Task scheduling method and device for artificial intelligence (AI) network function service

Abstract

The invention provides a task scheduling method and device for an artificial intelligence AI network function service, and relates to the technical field of mobile communication, according to the task scheduling method for the artificial intelligence AI network function service provided by the embodiment of the invention, a first network node can respond to an artificial intelligence AI network function service establishment request; screening out at least two second network nodes from a plurality of second network nodes, the node level of the plurality of second network nodes being smaller than the node level of the first network node; and sending a target task corresponding to the AI network function service establishment request to the at least two second network nodes. According to the method and the device, the artificial intelligence AI network function can be subdivided according to a specific algorithm and task types, so that the task scheduling efficiency can be maximized, and the AI service can be efficiently and flexibly carried out.

Description

Task scheduling method and device for artificial intelligence (AI) network function services Technical Field

[0001] The present disclosure relates to the field of mobile communication technology, and in particular to a task scheduling method and device for artificial intelligence (AI) network function services. Background Art

[0002] Artificial Intelligence (AI) will become one of the core technologies of future communications. Typical application scenarios for 6G and AI overlap by over 80%, demonstrating a deep integration between the two. Furthermore, the massive coverage of 6G networks will provide a ubiquitous platform for AI, addressing the significant pain point of a lack of carriers and channels for AI implementation and significantly promoting the development and prosperity of the AI ​​industry. Currently, numerous automated methods are being employed across network planning, construction, maintenance, and optimization stages to improve operational efficiency. However, the overall level of network autonomy is still relatively low, leaving significant room for improvement. Software-defined Networks (SDN) and Network Function Virtualization (NFV) currently play a crucial role in the evolution of network architectures. While SDN and NFV architectures offer networks a high degree of flexibility, they also increase complexity. Consequently, more factors need to be considered in aspects such as the scheduling and allocation of network resources, transmission paths, and optimization algorithm design, necessitating more intelligent approaches. AI technology can help networks achieve a higher level of autonomy, reducing costs and increasing efficiency.

[0003] However, the current task scheduling for AI network function services often lacks a common AI workflow and a unified technical framework, resulting in insufficient granularity in the division of AI network functions and fragmented application scenarios. This in turn fails to meet users' personalized AI network service needs and cannot reasonably schedule the tasks of AI network function resources.

[0004] Summary of the Invention

[0005] The present disclosure provides a task scheduling method and device for artificial intelligence (AI) network function services, which can subdivide artificial intelligence (AI) network functions according to specific algorithms and task types, thereby maximizing task scheduling efficiency and enabling AI services to be carried out efficiently and flexibly.

[0006] A first embodiment of the present disclosure provides a task scheduling method for an artificial intelligence (AI) network function service, the method being applied to a first network node, the method comprising:

[0007] In response to the artificial intelligence (AI) network function service establishment request, screening at least two second network nodes from a plurality of second network nodes, wherein the node level of the plurality of second network nodes is lower than the node level of the first network node;

[0008] Sending the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes.

[0009] In some embodiments of the present disclosure, the method further includes:

[0010] Receive the artificial intelligence AI network function service establishment request sent by the access and mobility management function AMF.

[0011] In some embodiments of the present disclosure, the AI ​​network function service establishment request carries the AI ​​network function service type, and screening at least two second network nodes from the plurality of second network nodes includes:

[0012] Determine multiple second network nodes that match the AI ​​network function service type, and receive node status information of the multiple second network nodes, where the node status information includes at least CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information;

[0013] Based on the node status information and a preset deep reinforcement learning algorithm, at least two second network nodes are selected from the plurality of second network nodes.

[0014] In some embodiments of the present disclosure, the method further includes:

[0015] Receiving a preset task type regarding an AI network function service uploaded by a second network node;

[0016] The determining of multiple second network nodes matching the AI ​​network function service type includes:

[0017] According to the preset task type, multiple second network nodes matching the AI ​​network function service type are determined.

