 401b, the RAN node receives a signaling-based QoE measurement request from the core network node. The QoE measurement request carries the QoE measurement configuration information.

[0098] Step 402: The RAN node sends the QoE measurement configuration information to the terminal's access stratum (AS).

[0099] For example, the RAN node sends QoE measurement configuration information to the terminal via radio resource control (RRC) messages. For example, the message also carries the service type corresponding to the QoE measurement configuration information.

[0100] Step 403: The terminal's access layer sends the QoE measurement configuration information received from the RAN node to the upper layer of the terminal's access layer.

[0101] The layer above the terminal's access layer can be the application layer or a layer that performs QoE measurement. Figure 4 shows the application layer as an example.

[0102] Optionally, the terminal's access layer can send QoE measurement configuration information to the upper layer of the access layer via a method called the attention command (AT).

[0103] Step 404: The upper layer of the terminal's access layer performs QoE measurement and obtains the QoE measurement result information.

[0104] Step 405: The upper layer of the terminal's access layer sends the QoE measurement result information to the terminal's access layer.

[0105] For example, the upper layer of the terminal's access layer periodically reports QoE measurement results.

[0106] For example, the upper layer of the terminal's access layer can report QoE measurement results only after a session has ended.

[0107] When the upper layer of the terminal's access layer needs to report QoE measurement results, the upper layer of the terminal's access layer sends the QoE measurement results to the terminal's access layer. The QoE measurement results can be reported in the form of a container.

[0108] Optionally, the upper layer of the access layer sends the QoE measurement results to the access layer via a method called the attention command (AT).

[0109] Step 406: The terminal's access layer sends the QoE measurement result information to the RAN node.

[0110] For example, QoE measurement result information may be carried in the uplink RRC message. The QoE measurement result information may be sent to the RAN node in a container form or in a non-container form.

[0111] Optionally, the RAN node that sends the QoE measurement configuration information may be different from the RAN node that receives the QoE measurement result information (e.g., the RAN node has been changed due to mobility), or it may be the same RAN node.

[0112] Step 407: The RAN node sends the QoE measurement result information to the measurement collection entity (MCE).

[0113] Due to the limited computing power of terminals, AI nodes in the network can assist in executing the AI ​​tasks required by the terminal. A network can include multiple AI nodes, and typically the RAN node selects an AI node to provide services to the terminal based on the resource availability (e.g., computing resources, storage resources) of the AI ​​nodes in the network. This selected AI node then assists in executing the AI ​​tasks required by the terminal. However, the AI ​​node selected in this way is unlikely to provide a high QoE (Quality of Service) for the terminal. Therefore, this application provides a communication method in which the terminal triggers an AI node to perform a QoE measurement, thereby selecting an AI node capable of providing a better QoE (Quality of Service) for the terminal.

[0114] It is understood that this application uses RAN nodes, AI nodes, and terminals as examples to illustrate the execution of the interaction, but this application does not limit the execution of the interaction.

[0115] For example, the method executed by the RAN node in this application can also be implemented by a module (e.g., a circuit, chip, or chip system) in the RAN node, or a logic node, logic module, or software that can implement all or part of the functions of the RAN node, or a circuit or chip (e.g., a GPU, AI processor, or ASIC) in the RAN node that is responsible for communication and / or computing functions.

[0116] The methods executed by the AI ​​node in this application can also be implemented by modules (such as circuits, chips, or chip systems) in the AI ​​node, or by logic nodes, logic modules, or software that can implement all or part of the functions of the AI ​​node, or by circuits or chips (such as GPUs, AI processors, or ASICs) in the AI ​​node that are responsible for communication and / or computing functions.

[0117] The method executed by the terminal in this application may also be implemented by a communication and / or computing module in the terminal, or a circuit or chip in the terminal responsible for communication and / or computing functions (such as a modem chip (also known as a baseband chip), or a SoC chip / SIP chip containing a modem core, or a GPU / AI processor / ASIC), or a logical node, logical module, or software that can implement all or part of the terminal functions.

[0118] Figure 5 is a flowchart illustrating a communication method exemplarily provided in this application, in which:

[0119] Step 501: The terminal obtains the first QoE measurement configuration information.

[0120] For example, the terminal's chip generates the first QoE measurement configuration information; or, the terminal's chip reads the first QoE measurement configuration information from the terminal's memory; or, the terminal requests the first QoE measurement configuration information from another device.

[0121] The first QoE measurement configuration information is associated with the first AI task. This can be understood as follows: when the terminal needs to deploy the first AI task, it can first obtain the first QoE measurement configuration information. This first QoE measurement configuration information is used to obtain the first QoE measurement result information, which is then used to select the target AI node. The target AI node is the AI ​​node that actually (finally / officially) executes the first AI task. In other words, the target AI node is used to provide the terminal with the service of the first AI task. For example, the target AI node can also send the results obtained from executing the first AI task to the terminal.

[0122] In this application, the target AI node that provides the service for the first AI task to the terminal can be referred to as the first target AI node.

[0123] For example, the first QoE measurement configuration information is used to indicate the model of the first AI task (hereinafter referred to as the model).

[0124] For example, the first QoE measurement configuration information includes a description of the model and / or the model itself.

[0125] For example, when the first QoE measurement configuration information includes a model, the model can be a lightweight model of the target model, and the QoE of the lightweight model can be used to characterize the QoE of the target model, where the target model is the model that is actually executed in the first target AI node.

[0126] For example, the model description may include at least one of the following: structural parameters of the model (e.g., number of neural network layers, neural network width, inter-layer connections, neuron weights, neuron activation functions, or biases in activation functions), input parameter descriptions (e.g., type and / or dimension of input parameters), or output parameter descriptions (e.g., type and / or dimension of output parameters).

[0127] For example, the first QoE measurement configuration information may also include one or more of the following: the type of the first AI task, the task identifier of the first AI task, the data required for QoE measurement of the first AI task (which can also be considered as the data of the first AI task), a description of the data of the first AI task, and the objective of the first AI task or the first QoE measurement parameters. The first QoE measurement parameters indicate which QoE parameters are included in the first QoE measurement result information. The description of the data of the first AI task may include, for example, data classification, data source, and data volume. The objective of the first AI task may include, for example, the accuracy requirements to be achieved by the trained model or the objective of data collection.

[0128] The following example further illustrates the configuration information for the first QoE measurement, taking into account different types of first AI tasks.

[0129] (1) When the first AI task is the training task:

[0130] The first QoE measurement configuration information includes a description of the initial model, a description of the data for the first AI task, the objective of the first AI task, and the first QoE measurement parameters. The description of the data for the first AI task includes descriptions of the training data and the test data. The objective of the first AI task is the required accuracy of the model. The first QoE measurement parameters include, for example, the training time, overhead, computational efficiency, and power consumption required to train the model. Training time may be the time consumed per epoch or the amount of data processed per unit time; training overhead may be the computational load required per epoch; and computational efficiency may be the computational load per unit time or the theoretical computational load per unit time. An epoch refers to training once using all samples in the training data; simply put, the epoch value is the number of times the entire training data is processed.

[0131] (2) When the first AI task is a reasoning task:

[0132] The first QoE measurement configuration information includes a description of the inference model and first QoE measurement parameters. These parameters may include, for example, inference accuracy, inference time, and throughput based on different types of input data. For example, the inference time may include the average inference time, the maximum inference time, and the time required to generate the first inference result. Throughput may be, for example, the amount of data processed per unit time.

[0133] (3) When the first AI task is a data acquisition task:

[0134] The first QoE measurement configuration information includes the target of data collection and the first QoE measurement parameters. These parameters include, for example, the diversity level of the collected data, the accuracy of the collected data, the similarity of the collected data, data labeling capability, data acquisition capability, and data processing capability. Specifically, the diversity level of the collected data is related to the data source; the similarity of the collected data refers to the presence of a large amount of duplicate data; the data labeling capability refers to the ability to accurately label the data; the data acquisition capability is the amount of data collected per unit time; and the data processing capability refers to the ability to extract high-quality data from low-quality datasets.

[0135] Step 502: The terminal sends the first QoE measurement configuration information.

[0136] In one possible implementation, the terminal sends first QoE measurement configuration information to a first RAN node. The first RAN node forwards the first QoE measurement configuration information to an AI node. Here, the AI ​​node is, for example, a candidate AI node, and there can be one or more candidate AI nodes. The first RAN node is, for example, the terminal's serving base station.

[0137] The first QoE measurement configuration information can be carried in either the SRB or the DRB.

[0138] For example, when the first QoE measurement configuration information includes the model and / or data of the first AI task, the first QoE measurement configuration information is carried in the DRB. For instance, when a terminal sends service data to a first RAN node, the service data is the first QoE measurement configuration information.

[0139] For example, when the first QoE measurement configuration information does not include the model and data of the first AI task, but includes, for example, a model description and / or a data description of the first AI task, the first QoE measurement configuration information is carried in the SRB. For example, the terminal sends an RRC message to the first RAN node, and the RRC message includes the first QoE measurement configuration information.

[0140] When the first RAN node forwards the first QoE measurement configuration information to the AI ​​node, there are two examples, such as the following:

[0141] In Example A, the first RAN node also obtains the identifier of the AI ​​node from the terminal, and forwards the first QoE measurement configuration information to the AI ​​node based on the AI ​​node's identifier. Prior to this, the terminal also obtains topology information from the first RAN node, and determines the identifier of the AI ​​node based on the topology information.

[0142] Example B: The first RAN node stores topology information. Based on the topology information, the first RAN node determines the identifier of the AI ​​node, and then forwards the first QoE measurement configuration information to the AI ​​node based on the identifier of the AI ​​node.

[0143] The topology information includes the identifiers of multiple AI nodes in the network, all of which can obtain the first QoE measurement configuration information.

[0144] Alternatively, the topology information may also include identifiers of the AI ​​tasks that the AI ​​nodes can execute. The terminal or the first RAN node can then use the identifiers of the first AI task and the topology information to identify multiple AI nodes in the network that can execute the first AI task as candidate nodes. For example, if the topology information includes identifiers of AI nodes 1 to 10, AI nodes 1 to 5 can execute AI task 1, while AI nodes 6 to 10 cannot execute AI task 1, and the first AI task is AI task 1, then the terminal or the first RAN node can identify AI nodes 1 to 5 as candidate nodes, and thus these AI nodes 1 to 5 can obtain the first QoE measurement configuration information.

[0145] Alternatively, the topology information may also include the geographical region to which the AI ​​nodes belong. The terminal or the first RAN node can then select AI nodes located in the target geographical region as candidate nodes based on the destination geographical region and the topology information. The target geographical region is, for example, the geographical region where the terminal is located. For instance, if the topology information includes the identifiers of AI nodes 1 to 10, AI nodes 1 to 5 belong to geographical region 1, AI nodes 6 to 10 belong to geographical region 2, and the target geographical region is geographical region 1, then the terminal or the first RAN node can select AI nodes 1 to 5 as candidate nodes, and these AI nodes 1 to 5 can then obtain the first QoE measurement configuration information.

[0146] Furthermore, the topology information may also include the identifiers of the AI ​​tasks that the AI ​​node can perform and the geographical region to which the AI ​​node belongs. Alternatively, the topology information may also include information about other AI nodes, which are used by the terminal or the first RAN node to determine candidate nodes.

[0147] In addition, the terminal also sends a first QoE measurement identifier, which is used to associate the current QoE measurement. For example, the first QoE measurement configuration information and the first QoE measurement identifier can be carried together in the SRB or in the DRB.

[0148] For example, a terminal can initiate different QoE measurements for the same AI task; that is, the same task identifier can correspond to different QoE measurement identifiers. For instance, a terminal can initiate three QoE measurements for an inference task, where the first QoE measurement is used to obtain the inference accuracy of the inference task (i.e., the first QoE measurement parameter includes inference accuracy), the second QoE measurement task is used to obtain the inference latency of the inference task (i.e., the first QoE measurement parameter includes inference latency), and the third QoE measurement task is used to obtain the throughput of the inference task (i.e., the first QoE measurement parameter includes throughput).

[0149] In other possible examples, the terminal may also send the first QoE measurement configuration information to other nodes, which then forward the information to the AI ​​node. These other nodes could be, for example, access and mobility management function (AMF) network elements.

[0150] Step 503: The AI ​​node obtains the first QoE measurement result information based on the first QoE measurement configuration information.

[0151] For example, the AI ​​node generates the first QoE measurement result information based on the first QoE measurement configuration information.

[0152] The first QoE measurement result information can be the measurement result of the test task of the first AI task (hereinafter referred to as the first AI test task) or the measurement result of the second AI task. The second AI task is a reference task for the first AI task.

[0153] The difference between the first AI test task and the first AI task can be considered as follows: the first AI test task is executed before the first AI task; the first AI test task is used to test the QoE of the AI ​​node; and the first AI task is actually executed on an AI node with a better QoE. For example, the first AI test task and the first AI task are associated with the same QoE measurement configuration information; that is, the first QoE measurement configuration information is also associated with the first AI test task. Alternatively, after receiving the first QoE measurement configuration information, the AI ​​node can start the first AI test task to obtain the first QoE measurement result information. This first QoE measurement result information is used to select the first target AI node (see step 505a or step 505b for details), and the first target AI node is used to actually execute the first AI task.

[0154] When the first QoE measurement result information is the measurement result of the first AI test task, for example, the AI ​​node obtains the first QoE measurement result information based on the first QoE measurement parameters and the executed first AI test task during the execution of the first AI test task. Alternatively, the AI ​​node obtains the first QoE measurement result information based on the first QoE measurement parameters and the result of executing the first AI test task after it has been executed.

[0155] Based on the first QoE measurement configuration information in step 501, the following is an exemplary description of how the AI ​​node obtains the first QoE measurement result information.

[0156] (1) When the first AI task (or the first AI test task) is a training task: The AI ​​node constructs an initial model based on the description of the initial model, and then collects training data and test data respectively based on the descriptions of the training data and the test data. The AI ​​node trains the initial model based on the training data and tests the trained model based on the test data until a model that meets the accuracy requirements is obtained. The AI ​​node uses the training time and training cost required for model training as the first QoE measurement result information based on the first QoE measurement parameters.

[0157] (2) When the first AI task (or the first AI test task) is an inference task: The AI ​​node constructs an inference model based on the description of the inference model, and then collects test data based on the description of the test data. The AI ​​node performs inference based on the test data and the inference model to obtain the inference result, and then uses the inference accuracy, inference latency, and throughput as the first QoE measurement result information based on the first QoE measurement parameters.

[0158] (3) When the first AI task (or the first AI test task) is a data acquisition task: the AI ​​node acquires data according to the target of the data acquisition, and uses the diversity level of the acquired data, the accuracy of the acquired data, the similarity of the acquired data, the data labeling ability, the data acquisition ability, and the data processing ability as the first QoE measurement result information according to the first QoE measurement parameters.

[0159] The second AI task is a reference task for the first AI task. For example, the first AI task and the second AI task are AI tasks of the same type; or the model information of the first AI task is the same as that of the second AI task; or the model information of the first AI task is the same as that of the second AI task, and the data information of the first AI task is the same as that of the second AI task, etc.

[0160] When the first QoE measurement result is the measurement result of the second AI task, for example, the AI ​​node can obtain the first QoE measurement result information based on the first QoE measurement parameters and the executed second AI task during the execution of the second AI task. In other words, the AI ​​node measures the second AI task based on the first QoE measurement parameters to obtain the first QoE measurement result information. Alternatively, the first QoE measurement result information can be obtained after the execution of the second AI task, based on the first QoE measurement parameters and the result of the execution of the second AI task.

[0161] Here, the AI ​​node can be the target AI node (or the second target AI node) that provides the second AI task service to the terminal (or other terminals), that is, the AI ​​node officially executes the second AI task.

[0162] Alternatively, the AI ​​node may also execute a second AI task to obtain second QoE measurement result information. That is, the second AI task is also an AI testing task. The AI ​​node also receives second QoE measurement configuration information and obtains second QoE measurement result information (as the result of executing the second AI task) based on the second QoE measurement configuration information. This second QoE measurement result information is used to select a second target AI node.