[0018] In some embodiments of the present disclosure, the preset deep reinforcement learning algorithm includes a dual deep Q learning DDQN algorithm, and the selecting at least two second network nodes from the plurality of second network nodes based on the node state information and the preset deep reinforcement learning algorithm includes:

[0019] With the goal of minimizing system latency and energy consumption and maximizing the number of second network nodes participating in the service, the state space is defined as the node state information of the second network node, the action space is defined as the second network node combination selected for the AI ​​network function service establishment request, and a deep reinforcement learning model is modeled based on the DDQN algorithm;

[0020] The deep reinforcement learning model is used to determine the reward function of each action in the action set based on the node state information, and the optimal selection decision of the second network node corresponding to the AI ​​network function service is determined based on the reward function, wherein the action is used to represent the selection decision of the second network node corresponding to the AI ​​network function service, and the optimal selection decision includes at least two second network nodes and the task allocation weights of the at least two second network nodes.

[0021] In some embodiments of the present disclosure, determining a reward function for each action in an action set based on the node state information using the deep reinforcement learning model includes:

[0022] Determine an action set based on the node state information using the deep reinforcement learning model, and repeat the reward function determination process until the current action is determined to be the last action in the action set:

[0023] The reward function determination process includes: taking the next action as the current action according to the action selection order in the action set; receiving the local model parameters uploaded by the second network node included in the current action, and based on the federated average FedAvg algorithm, weightedly aggregating the local model parameters according to the local data size to obtain global model parameters; sending the global model parameters to the second network node included in the current action to continue training the task model of the second network node based on the global model parameters, and obtaining the reward function of the current action after the training is completed.

[0024] In some embodiments of the present disclosure, determining the optimal selection decision for the target task with respect to the second network node according to the reward function includes:

[0025] A target action in the action set that corresponds to the maximum reward function is determined, and a selection strategy represented by the target action is determined as an optimal selection decision for the target task with respect to the second network node.

[0026] In some embodiments of the present disclosure, sending the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes includes:

[0027] Dividing the target task into at least two subtasks according to the task allocation weights of the at least two second network nodes;

[0028] The subtasks are respectively sent to corresponding second network nodes.

[0029] In some embodiments of the present disclosure, the method further includes:

[0030] Receive and aggregate the task execution results sent by the at least two second network nodes.

[0031] In some embodiments of the present disclosure, the method further includes:

[0032] The aggregated task execution results are sent to the access and mobility management function AMF.

[0033] In some embodiments of the present disclosure, the method further includes:

[0034] Receive feedback on aggregated task execution results.

[0035] A second aspect of the present disclosure provides a method for scheduling tasks for an artificial intelligence (AI) network function service, the method being applied to a second network node, the method comprising:

[0036] Receiving subtasks of a target task sent by a first network node, wherein the subtasks are obtained by dividing the target task by the first network node according to task allocation weights in an optimal selection decision;

[0037] Execute the subtask according to the task model completed by local training;

[0038] Sending the task execution result to the first network node.

[0039] In some embodiments of the present disclosure, the method further includes:

[0040] Send node status information and a preset task type for the AI ​​network function service to the first network node, where the node status information includes at least CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information.

[0041] In some embodiments of the present disclosure, before executing the subtask based on the locally trained task model, the method further includes:

[0042] sending local model parameters to the first network node;

[0043] Receiving global model parameters sent by the first network node, wherein the global model parameters are obtained by weightedly aggregating the local model parameters according to local data size based on a federated average FedAvg algorithm after the first network node receives the local model parameters uploaded by the second network node included in the current action;

[0044] Based on the global model parameters and using a sample set matching the target task to iteratively train the task model, a locally trained task model is obtained.