[0163] For example, after obtaining the second QoE measurement result information, the AI ​​node obtains the first QoE measurement result information based on the first QoE measurement parameters and the second QoE measurement result information. For example, if the second AI task is an inference task, the AI ​​node first obtains the second QoE measurement result information based on the second QoE measurement parameters, including inference accuracy, inference latency, and throughput; then, the AI ​​node obtains the inference accuracy from the second QoE measurement result information based on the first QoE measurement parameters, and uses the inference accuracy as the first QoE measurement result information.

[0164] In one possible approach, the AI ​​node receives first QoE measurement configuration information, determines whether a second AI task exists among the multiple AI tasks being executed, based on the first QoE measurement configuration information, and obtains the first QoE measurement result information based on the executed second AI task if the second AI task exists; otherwise, it starts a first AI test task and obtains the first QoE measurement result information based on the executed first AI test task.

[0165] In the above scheme, after the AI ​​node receives the first QoE measurement configuration information, if it determines that there is no second AI task among the multiple AI tasks being executed, it needs to start the first AI test task in order to obtain the first QoE measurement result information.

[0166] For example, the first QoE measurement configuration information may be QoE measurement request information. It is understood that information with other names that have this function may also be considered as QoE measurement request information.

[0167] In another possible approach, the AI ​​node receives first QoE measurement configuration information. Based on this information, it determines whether a second AI task exists among the multiple AI tasks currently being executed. If a second AI task exists, the first QoE measurement result information is obtained based on the executed second AI task. If no second AI task exists, the first QoE measurement result information is obtained again during the execution of the second AI task. For example, after receiving the first QoE measurement configuration information, the AI ​​node determines that no second AI task exists among the multiple AI tasks currently being executed. Subsequently, the AI ​​node receives a start request for the second AI task, or a QoE measurement request for the second AI task, executes the second AI task, and obtains the first QoE measurement result information based on the executed second AI task and the first QoE measurement configuration information.

[0168] In the above scheme, after the AI ​​node receives the first QoE measurement configuration information, if it determines that there is no second AI task among the multiple AI tasks being executed, it can wait until the second AI task is executed before obtaining the first QoE measurement result information, without having to start the first AI test task in order to obtain the first QoE measurement result information.

[0169] For example, the first QoE measurement configuration information may be QoE subscription request information. It is understood that information with other names that have this function may also be considered as QoE subscription request information.

[0170] Step 504: The AI ​​node sends the first QoE measurement result information.

[0171] In the first possible implementation, the AI ​​node sends the first QoE measurement result information to the first RAN node. Correspondingly, the first RAN node receives the first QoE measurement result information from the AI ​​node.

[0172] Optional, also includes:

[0173] Step 505a: The first RAN determines the first target AI node (not shown in Figure 5) based on the first QoE measurement result information.

[0174] For example, the first RAN node determines (or selects) the first target AI node from the multiple AI nodes based on the first QoE measurement results of each of the multiple AI nodes.

[0175] Example 1: The first RAN node determines the AI ​​node with the better QoE as the first target AI node based on the first QoE measurement results of multiple AI nodes.

[0176] For example, the first RAN node sorts the multiple AI nodes based on their respective first QoE measurement results, where AI nodes with higher QoE are ranked higher in the ranking result, and vice versa. The first RAN node then selects one of the AI ​​nodes as the first target AI node based on the ranking result.

[0177] For example, the first RAN node determines the comprehensive attribute value of the AI ​​node based on the first QoE measurement result information corresponding to the AI ​​node, and then sorts the comprehensive attribute values ​​of multiple AI nodes. For example, the first QoE measurement result information includes attribute values ​​corresponding to multiple attributes, and the first RAN node can determine the comprehensive attribute value based on the weights of these multiple attributes and their respective attribute values. For example, if the first AI task (or the first AI test task) is an inference task, the first QoE measurement result information includes inference accuracy, inference latency, and throughput, where the weights for inference accuracy, inference latency, and throughput are 0.3, 0.3, and 0.4, respectively.

[0178] For example, the first RAN node inputs the first QoE measurement results of multiple AI nodes into the ranking model, and the ranking model outputs the ranking result.

[0179] The first target AI node can be managed by a first RAN node or by other RAN nodes (referred to as the second RAN node); RAN nodes include, for example, base stations, duplexes (DUs), and cubes (CUs). Alternatively, the first target AI node can be managed by a first core network node or by other core network nodes (referred to as the second core network node). The first core network node is a core network node capable of establishing a direct or indirect connection with the first target AI node. For example, the first core network node is an AMF (Active Network Function) or Session Management Function (SMF) network element that establishes a session with the terminal, such as a Protocol Data Unit (PDU) session. In this application, the first RAN node and the first core network node can be collectively referred to as the first network node, and the second RAN node and the second core network node can be collectively referred to as the second network node. For example, when the first target AI node is managed by the second network node, the first network node can also send handover information to the terminal, which instructs the terminal to switch to the second network node. Alternatively, after receiving the traffic of the first AI task from the terminal, the first network node forwards the traffic of the first AI task to the second network node, and then the second network node forwards the traffic of the first AI task to the first target AI node.

[0180] For example, the first network node is the first RAN node, and the second network node is the second RAN node. Depending on whether the first target AI node is managed by the first RAN node or the second RAN node, the following are two scenarios: Scenario 1 and Scenario 2.

[0181] Scenario 1: The first target AI node is managed by the first RAN node.

[0182] For example, a first RAN node receives traffic from a terminal's first AI task and sends the traffic of the first AI task to a first target AI node. For instance, the first AI task is a one-sided inference task (e.g., inference is on the first target AI node side). The first target AI node executes the first AI task. The first RAN node receives the data to be inferred from the terminal and forwards it to the first target AI node. The first target AI node obtains the inference result based on the data to be inferred and the first AI task, and sends the inference result back to the first RAN node, which then forwards the inference result back to the terminal. Alternatively, the first AI task is a multi-sided inference task (also known as distributed inference). For example, a two-sided inference task may involve partial inference on the terminal side and other inference on the first target AI node side. The first target AI node executes the first AI task. The first RAN node receives the first inference result from the terminal, obtained through inference by the terminal. The first RAN node forwards the first inference result back to the first target AI node. The first target AI node obtains a second inference result based on the first inference result and the first AI task, and sends the second inference result back to the first RAN node, which then forwards the second inference result back to the terminal.

[0183] Scenario 2: The first target AI node is managed by the second RAN node.

[0184] In one possible example, the first RAN node receives traffic for the first AI task from the terminal and sends the traffic for the first AI task to the second RAN node. The second RAN node then receives the traffic for the first AI task and sends the traffic for the first target AI node.

[0185] In another possible example, the first RAN node sends handover information to the terminal, including the identifier of the second RAN node. This handover information instructs the terminal to switch from the first RAN node to the second RAN node. Subsequently, the terminal accesses the second RAN node and sends traffic for the first AI task to it. The second RAN node receives this traffic and then sends it to the first target AI node. For example, the first RAN node also sends the identifier of the first target AI node to the second RAN node.

[0186] Example 2: The first RAN node determines N AI nodes with better QoE as N first target AI nodes based on the first QoE measurement results of multiple AI nodes. The N first target AI nodes have a priority order and can form a priority list, where N is an integer greater than 1.

[0187] For example, if the first target AI nodes are AI nodes 1 to 3, and their priority order is AI node 1, AI node 2 and AI node 3 respectively, then the priority list will include the identifiers of AI node 1, AI node 2 and AI node 3 respectively.

[0188] For example, if the first target AI nodes are AI nodes 1 to 3, and their priority order is AI node 1, AI node 3 and AI node 2 respectively, then the priority list will include the identifiers of AI node 1, AI node 3 and AI node 2 respectively.

[0189] For example, the first RAN node sorts the multiple AI nodes based on their respective first QoE measurement results. AI nodes with higher QoE are ranked higher in the sorting result, and vice versa. The first RAN node then selects N AI nodes as the N first target AI nodes based on the sorting result.

[0190] For example, the first RAN node determines the comprehensive attribute value of the AI ​​node based on the first QoE measurement result information corresponding to the AI ​​node, and sorts the comprehensive attribute values ​​of multiple AI nodes to obtain the sorting result of the multiple AI nodes. See Example 1 above for details.

[0191] For example, the first RAN node inputs the first QoE measurement results information corresponding to multiple AI nodes into the ranking model, and the ranking model outputs the ranking results.

[0192] In this scenario, all N first target AI nodes are managed by a first network node, or some are managed by the first network node while others are managed by a second network node, or all N first target AI nodes are managed by the second network node. For example, in the latter two possibilities, the first network node may also send switching information to the terminal, instructing the terminal to switch to the second network node. Alternatively, after receiving traffic from the terminal for the first AI task, the first network node forwards the traffic to the second network node, which then forwards the traffic to the first target AI nodes.

[0193] For example, the first network node is the first RAN node, and the second network node is the second RAN node. Depending on whether all N first target AI nodes are managed by the first RAN node, some AI nodes are managed by the first RAN node while others are managed by the second RAN node, or all N first target AI nodes are managed by the second RAN node, the following are cases (I) to (III):

[0194] In scenario (1), all N target AI nodes are managed by the first RAN node.

[0195] For example, the first RAN node receives the traffic of the first AI task from the terminal, and sends the traffic of the first AI task to one or more of the N first target AI nodes according to the priority list and the load of the N AI nodes.

[0196] For example, if the traffic of the first AI task sent by the terminal is 10G, and the priority list includes the identifiers of AI node 1, AI node 3 and AI node 2 in sequence, the first RAN node can forward the first 3G of traffic of the first AI task to AI node 1, forward the 4G to 8G of traffic of the first AI task to AI node 2, and forward the 9G to 10G of traffic of the first AI task to AI node 3, according to the priority list and the load of the three AI nodes.

[0197] In scenario (ii), some of the N target AI nodes are managed by the first RAN node, while the other AI nodes are managed by the second RAN node. Here, the second RAN node can be one or more.

[0198] Example 1: The priority list includes the identifiers of AI node 1, AI node 2 and AI node 3 in sequence. AI node 1 and AI node 2 are managed by the first RAN node, and AI node 3 is managed by the second RAN node 1.

[0199] Example 2: The priority list includes the identifiers of AI node 1, AI node 2 and AI node 3 in sequence. AI node 1 is managed by the first RAN node, AI node 2 is managed by the second RAN node 1, and AI node 3 is managed by the second RAN node 2.

[0200] In one possible example, a first RAN node receives traffic from a terminal's first AI task. Based on a priority list and the load of N first target AI nodes, it sends a portion of the traffic to the first target AI nodes managed by the first RAN node, and sends the remaining traffic to a second RAN node, which then forwards it to the first target AI nodes managed by the second RAN node. Exemplarily, the first RAN node also requests the load of the first target AI nodes managed by the second RAN node from the second RAN node. Exemplarily, the first RAN node also sends the identifier of the first target AI nodes managed by the second RAN node to the second RAN node.

[0201] Referring to Example 1 above, the traffic for the first AI task sent by the terminal is 10G. Based on the priority list and the load of the three AI nodes, the first RAN node forwards the first 3G of traffic from the first AI task to AI node 1, the 4th to 8thG of traffic from the first AI task to AI node 2, and the 9th to 10thG of traffic from the first AI task to the second RAN node 1, which then forwards it to AI node 3. For example, the first RAN node also sends the identifier of AI node 3 to the second RAN node 1.

[0202] Referring to Example 2 above, the traffic of the first AI task sent by the terminal is 10G. The first RAN node can, based on the priority list and the load of the three AI nodes, forward the first 3G of the first AI task's traffic to AI node 1, forward the 4G to 8G of the first AI task's traffic to the second RAN node 1, which then forwards it to AI node 2; and forward the 9G to 10G of the first AI task's traffic to the second RAN node 2, which then forwards it to AI node 3. For example, the first RAN node also sends the identifier of AI node 2 to the second RAN node 1 and the identifier of AI node 3 to the second RAN node 2.

[0203] In scenario (3), all N first target AI nodes are managed by second RAN nodes. Here, there can be one or more second RAN nodes.

[0204] Example (1): The priority list includes the identifiers of AI node 1, AI node 2 and AI node 3 in sequence. AI node 1 to AI node 3 are all managed by the second RAN node 1.

[0205] Example (2): The priority list includes the identifiers of AI node 1, AI node 2 and AI node 3 in sequence. AI node 1 and AI node 2 are managed by the second RAN node 1, and AI node 3 is managed by the second RAN node 2.

[0206] In one possible example, the first RAN node receives the traffic of the terminal's first AI task, sends the traffic and priority list of the first AI task to the second RAN node, and the second RAN node, based on the priority list and the load of the N first target AI nodes, sends the traffic of the first AI task to the N first target AI nodes. For example, the second RAN node also receives the load of the N first target AI nodes.

[0207] In conjunction with the above example (1), the traffic of the first AI task sent by the terminal is 10G. The first RAN node sends the traffic and priority list of the first AI task to the second RAN node 1. According to the priority list and the load of the three AI nodes, the second RAN node 1 forwards the first 3G of traffic of the first AI task to AI node 1, forwards the 4G to 8G of traffic of the first AI task to AI node 2, and forwards the 9G to 10G of traffic of the first AI task to AI node 3.

[0208] Combining the above example (2), the traffic of the first AI task sent by the terminal is 10G. The first RAN node sends the traffic and priority list of the first AI task to the second RAN node 1. The second RAN node 1 forwards the first 3G of traffic of the first AI task to AI node 1 according to the priority list and the load of the three AI nodes, forwards the 4G to 8G of traffic of the first AI task to AI node 2, and forwards the 9G to 10G of traffic of the first AI task to the second RAN node 2, which then forwards it to AI node 3. For example, the second RAN node 1 also sends the identifier of AI node 3 to the second RAN node 2.

[0209] In another possible example, the first RAN node sends handover information to the terminal, including an identifier of a second RAN node. This handover information instructs the terminal to switch from the first RAN node to the second RAN node. Subsequently, the terminal sends traffic for the first AI task to the second RAN node. The second RAN node receives the traffic for the first AI task from the terminal and sends the traffic for the first AI task to the first target AI node. The specific implementation is similar to that in case (I) or case (II), and will not be described in detail here. For example, the first RAN node also sends a priority list to the second RAN node.

[0210] It should be noted that after step 505a, the AI ​​node can also periodically or non-periodically (e.g., based on conditional triggering) obtain the first QoE measurement result information and send it to the terminal's serving RAN node (when the terminal has not switched over, the serving RAN node is the first RAN node; when the terminal has switched over, the serving RAN node is the second RAN node). Correspondingly, the terminal's serving RAN node, based on the received first QoE measurement result information, again determines the first target AI node from among the multiple AI nodes. The determined first target AI node can be the same as or different from the first target AI node determined in the embodiment of Figure 5. That is, the serving RAN node can also determine a new first target AI node and forward the terminal's first AI task traffic according to the new first target AI node. For example, if there are N first target AI nodes, where N is an integer greater than 1, the serving RAN node updates the priority list based on the identifiers of the N new first target AI nodes, and then forwards the terminal's first AI task traffic according to the updated priority list. Alternatively, if there is only one first target AI node, the serving RAN node forwards the terminal's first AI task traffic based on the identifier of this new first target AI node.

[0211] For example, the serving RAN node may forward all or part of the traffic forwarded to the current first target AI node to other AI nodes. These other AI nodes can then obtain the first QoE measurement result information based on the received traffic, thereby achieving the effect of detecting (or monitoring) the QoE of the AI ​​node performing the first AI task. For instance, the serving node may copy all or part of the traffic forwarded to the current first target AI node and then forward the copied traffic to other AI nodes.

[0212] In this implementation, the first target AI node and other AI nodes all perform the first AI task. That is, the first QoE measurement result information can also be the measurement result of the first AI task.