[0045] In some embodiments of the present disclosure, performing the subtask based on the locally trained task model includes:

[0046] Retrieving structured data matching the subtask in a user data register UDR, and retrieving unstructured data matching the subtask in an unstructured data storage function UDSF;

[0047] The structured data and the unstructured data are input into a task model that has been trained locally, and the task execution result of the subtask is output using the task model that has been trained locally.

[0048] A third aspect of the present disclosure provides a method for scheduling tasks for an artificial intelligence (AI) network function service, the method being applied to a user equipment (UE), and the method including:

[0049] Send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF);

[0050] Receive the aggregated task execution result transparently transmitted by the access and mobility management function AMF.

[0051] In some embodiments of the present disclosure, the method further includes:

[0052] The structured data set of the target task corresponding to the AI ​​network function service establishment request is stored in the user data register UDR, and the unstructured data set of the target task corresponding to the AI ​​network function service establishment request is stored in the unstructured data storage function UDSF.

[0053] In some embodiments of the present disclosure, the method further includes:

[0054] Send a reception response message of the task execution result and feedback on the aggregated task execution result to the access and mobility management function AMF.

[0055] A fourth embodiment of the present disclosure provides a task scheduling method for an artificial intelligence (AI) network function service, the method being applied to an access and mobility management function (AMF), the method including:

[0056] Receiving an artificial intelligence (AI) network function service establishment request sent by a user equipment (UE);

[0057] Send the AI ​​network function service establishment request to the first network node.

[0058] In some embodiments of the present disclosure, the method further includes:

[0059] receiving the aggregated task execution result sent by the first network node;

[0060] The aggregated task execution result is transparently transmitted to the user equipment UE.

[0061] In some embodiments of the present disclosure, the method further includes:

[0062] Receive response messages for task execution results and feedback on aggregated task execution results;

[0063] The feedback on the aggregated task execution results is sent to the first network node.

[0064] A fifth embodiment of the present disclosure provides a task scheduling device for an artificial intelligence (AI) network function service, the device being applied to a first network node and including:

[0065] a screening module, configured to, in response to the artificial intelligence (AI) network function service establishment request, screen out at least two second network nodes from a plurality of second network nodes, wherein the node level of the plurality of second network nodes is lower than the node level of the first network node;

[0066] A sending module is used to send the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes.

[0067] A sixth aspect of the present disclosure provides a task scheduling device for an artificial intelligence (AI) network function service, the device being applied to a second network node and including:

[0068] a receiving module, configured to receive subtasks of a target task sent by a first network node, wherein the subtasks are obtained by dividing the target task by the first network node according to task allocation weights in an optimal selection decision;

[0069] An execution module, configured to execute the subtask according to the task model completed through local training;

[0070] A sending module is used to send the task execution result to the first network node.

[0071] A seventh aspect of the present disclosure provides a task scheduling device for an artificial intelligence (AI) network function service, the device being applied to a user equipment (UE), and comprising:

[0072] A sending module is used to send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF);

[0073] The receiving module is used to receive the aggregated task execution results transparently transmitted by the access and mobility management function AMF.

[0074] An eighth embodiment of the present disclosure provides a task scheduling device for an artificial intelligence (AI) network function service, the device being applied to an access and mobility management function (AMF), the device including:

[0075] A receiving module, configured to receive an artificial intelligence (AI) network function service establishment request sent by a user equipment (UE);

[0076] A sending module is used to send the AI ​​network function service establishment request to the first network node.

[0077] The ninth aspect embodiment of the present disclosure provides a communication device, which includes: a transceiver; a memory; and a processor, which is connected to the transceiver and the memory respectively, and is configured to control the wireless signal reception and transmission of the transceiver by executing computer-executable instructions on the memory, and can implement the method of the first aspect embodiment, the second aspect embodiment, the third aspect embodiment, or the fourth aspect embodiment of the present disclosure.

[0078] The tenth aspect embodiment of the present disclosure provides a computer storage medium, wherein the computer storage medium stores computer-executable instructions; after the computer-executable instructions are executed by the processor, the method of the first aspect embodiment, the second aspect embodiment, the third aspect embodiment, or the fourth aspect embodiment of the present disclosure can be implemented.

[0079] An eleventh embodiment of the present disclosure provides a communication system, comprising at least one of the following network elements: a task scheduling device for an AI network function service applied to a first network node as described above, and a task scheduling device for an AI network function service applied to a second network node as described above.

[0080] In some embodiments of the present disclosure, the communication system further includes an access and mobility management function AMF, and the AMF includes the above-mentioned task scheduling device applied to the access and mobility management function AMF.

[0081] The disclosed embodiments provide a method and device for scheduling tasks for an artificial intelligence (AI) network function service. The method can consider refining AI network functions, introducing a hierarchical relationship, and utilizing a high-level first network node to be responsible for signaling analysis of the AI ​​network function service, thereby realizing resource allocation and distribution scheduling for a low-level second network node. Specifically, the first network node can respond to an artificial intelligence (AI) network function service establishment request, screen out at least two second network nodes from a plurality of second network nodes for providing the corresponding AI network function service, and send the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes. By subdividing the artificial intelligence (AI) network function and hierarchical deployment of network nodes, the efficiency of task scheduling can be maximized, enabling AI services to be carried out efficiently and flexibly.

[0082] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The above and / or additional aspects and advantages of the present disclosure will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0084] FIG1 is a flow chart of a method for scheduling tasks of an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0085] FIG2 is a flowchart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0086] FIG3 is a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0087] FIG4 is a flowchart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0088] FIG5 is a flowchart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0089] FIG6 is a flowchart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0090] FIG7 is a flowchart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0091] FIG8 is a timing diagram of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0092] FIG9 is a block diagram of a task scheduling device for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0093] FIG10 is a block diagram of a task scheduling device for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0094] FIG11 is a block diagram of a task scheduling device for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0095] FIG12 is a block diagram of a task scheduling device for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure;

[0096] FIG13 is a schematic structural diagram of a communication device according to an embodiment of the present disclosure;

[0097] FIG14 is a schematic structural diagram of a chip provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0098] The following describes in detail embodiments of the present disclosure, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0099] Currently, numerous automated methods are being employed to improve operational efficiency in network planning, construction, maintenance, and optimization. However, the overall level of network autonomy is still low, leaving significant room for improvement. While software-defined networking (SDN) and network function virtualization (NFV) architectures offer networks a high degree of flexibility, they also increase complexity. Considerations for factors such as network resource allocation, transmission paths, and optimization algorithm design require more intelligent approaches. Artificial intelligence (AI) technology can help networks achieve a higher level of autonomy, reducing costs and increasing efficiency. Due to the relatively recent application of AI technology in communications networks, existing intelligent network applications are primarily plug-in applications that optimize and transform traditional network architectures. The lack of a common AI workflow and unified technical framework has led to fragmented network AI application scenarios and siloed R&D. Network AI functions are simply overlaid on existing network processes, and cross-domain and cross-layer intelligent application collaboration is difficult. The Network Data Analysis Function (NWDAF) collects data, performs analysis, and provides the results to other network functions. However, there is no detailed breakdown of the types of data analysis, nor is there any classification of specific AI algorithms.

[0100] To this end, the present disclosure proposes a task scheduling method and device for artificial intelligence (AI) network function services, which can subdivide artificial intelligence (AI) network functions according to specific algorithms and task types, thereby maximizing task scheduling efficiency and enabling AI services to be carried out efficiently and flexibly.

[0101] The switching method and device provided in this application are described in detail below with reference to the accompanying drawings.

[0102] Figure 1 shows a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. As shown in Figure 1 , the method is applied to a first network node and may include the following steps.