[0213] For example, the network includes AI nodes 1 to 10. The terminal obtains first QoE measurement configuration information and sends it to AI nodes 1 to 10. Each AI node 1 to 10 obtains its own first QoE measurement result information 1 based on the first QoE measurement configuration information and sends it to the first RAN node. The first RAN node determines that AI node 1 is the first target AI node based on the first QoE measurement result information 1 of AI nodes 1 to 10. The terminal sends traffic for the first AI task to the first RAN node, which forwards the traffic to AI node 1. Accordingly, AI node 1 executes the first AI task based on the traffic and obtains the first QoE measurement result information 2 from AI node 1. Furthermore, the first RAN node can also copy a portion of the traffic from the first AI task and forward the copied traffic to AI nodes 2 through 10. Each of AI nodes 2 through 10 obtains its own first QoE measurement result information 2 based on the first QoE measurement configuration information and the traffic, and sends their respective first QoE measurement result information 2 to the first RAN node. Based on the first QoE measurement result information 2 from AI nodes 1 through 10, the first RAN node determines that AI node 2 is the first target AI node. Therefore, when there is traffic from the first AI task originating from the terminal, the first RAN node forwards the traffic of the first AI task to AI node 2, and so on. This ensures that the AI ​​node providing services to the terminal is the one with better QoE performance. The traffic here is equivalent to the data in the first QoE measurement configuration information, such as training data in a training task, data input into the inference model in an inference task, and traffic to be collected in a data acquisition task.

[0214] In a second possible implementation, the AI ​​node sends the first QoE measurement result information to the terminal. For example, the AI ​​node sends the first QoE measurement result information to the first RAN node, and the first RAN node forwards the first QoE measurement result information to the terminal. Correspondingly, the terminal receives the first QoE measurement result information from the AI ​​node.

[0215] Optional, also includes:

[0216] Step 505b: The terminal determines the first target AI node (not shown in Figure 5) based on the first QoE measurement result information.

[0217] For example, the terminal determines (or selects) the first target AI node from multiple AI nodes based on the first QoE measurement results of each AI node.

[0218] Example 1: The terminal determines the AI ​​node with the better QoE as the first target AI node based on the first QoE measurement results of multiple AI nodes.

[0219] The way the terminal determines the first target AI node is similar to the way the first RAN node determines the first target AI node.

[0220] Similarly, the first target AI node can be managed by either the first network node or the second network node. For example, when the first target AI node is managed by the second network node, the terminal can also actively request access to the second network node. Alternatively, after receiving traffic from the terminal's first AI task, the first network node forwards the traffic to the second network node, which then forwards the traffic to the first target AI node.

[0221] For example, the first network node is the first RAN node, and the second network node is the second RAN node. Depending on whether the first target AI node is managed by the first RAN node or the second RAN node, the following are two cases: Case 1 and Case 2.

[0222] Case 1: The first target AI node is managed by the first RAN node.

[0223] The terminal sends the traffic of the first AI task and the identifier of the first target AI node to the first RAN node. Accordingly, the first RAN node forwards the traffic of the first AI task to the first target AI node based on the identifier of the first target AI node.

[0224] Scenario 2: The first target AI node is managed by the second RAN node.

[0225] In one possible example, the terminal sends the traffic of the first AI task and the identifier of the first target AI node to the first RAN node. The first RAN node forwards the traffic of the first AI task and the identifier of the first target AI node to the second RAN node based on the identifier of the first target AI node; the second RAN node then sends the traffic of the first AI task to the first target AI node based on the identifier of the first target AI node.

[0226] In another possible example, the terminal sends an access request to the second RAN node to request access to the second RAN node. Subsequently, the terminal sends the traffic for the first AI task and the identifier of the first target AI node to the second RAN node. Based on the identifier of the first target AI node, the second RAN node sends the traffic for the first AI task to the first target AI node.

[0227] Example 2: The terminal determines N first target AI nodes based on the first QoE measurement results of multiple AI nodes. The N first target AI nodes have a priority order and can form a priority list. N is an integer greater than 1.

[0228] The way the terminal determines the first target AI node is similar to the way the first RAN node determines the first target AI node.

[0229] Similarly, all N first target AI nodes can be managed by the first network node, or some can be managed by the first network node while others are managed by the second network node, or all N first target AI nodes can be managed by the second network node. For example, in the latter two cases, the terminal can also actively request access to the second network node. Alternatively, after receiving traffic from the terminal's first AI task, the first network node forwards the traffic to the second network node, which then forwards the traffic to the first target AI node.

[0230] For example, the first network node is the first RAN node, and the second network node is the second RAN node. Based on whether all N first target AI nodes are managed by the first RAN node, or some AI nodes are managed by the first RAN node while others are managed by the second RAN node, or all N first target AI nodes are managed by the second RAN node, the following cases (1) to (3) are considered:

[0231] In case (1), all N first target AI nodes are managed by the first RAN node.

[0232] The terminal sends a traffic and priority list for the first AI task to the first RAN node. The priority list includes the identifiers and priorities of N first target AI nodes. The first RAN node obtains the load of the N first target AI nodes based on their identifiers, and then forwards the traffic of the first AI task to the N first target AI nodes based on their load and priorities.

[0233] In case (2), some of the N first target AI nodes are managed by the first RAN node, while the other AI nodes are managed by the second RAN node. Here, the second RAN node can be one or more.

[0234] The terminal sends a traffic and priority list for the first AI task to the first RAN node. The priority list includes the identifiers and priorities of N first target AI nodes. Accordingly, the first RAN node obtains the load of the N first target AI nodes based on their identifiers, and then forwards a portion of the traffic of the first AI task to the first target AI nodes managed by the first RAN node based on their load and priorities. The second RAN node forwards the remaining traffic to the second target AI node, which in turn forwards it to the first target AI nodes managed by the second RAN node.

[0235] In case (3), all N first target AI nodes are managed by second RAN nodes. Here, the second RAN node can be one or more.

[0236] The terminal sends a traffic and priority list for the first AI task to the first RAN node. The priority list includes the identifiers and priorities of N first target AI nodes. The first RAN node sends the traffic and priority list for the first AI task to the second RAN node. The second RAN node obtains the load of the N first target AI nodes based on their identifiers, and then forwards the traffic of the first AI task to the N first target AI nodes based on their load and priorities.

[0237] Alternatively, the terminal switches from the first RAN node to the second RAN node. Subsequently, the terminal sends the traffic and priority list of the first AI task to the second RAN node. The second RAN node receives the traffic and priority list of the first AI task from the terminal and sends the traffic of the first AI task to the first target AI node according to the priority list. The specific implementation is similar to case (1) or case (2), and will not be described in detail here.

[0238] It should be noted that after step 505b, the AI ​​node can also periodically or non-periodically (e.g., based on conditional triggering) obtain the first QoE measurement result information and send it to the terminal. Correspondingly, the terminal, based on the received first QoE measurement result information, again determines the first target AI node from among multiple AI nodes. For specific implementation details, please refer to the method of the serving RAN node updating the first target AI node. This ensures that the AI ​​node providing services to the terminal is the AI ​​node with better QoE performance.

[0239] Figures 6 to 9 illustrate four specific communication methods provided in this application, tailored to specific scenarios.

[0240] It is understood that any content not detailed in the embodiments shown in Figures 6 to 9 can be found in the descriptions in the embodiments shown in Figure 5. The step numbers in the flowcharts described in each embodiment are merely examples of the execution flow and do not constitute a restriction on the order of execution. There is no strict execution order between steps in this application that have no temporal dependency. Not all steps shown in the flowcharts are mandatory; some steps can be deleted or added as needed. This application focuses on describing the differences between different embodiments. Except for the differences, the embodiments can be referenced interchangeably; within the same embodiment, different implementations or examples can also be referenced interchangeably.

[0241] Figure 6 is a flowchart illustrating the communication method provided by this application. In this communication method, the first QoE measurement configuration information is the QoE measurement request information. The second AI task is not executed in the AI ​​node. The first RAN node forwards the QoE measurement request information from the terminal. The terminal determines the first target AI node, and the first target AI node is managed by the first RAN node.