[0103] Step 101: In response to an artificial intelligence (AI) network function service establishment request, at least two second network nodes are selected from a plurality of second network nodes, wherein the node level of the plurality of second network nodes is lower than the node level of the first network node, and the node level is used to identify the AI ​​task management level of the network node.

[0104] In the embodiments of the present disclosure, it is possible to consider refining the AI ​​network function and introducing a hierarchical relationship. The first network node is a high-level AI management-level function in the core network, which can act as an AI task manager to perform task scheduling and communication computing resource allocation. It can be represented by AI0 in the following embodiments; one first network node can correspond to multiple second network nodes at the next level. The second network node is a low-level sub-AI network function in the core network, which can act as an AI service implementer. It can be represented by AI1\AI2\...\AIN in the following embodiments.

[0105] In response to an artificial intelligence (AI) network function service establishment request, the first network node may screen out multiple second network nodes that match the AI ​​network function service type from the subordinate second network nodes based on the AI ​​network function service type carried in the AI ​​network function service establishment request, and further screen out at least two second network nodes from the multiple second network nodes for jointly participating in task calculation and communication in this round of task execution, so as to maximize the overall reward function.

[0106] Step 102: Send the target task corresponding to the AI ​​network function service establishment request to at least two second network nodes.

[0107] In an embodiment of the present disclosure, after at least two second network nodes are screened out based on step 101 of the embodiment, when the first network node performs task allocation, the target task corresponding to the AI ​​network function service establishment request can be dispatched and scheduled, and the at least two second network nodes jointly complete the overall target task, and each second network node performs a part of the sub-AI network function of the target task. Among them, the target task can correspond to a fine-grained AI algorithm function service, including at least one of the following AI algorithm function services: classification, regression, clustering, etc., and can also provide personalized AI services according to user scenario requirements, such as image processing, speech recognition, machine translation, business recommendations, etc. In all embodiments of the present disclosure, the second network node under the first network node refers to a second network node whose node level is lower than that of the first network node.

[0108] In summary, according to the task scheduling method for the artificial intelligence (AI) network function service provided by the embodiment of the present disclosure, it is possible to consider refining the AI ​​network function, introducing a hierarchical relationship, and using the high-level first network node to be responsible for the signaling analysis of the AI ​​network function service to realize resource allocation and distribution scheduling of the low-level second network node. The first network node can specifically respond to the artificial intelligence (AI) network function service establishment request, select at least two second network nodes for providing the corresponding AI network function service from multiple second network nodes, and send the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes. By subdividing the artificial intelligence (AI) network function and the hierarchical deployment of network nodes, the task scheduling efficiency can be maximized, so that the AI ​​service can be carried out efficiently and flexibly.

[0109] Figure 2 shows a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to a first network node, based on the embodiment shown in Figure 1 , as shown in Figure 2 , and may include the following steps.

[0110] 201. Receive an artificial intelligence (AI) network function service establishment request sent by the access and mobility management function (AMF), where the AI ​​network function service establishment request carries an AI network function service type.

[0111] In an embodiment of the present disclosure, any user equipment (UE) that has completed the initial registration process and is connected to the core network can send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF) through the radio access network (RAN). The access and mobility management function (AMF) can further send an artificial intelligence (AI) network function service establishment request to the first network node, that is, the manager of the AI ​​service, to request the user equipment to provide the AI ​​network function service. Furthermore, the first network node can receive the artificial intelligence (AI) network function service establishment request sent by the access and mobility management function (AMF). The artificial intelligence (AI) network function service establishment request may include at least one of the following parameters: AI network function service type, AI network function service identifier, user equipment (UE) information, etc. Of course, those skilled in the art will understand that the AI ​​network function service establishment request may include other parameters, or a combination of the aforementioned parameters and other parameters, and the embodiments of the present disclosure are not limited to this.

[0112] 202. In response to an artificial intelligence (AI) network function service establishment request, determine multiple second network nodes that match the AI ​​network function service type, and receive node status information of the multiple second network nodes.