[0242] Step 601: The terminal obtains QoE measurement request information. See step 501 for details.

[0243] Step 602: The terminal sends a QoE measurement request message to the first RAN node. The QoE measurement request message may be carried in an SRB. See step 502 for a detailed implementation.

[0244] Step 603: The first RAN node sends QoE measurement request information to AI node 1 and AI node 2 respectively. For ease of description, Figure 6 only shows AI node 1 and AI node 2, but this application does not limit the number of AI nodes in the network. For specific implementation, please refer to step 502.

[0245] Step 604: AI node 1 executes and measures the first AI test task according to the QoE measurement request information to obtain the first QoE measurement result information of AI node 1. AI node 2 executes and measures the first AI test task according to the QoE measurement request information to obtain the first QoE measurement result information of AI node 2. For specific implementation details, please refer to step 503.

[0246] Step 605: AI Node 1 sends its first QoE measurement result information to the first RAN node. AI Node 2 sends its first QoE measurement result information to the first RAN node. See step 504 for details.

[0247] Step 606: The first RAN node sends measurement result information to the terminal. The measurement result information includes the first QoE measurement result information of AI node 1 and the first QoE measurement result information of AI node 2. For specific implementation details, please refer to step 504.

[0248] Both the QoE measurement request information and the measurement result information can be RRC messages.

[0249] Step 607: The terminal determines the first target AI node based on the first QoE measurement result information of AI node 1 and the first QoE measurement result information of AI node 2, wherein the first target AI node is, for example, AI node 1. For specific implementation, please refer to step 505b.

[0250] Step 608: The terminal sends the identifier of the first target AI node and the traffic of the first AI task to the first RAN node.

[0251] Step 609: The first RAN node sends the traffic of the first AI task to the first target AI node according to the identifier of the first target AI node.

[0252] Figure 7 is a flowchart illustrating the communication method provided by this application. In this communication method, the first QoE measurement configuration information is the QoE measurement request information. The second AI task is not executed in the AI ​​node. The first RAN node forwards the QoE measurement request information from the terminal. The first RAN node determines the first target AI node. The first target AI node is managed by the first RAN node.

[0253] Steps 701 to 705 correspond to steps 601 to 605, respectively.

[0254] Step 706: The first RAN node determines the first target AI node based on the first QoE measurement result information of AI node 1 and the first QoE measurement result information of AI node 2, wherein the first target AI node is, for example, AI node 1.

[0255] Step 707: The terminal sends the traffic for the first AI task to the first RAN node.

[0256] Step 708: The first RAN node sends the traffic of the first AI task to the first target AI node according to the identifier of the first target AI node.

[0257] For the specific implementation of steps 706 to 708, please refer to step 505a.

[0258] Figure 8 is a flowchart illustrating the communication method provided by this application. In this communication method, the first QoE measurement configuration information is QoE measurement request information. The second AI task is not executed in the AI ​​node. The first RAN node forwards the QoE measurement request information from the terminal. The first RAN node determines the first target AI node. The first target AI node is managed by the second RAN node.

[0259] Steps 801 to 805 correspond to steps 601 to 605, respectively.

[0260] Step 806: The first RAN node determines the first target AI node based on the first QoE measurement result information of AI node 1 and the first QoE measurement result information of AI node 2, wherein the first target AI node is, for example, AI node 1.

[0261] Step 807: The first RAN node sends handover information to the terminal. The handover information instructs the terminal to switch to the second RAN node. For example, the handover information includes the identifier of the second RAN node. For instance, before sending the handover information to the terminal, the first RAN node may also determine that the first target AI node is an AI node managed by the second RAN node.

[0262] Step 808: The terminal sends an access request message to the second RAN node to access the second RAN node.

[0263] Step 809: The terminal sends the traffic for the first AI task to the second RAN node.

[0264] Step 810: The first RAN node sends the identifier of the first target AI node to the second RAN node.

[0265] Step 811: The second RAN node sends the traffic of the first AI task to the first target AI node according to the identifier of the first target AI node.

[0266] For the specific implementation of steps 806 to 811, please refer to step 505a.

[0267] Figure 9 is a flowchart illustrating the communication method exemplarily provided in this application. In this communication method, the first QoE measurement configuration information is QoE measurement request information, the AI ​​node is executing a second AI task, the first RAN node forwards the QoE measurement request information from the terminal, the first RAN node determines the first target AI node, and the first target AI node is managed by the first RAN node.

[0268] Step 900: AI node 1 is executing the second AI task, and AI node 2 is executing the second AI task.

[0269] Step 901: The terminal obtains QoE measurement request information. See step 501 for details.

[0270] Step 902: The terminal sends a QoE measurement request message to the first RAN node. The QoE measurement request message may be carried in an SRB. See step 502 for a detailed implementation.

[0271] Step 903: The first RAN node sends QoE measurement request information to AI node 1 and AI node 2 respectively. For ease of description, Figure 9 only shows AI node 1 and AI node 2, but this application does not limit the number of AI nodes in the network. For specific implementation, please refer to step 502.

[0272] Step 904: AI node 1 measures the second AI task according to the QoE measurement request information and obtains the first QoE measurement result information of AI node 1. AI node 2 is similar to AI node 1. For specific implementation details, please refer to step 503.

[0273] Step 905: AI Node 1 sends its first QoE measurement result information to the first RAN node. AI Node 2 sends its first QoE measurement result information to the first RAN node. See step 504 for details.

[0274] Steps 906 to 908 correspond to steps 706 to 708, respectively.

[0275] Figure 10 is a flowchart illustrating the communication method provided by this application. In this communication method, the first QoE measurement configuration information is QoE subscription request information. The second AI task is not executed in the AI ​​node. The first RAN node forwards the QoE subscription request information from the terminal. The first RAN node determines the first target AI node. The first target AI node is managed by the first RAN node.

[0276] Step 1001: The terminal obtains the QoE subscription request information. See step 501 for details.

[0277] Step 1002: The terminal sends a QoE subscription request to the first RAN node. The QoE subscription request is, for example, carried in an SRB. See step 502 for a detailed implementation.

[0278] Step 1003: The first RAN node sends QoE subscription request information to AI node 1 and AI node 2 respectively. For ease of description, Figure 10 only shows AI node 1 and AI node 2, but this application does not limit the number of AI nodes in the network. For specific implementation, please refer to step 502.

[0279] Step 1004: AI node 1 receives the start request information for the second AI task and starts the second AI task; AI node 2 receives the start request information for the second AI task and starts the second AI task. See step 503 for the specific implementation details.

[0280] Step 1005: AI node 1 measures the second AI task based on the QoE subscription request information, and obtains the first QoE measurement result information of AI node 1. AI node 2 is similar to AI node 1. For specific implementation details, please refer to step 503.

[0281] Step 1006: AI Node 1 sends its first QoE measurement result information to the first RAN node. AI Node 2 sends its first QoE measurement result information to the first RAN node. See step 504 for details.

[0282] Steps 1007 to 1009 correspond to steps 706 to 708, respectively.

[0283] Figure 11 shows a possible exemplary block diagram of the communication device involved in the embodiments of this application. As shown in Figure 11, the communication device 1100 may include modules or units for implementing the method embodiments described above.

[0284] In one possible design, the communication device 1100 includes a processing unit 1102 and a communication unit 1103. Optionally, the communication device 1100 may also include a storage unit 1101 for storing device program code and / or data.

[0285] The communication device 1100 can be a terminal-side device as described in the above embodiments, such as a terminal or a communication module in a terminal, or a circuit or chip in a terminal that is responsible for communication functions.

[0286] For example, in one embodiment, the processing unit 1102 is used to: obtain QoE measurement configuration information, the QoE measurement configuration information is associated with a first AI task, the QoE measurement configuration information is used to obtain QoE measurement result information, the QoE measurement result information is used to determine a target AI node, and the target AI node is used to execute the first AI task; the communication unit 1103 is used to: send the QoE measurement configuration information.

[0287] In one possible design, the communication unit 1103 is further configured to: receive QoE measurement result information; and the processing unit 1102 is further configured to: determine the target AI node based on the QoE measurement result information.