[0113] The node status information includes at least one of the following parameters: CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information.

[0114] In an embodiment of the present disclosure, before receiving an artificial intelligence (AI) network function service establishment request, the first network node may also receive preset task types for the AI ​​network function service uploaded by all subordinate second network nodes. Since the subordinate second network nodes of the first network node may cover multiple task types, when responding to the artificial intelligence (AI) network function service establishment request and determining multiple second network nodes that match the AI ​​network function service type, to save computing resources, the node may first screen out multiple second network nodes whose corresponding preset task types are the same as the AI ​​network function service type from all second network nodes. The node status information of the screened multiple second network nodes is then received, so that a more refined screening can be performed among the multiple second network nodes corresponding to the same AI network function service type based on the node status information.

[0115] 203. Based on the node status information and a preset deep reinforcement learning algorithm, select at least two second network nodes from the plurality of second network nodes.

[0116] Among them, the preset deep reinforcement learning algorithm can be a double deep Q learning (Double Deep Q Learning, DDQN) algorithm. By designing a reward mechanism, the first network node can find the strategy combination of the second network node corresponding to the maximum reward value, thereby reasonably performing task scheduling and resource allocation, and providing users with flexible and efficient AI services. It should be noted that the preset deep reinforcement learning algorithm can also use other implementable deep reinforcement learning algorithms. In this embodiment, the double deep Q learning DDQN algorithm is used as an example to illustrate the technical solution in this disclosure, but it does not constitute a specific limitation on the technical solution in this application.

[0117] Task scheduling for AI network function services primarily involves three steps: i) the first network node sends input data; ii) the second network node performs local training and computation; and iii) the first network node receives the output results. During this process, the two main objectives are: i) minimizing energy consumption; and ii) ensuring that as many second network nodes as possible participate in training. In embodiments of the present disclosure, the task scheduling requirements between AI network function services are modeled as optimization objectives and constraints, thereby seeking the optimal solution. The optimization metrics primarily include at least one of the following parameters: latency, energy consumption, benefits and costs, and physical equipment. This disclosure describes the modeling methods for optimization objectives and constraints from the perspectives of performance and energy consumption. Performance can include service latency and service deadlines. Service latency refers to the time between an application submitting a request and receiving a response. Service latency is a key metric for resource scheduling optimization. In this disclosure, service latency is categorized into two types: time spent on the compute node and time spent transmitting between nodes, namely, computational latency and communication latency. Through appropriate optimization strategies and task scheduling methods, service latency can be effectively reduced, thereby improving system performance. In addition to minimizing latency, task deadlines can also indicate the urgency of the task. Task completion deadlines can be categorized as hard or soft. Different tasks have varying degrees of latency sensitivity, and failure to meet deadlines can have serious consequences. These tasks are therefore defined as hard-deadline-constrained; otherwise, they are soft-deadline-constrained. Energy consumption is a major overhead in data centers, including power consumption from computing machines and cooling and heat dissipation equipment. Scheduling the target task corresponding to the AI ​​network function service establishment request to the second network node is crucial to ensuring the normal operation of the physical device hosting the second network node. Energy consumption primarily refers to battery power consumption in servers, mobile devices, and other devices, and is divided into four components: monitoring, computing, communication, and execution. Monitoring power consumption is related to the size and duration of its data packets; computing power consumption depends on the specific hardware parameters of the entity; communication power consumption is divided into uploading and receiving; and execution power consumption is positively correlated with the specific tasks and number of executions. Effectively conserving energy while maintaining system stability will be key considerations.

[0118] For time constraints, the total delay mainly includes the local model training time and parameter result upload time of the second network node. Since the downlink communication rate is much greater than the uplink rate, the time it takes for the first network node to send instructions to the second network node can be ignored. Each time a task is sent, each second network node processes its own task in parallel. and They respectively represent the model local training time and result upload time when the second network node executes the task. It depends on two things: i) the computation time; ii) the waiting time in the task queue of the second network node. The latter reflects the queuing time of the remaining workload in progress on the second network node. Therefore, It can be expressed as: D i Indicates the computing time of the current task. T i represents the waiting delay. Let y i ∈{0,1} is a binary variable, indicating whether the i-th second network node will execute the task in the current round, 1 means execution, 0 means no, that is, not participating in this task scheduling. The following conditions are met: y i ∈{0,1}, Because each second network node works in parallel, the total computing time and transmission time of each second network node to complete the task should meet the following upper limit: The heterogeneous computing capabilities of the second network node are reflected in The value is different, and because the amount of training task data carried by each second network node is different and the communication channel quality is also different, the time required to upload the result is There are also differences.

[0119] Regarding energy consumption, the energy consumption of each edge device includes two aspects: on the one hand, the energy consumption of uploading the model (result) from the second network node to the first network node, and on the other hand, the energy consumption of local model training of the second network node. The energy consumption of local training of the second network node depends on the time and space complexity of the specific AI algorithm and the size of the model parameters. Since there are differences in the tasks dispatched by the first network node each time, and the set of second network nodes involved in the execution of each task is also different, the amount of subtask data carried by each second network node is also significantly different, so the local training calculation energy consumption also varies, which is expressed here as This item is related to physical entity parameters such as clock frequency, operating power, etc. The total local training energy consumption of each second network node is: Assume that the minimum transmission time for sending 1 bit of information between the second network node and the first network node is Ti. Then the energy consumption related to uploading data is: where p i is the information transmission power, N i is the amount of result data uploaded by the second network node to the first network node. Therefore, the total energy consumption during the upload process is: Furthermore, the total energy consumption of each second network node in executing tasks in one round of iteration is E=E comp +E up .

[0120] In the embodiments of the present disclosure, the main goal is to enable more second network nodes to participate in training while meeting constraints, and to minimize system latency and energy consumption, thereby completing the AI ​​network function service faster and more efficiently. A deep reinforcement learning algorithm, namely the DDQN algorithm, is used here to help the first network node interact with the system environment and select a strategy that can obtain the maximum reward value. The specific steps of the embodiment may include: with the purpose of minimizing system latency and energy consumption and maximizing the number of second network nodes participating in the service, defining the state space as the node state information of the second network node, the action space as the combination of second network nodes selected for the AI ​​network function service establishment request, modeling a deep reinforcement learning model based on the DDQN algorithm; using the deep reinforcement learning model to determine the reward function for each action in the action set based on the node state information, and determining the optimal selection decision for the AI ​​network function service corresponding to the second network node based on the reward function, wherein the action is used to represent the selection decision for the AI ​​network function service corresponding to the second network node, and the optimal selection decision includes at least two second network nodes and the task allocation weights of at least two second network nodes.

[0121] Accordingly, when using a deep reinforcement learning model to determine the reward function of each action in the action set based on node state information, the overall algorithmic approach of federated learning can be adopted. The first network node uses a weighted average algorithm to aggregate and update the local model parameters trained by the second network nodes participating in each iteration, and then sends the global model to each second network node. The second network node continues to train with the updated model, and iterates multiple rounds until the requirements of the first network node are met. Because each task is different, and there are many second network nodes, each suitable for different task types. Therefore, each round needs to find a suitable second network node to participate in the training to maximize resource efficiency and the ultimate AI service quality. The steps of the embodiment may specifically include: using a deep reinforcement learning model to determine an action set based on node state information, and repeatedly executing a reward function determination process until the current action is determined to be the last action in the action set: The reward function determination process includes: taking the next action as the current action according to the action selection order in the action set; receiving local model parameters uploaded by the second network node included in the current action, and based on the federated average FedAvg algorithm, weightedly aggregating the local model parameters according to the local data size to obtain global model parameters; sending the global model parameters to the second network node included in the current action to continue training the task model of the second network node based on the global model parameters, and obtaining the reward function of the current action after the training is completed.