[0288] In one possible design, the communication unit 1103 is also used to: receive switching information, which indicates switching to a network node for managing the target AI node.

[0289] In one possible design, when the communication device 1100 is a terminal or a communication module within a terminal, the function of the processing unit 1102 can be implemented by one or more processors. Specifically, the processor may include a modem chip, or a system-on-a-chip (SoC) chip or a SIP chip containing a modem core. The function of the communication unit 1103 can be implemented by transceiver circuitry.

[0290] In one possible design, when the communication device 1100 is a circuit or chip in a terminal responsible for communication functions, such as a modem chip or a system-on-a-chip (SoC) or SIP chip containing a modem core, the function of the processing unit 1102 can be implemented by a circuit system in the aforementioned chip that includes one or more processors or processor cores. The function of the communication unit 1103 can be implemented by an interface circuit or data transceiver circuit on the aforementioned chip.

[0291] In one possible design, when the communication device 1100 is a terminal or a communication and / or computing module within a terminal, the functionality of the processing unit 1102 can be implemented by one or more processors. Specifically, the processor may include a GPU, or a system-on-a-chip (SoC) or SIP chip containing a GPU. Alternatively, the processor may include an AI processor, or a SoC or SIP chip containing an AI processor. Or, the processor may include an ASIC, or a SoC or SIP chip containing an ASIC. The functionality of the communication unit 1103 can be implemented by transceiver circuitry.

[0292] In one possible design, when the communication device 1100 is a circuit or chip in a terminal responsible for communication and / or computing functions, such as a GPU or a system-on-a-chip (SoC) or SIP chip containing a GPU, an AI processor or a SoC or SIP chip containing an AI processor, or an ASIC or a SoC or SIP chip containing an ASIC, the function of the processing unit 1102 can be implemented by a circuit system in the aforementioned chip that includes one or more processors or processor cores. The function of the communication unit 1103 can be implemented by interface circuits or data transceiver circuits on the aforementioned chip.

[0293] The communication device 1100 can be an AI node-side device as described in the above embodiments, such as a communication module in an AI server, or a circuit or chip in an AI server responsible for communication functions. For example, in one embodiment, the communication unit 1103 is used to: receive QoE measurement configuration information, which is associated with a first AI task; the processing unit 1102 is used to: obtain QoE measurement result information based on the QoE measurement configuration information, which is used to determine the target AI node, and the target AI node is used to execute the first AI task; the communication unit 1103 is also used to: send the QoE measurement result information.

[0294] It is understood that the division of units in the above-described device is merely a logical functional division. One function can correspond to one functional unit, or two or more functions can be integrated into one functional unit. In actual implementation, all or some units can be integrated onto a single physical entity, or distributed across different physical entities. Furthermore, the aforementioned functional units can be implemented in hardware, software, or a combination of both. Whether a function is executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for specific applications, but such implementations should not be considered beyond the scope of this application.

[0295] In one example, the functional unit in any of the above devices may be one or more integrated circuits configured to implement the above methods, such as: one or more application-specific integrated circuits (ASICs), or one or more central processing units (CPUs), one or more microcontroller units (MCUs), one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0296] In one example, storage unit 1101 may include random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory and / or registers, etc.

[0297] Referring to Figure 12, it is a structural schematic diagram of a terminal 1200 provided in an embodiment of this application. The terminal 1200 can correspond to the terminals shown in Figures 1 to 3 and is used to implement the operation of the terminal in the above embodiments.

[0298] As shown in Figure 12, the terminal includes: one or more antennas 1210, a radio frequency processing system 1220, and a processor system 1230.

[0299] In the downlink or sidelink direction, the RF processing system 1220 receives RF signals through the antenna 1210 and sends the RF-processed signals to the processor system 1230 for further processing. In the uplink or sidelink direction, the processor system 1230 processes the terminal-side information and sends it to the RF processing system 1220, which then processes the signal and transmits it through the antenna 1210.

[0300] In one example, the radio frequency (RF) processing system 1220 serves as the communication interface for external communication of the terminal and may include an RF front end (RFFE) 1221 and an RF transceiver 1222. The RFFE 1221 is primarily used for one or more processing operations, such as shaping, passband selection, or gain adjustment, on the RF signals received by the antenna or those to be transmitted through the antenna. It may include one or more components such as RF switches, duplexers, filters, power amplifiers, antenna tuners, and low-noise amplifiers. The RFFE 1221 can be a circuit system composed of multiple discrete devices or integrated into one or more chips. The transceiver 1222 processes the RF signals received by the RFFE into baseband / IF signals for further processing by the processor system 1230, and processes the baseband / IF signals provided by the processor system 1230 into RF signals for transmission to the RFFE 1221. The baseband / IF signals transmitted between the transceiver 1222 and the processor system 1230 can be digital or analog signals. The transceiver 1222 can be implemented by one or more chips, which are commonly referred to as radio frequency ICs (RFICs).

[0301] In one example, processor system 1230 may include one or more processors for processing signals and executing one or more communication protocols. Optionally, processor system 1230 may also include memory 1236. In one example, the one or more processors include at least one baseband processor 1231 (also known as a modem processor). Memory 1236 is used to store data and / or computer program instructions. Optionally, processor system 1230 may also include one or more application processors 1232 for implementing processing of the terminal operating system and application layer. Application processor 1232 may include, for example, a GPU, AI processor, or ASIC. Optionally, processor system 1230 may also include one or more of a voice subsystem 1233, a multimedia subsystem 1234, or an interface circuit 1235. The voice subsystem 1233 is used to process voice signals, the multimedia subsystem 1234 is used to handle multimedia-related operations, such as video encoding / decoding, image processing, etc., and the interface circuit 1235 is used to implement communication with other terminal components, such as a display 1240, an input device 1250, memory 1260, etc. The aforementioned components in the processor system 1230 can communicate with each other via a bus or communication interface circuit.

[0302] In one example, the processor system 1230 can be packaged as a single processor chip, such as a SoC chip or a SIP chip. In another example, the processor system 1230 can be a system composed of multiple chips, for example, the baseband processor 1231 can be packaged as a single chip, or packaged with part or all of the circuitry of the radio frequency processing system into a single chip.

[0303] In one example, memory 1236 can be on-chip memory, i.e., located on the processor system 1230 chip. In another example, memory 1260 can be off-chip memory, i.e. located outside the processor system 1230 chip.

[0304] In one example, the baseband processor 1231 may include one or more processor cores 12311 and interface circuitry 12314. The one or more processor cores 12311 are used to process signals and execute one or more communication protocols. Optionally, the baseband processor 1231 may also include a memory 12312 for storing at least a portion of the corresponding computer program instructions and / or data. In one example, the one or more processor cores 12311 execute the computer program instructions stored in the memory 12312 to perform the relevant operations in the above method embodiments (such as obtaining QoE measurement configuration information and sending QoE measurement configuration information). In this disclosure, memory 12312 is used to store corresponding computer program instructions and / or data. This can mean that memory 12312 stores all corresponding computer program instructions and / or data for execution by processor core 12311; or it can mean that memory 12312 stores a portion of corresponding computer program instructions and / or data, including the computer program instructions and / or data currently required to be executed by processor core 12311. Memory 12312 can store different portions of computer program instructions and / or data multiple times for execution by processor core 12311 to implement the relevant operations in the above method embodiments. Interface circuit 12314 serves as a communication interface for communication with other components, such as transmitting signals with radio frequency processing system 1220, communicating with other subsystems and related components of processor system 1230 via a bus, such as transmitting data control signals with application processor 1232, and transmitting data or computer program instructions with memory 1236 or memory 1260. Optionally, in order to reduce the load on the processor core, a baseband signal processing circuit 12313 can be set to perform at least some baseband signal processing, including one or more of signal demodulation, modulation, encoding or decoding.

[0305] In one example, the communication device provided in this application may be a terminal 1200, a communication module including a processor system 1230 and a radio frequency system 1220, the processor system 1230, or a baseband processor 1231.

[0306] The processor, processor system, application processor, baseband processor, processor circuit, or processor core mentioned above can be collectively referred to as a processor. The processor may include one or more of the following: central processing unit (CPU), digital signal processor (DSP), microprocessor unit (MPU), microcontroller unit (MCU), graphics processing unit (GPU), field programmable gate array (FPGA), application specific integrated circuit (ASIC), artificial intelligence processor (AI processor), or neural processing unit (NPU).

[0307] The aforementioned memory may include one or more of the following storage media: random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), phase-change memory (PCM), resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), cache, register, read-only memory (ROM), flash memory, erasable programmable read-only memory (EPROM), hard disk, etc. In one example, computer program instructions for executing the above embodiments may be stored on non-volatile memory, such as at least a portion of the aforementioned memory 1260 (e.g., one or more of ROM, flash memory, EPROM, or hard disk). When executed on the terminal, the corresponding computer program instructions may be partially or wholly loaded onto a memory with a faster transfer speed than the processor, such as at least a portion of memory 1236 and / or memory 12312 (e.g., one or more of RAM, SRAM, DRAM, PCM, RERAM, MRAM, FRAM, cache, or register), for the processor to execute in order to implement the steps in the above method embodiments.

[0308] In one example, the transceiver 1222 and the RF front-end 1221 can also be packaged in a single chip. In another example, the transceiver 1222, the RF front-end 1221, and the baseband processor 1231 can also be packaged in a single chip.

[0309] The names used in this application (e.g., parameter names, message names, network element / device names) are exemplary. For example, QoE measurement configuration information can also be called application layer measurement configuration information, measurement configuration information, etc. RAN nodes can also be called access network nodes, etc.

[0310] The terms "system" and "network" in this application embodiment are used interchangeably. "At least one" refers to one or more, and "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, "at least one of A, B, or C" includes A, B, C, AB, AC, BC, or ABC; "at least one of A, B, and C" can also be understood as including A, B, C, AB, AC, BC, or ABC. Furthermore, unless otherwise specified, the ordinal numbers such as "first" and "second" mentioned in this application embodiment are used to distinguish multiple objects and are not used to limit the order, sequence, priority, or importance of multiple objects.

[0311] In this application, "sending information" can be understood as one device sending information to another device, or it can also be understood as one logical module within a device sending information to another logical module. For example, "RAN node sending information" can be understood as the RAN node sending information to another device (such as a terminal), or it can be understood as logical module 1 in the RAN node sending information to logical module 2 in the RAN node.

[0312] In this application, "receiving information" can be understood as one device receiving information from another device, or it can also be understood as a logical module within a device receiving information from another logical module. For example, "RAN node receiving information" can be understood as the RAN node receiving information from another device (such as a terminal), or it can be understood as logical module 1 in the RAN node receiving information from logical module 2 in the RAN node.

[0313] In this application, phrases such as "sending information to... (e.g., a terminal)" or related illustrations in the accompanying drawings can be understood as indicating that the destination of the information is a terminal. This can include sending information directly or indirectly to a terminal. Similarly, phrases such as "receiving information from... (e.g., a terminal)," "receiving information from... (e.g., a terminal)," or "receiving information sent by (e.g., a terminal)," or related illustrations in the accompanying drawings, can be understood as indicating that the source of the information is a terminal. This can include receiving information directly or indirectly from a terminal. Information may undergo necessary processing between the source and destination, such as format changes, but the destination can understand the valid information from the source. Similar expressions in this application can be interpreted similarly and will not be elaborated further here.

[0314] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0315] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0316] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0317] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

Claims

1. A communication method, characterized in that, include: The QoE measurement configuration information is obtained, which is associated with a first artificial intelligence (AI) task. The QoE measurement configuration information is used to obtain QoE measurement result information, which is used to determine the target AI node. The target AI node is used to execute the first AI task. Send the QoE measurement configuration information.

2. The method as described in claim 1, characterized in that, Also includes: Receive the QoE measurement result information; The target AI node is determined based on the QoE measurement results.

3. The method as described in claim 1 or 2, characterized in that, Also includes: Receive switching information, which indicates switching to a network node for managing the target AI node.

4. The method according to any one of claims 1-3, characterized in that, The QoE measurement configuration information is used to indicate the model of the first AI task, and the QoE measurement configuration information is carried in the signaling radio bearer or the data radio bearer.

5. The method as described in claim 4, characterized in that, The QoE measurement configuration information includes the model of the first AI task, and the QoE measurement configuration information is carried in the data radio bearer; or, The QoE measurement configuration information includes a model description of the first AI task, and the QoE measurement configuration information is carried in the signaling radio bearer.

6. The method according to any one of claims 1-5, characterized in that, The QoE measurement result information is the measurement result of the second AI task, which is a reference task for the first AI task.

7. The method as described in claim 6, characterized in that, The second AI task is an AI task that is being executed in the AI ​​node or has already been executed in the AI ​​node.

8. The method as described in claim 6 or 7, characterized in that, The first AI task and the second AI task are of the same type.

9. A communication method, characterized in that, include: Receive Quality of Experience (QoE) measurement configuration information, which is associated with a first artificial intelligence (AI) task; Based on the QoE measurement configuration information, QoE measurement result information is obtained. The QoE measurement result information is used to determine the target AI node, and the target AI node is used to execute the first AI task. Send the QoE measurement result information.

10. The method as described in claim 9, characterized in that, The QoE measurement result information is the measurement result of the second AI task, which is a reference task for the first AI task.

11. The method as described in claim 10, characterized in that, The first AI task and the second AI task are of the same type.

12. A communication method, characterized in that, include: Receive Quality of Experience (QoE) measurement result information, wherein the QoE measurement result information is obtained by the AI ​​node based on the QoE measurement configuration information from the terminal, and the QoE measurement configuration information is associated with the first AI task; Based on the QoE measurement results, a target AI node is determined, which is used to execute the first AI task.

13. The method as described in claim 12, characterized in that, Also includes: Receive traffic from the first AI task from the terminal; Send the traffic of the first AI task to the target AI node.

14. The method as described in claim 13, characterized in that, The step of determining the target AI node based on the QoE measurement results includes: Based on the QoE measurement results, determine multiple target AI nodes and their priority order. The process of sending the first AI task to the target AI node includes: Based on the plurality of target AI nodes, the priority order of the plurality of target AI nodes, and the load of the plurality of target AI nodes, the traffic of the first AI task is sent to the plurality of target AI nodes.

15. The method according to any one of claims 12-14, characterized in that, Also includes: A switching message is sent to the terminal, indicating a switch to a network node for managing the target AI node.

16. A communication device, characterized in that, Includes a unit for performing the method as described in any one of claims 1 to 8.

17. A communication device, characterized in that, Includes units for performing the method as described in any one of claims 9 to 11.

18. A communication device, characterized in that, Includes units for performing the method as described in any one of claims 12 to 15.

19. A communication device, characterized in that, include: Interface circuitry and one or more processors; The interface circuit is used for communication within the communication device or between the communication device and other devices. The one or more processors are coupled to a memory for storing part or all of a necessary computer program or instructions. When the one or more processors execute part or all of the computer program or instructions, the communication device performs the method of any one of claims 1 to 8, or the communication device performs the method of any one of claims 9 to 11, or the communication device performs the method of any one of claims 12 to 15.

20. A computer-readable storage medium, characterized in that, The storage medium stores a computer program or instructions that, when executed, cause the method as described in any one of claims 1 to 8 to be implemented, or cause the method as described in any one of claims 9 to 11 to be implemented, or cause the method as described in any one of claims 12 to 15 to be implemented.

21. A computer program product, characterized in that, The computer program product includes a computer program or instructions that, when executed, cause the method as described in any one of claims 1 to 8 to be implemented, or cause the method as described in any one of claims 9 to 11 to be implemented, or cause the method as described in any one of claims 12 to 15 to be implemented.

22. A communication system, characterized in that, include: Terminals, AI nodes, and RAN nodes; The terminal is used to perform the method according to any one of claims 1 to 8, the AI ​​node is used to perform the method according to any one of claims 9 to 11, and the RAN node is used to perform the method according to any one of claims 12 to 15.

Images