[0122] Among them, when determining the optimal selection decision of the target task with respect to the second network node based on the reward function, the steps of the embodiment may specifically include: determining the target action with the largest corresponding reward function in the action set, and determining the selection strategy represented by the target action as the optimal selection decision of the target task with respect to the second network node.

[0123] Specifically, when at least two second network nodes are selected from a plurality of second network nodes based on the node state information and a preset deep reinforcement learning algorithm, the following iterative process may be repeatedly performed until the overall task objective of the first network node is achieved, and the trained deep reinforcement learning model is used to output the optimal selection decision for the second network node corresponding to the target task:

[0124] Step 1: With the goal of minimizing system latency and energy consumption and maximizing the number of second network nodes participating in the service, the state space is defined as the node state information of the second network node, and the action space is the selection decision of the second network node corresponding to the target task. A deep reinforcement learning model is modeled based on the DDQN algorithm. The model parameters include: maximum number of iterations T, action set, decay factor γ, exploration rate ∈, Q function, number of samples m for batch gradient descent representing the Markov decision process, state S, action A, reward function R after executing action A, and next state S′ after executing action A;

[0125] In step 1, the state space is determined by the resource state information of the N second network nodes. The system state of all devices includes the remaining battery power, channel bandwidth, channel gain, power, etc. The state S is expressed as:

[0126]

[0127] Where S i Represents the state of the second network node, which can be expressed as S i ={f i ,e i ,r i ,c i ,t i}, where f i Represents the CPU computing frequency of the second network node, e i Represents the energy consumption of the second network node, r i Represents the wireless bandwidth situation, c i represents the channel state information, t i It is the type of preset tasks that can be performed;

[0128] The action space is a combination of selection strategies of the first network node, indicating which second network nodes it selects to participate in the current round of training tasks. Action A can be expressed as:

[0129]

[0130] Where A i ={0}∪{1} is the action state of the second network node. Ai=0 means that the second network node AIi does not participate in this round of updating the global model, and Ai=1 means that the second network node AIi participates in this round of model updating to train its local model;

[0131] The reward function R is the immediate reward obtained by the system when it performs behavior A in state S. The reward function should be proportional to the number of second network nodes participating in each round of tasks and inversely proportional to energy loss and training delay. The reward function is defined as follows:

[0132]

[0133] Where m is the number of second network nodes participating in this round, Emax is the total energy of the system, and T is the maximum delay of the number of second network nodes participating in this round of iteration.

[0134] The discount factor γ is a discount factor between 0 and 1. The farther the reward is from the current time step, the less important it is. γ = 0 means that the policy is short-sighted and only considers the current immediate reward Rt. Balancing the current time reward and future rewards can set γ to a larger value, such as γ = 0.9. If the state and environment are model-based, the future cumulative rewards can be obtained in advance without discounting. Here it is set to 0.99;

[0135] The Q function is the long-term reward, or action value function, which defines the expected reward R obtained after taking action A in state S and continuously executing the strategy. The first network node updates the Q value based on the experience replay mechanism:

[0136]

[0137] Where β is the learning rate and γ is the discount factor. After updating the Q value, the first network node can rely on the Q value to make a judgment. Based on any state S, the first network node can select the action A with the maximum cumulative reward R as the optimal choice decision for the target task with respect to the second network node.

[0138] Step 2: Initialize state S to the first state of the current state sequence and obtain its eigenvector φ(S);

[0139] Step 3: Use φ(S) as input in the network structure Q to obtain the Q-value output corresponding to all actions of the network structure Q, and use the ∈-greedy method to select the corresponding action A in the current Q-value output;

[0140] Step 4: Execute the current action A in state S. After the execution is completed, obtain the reward function R and the feature vector φ(S′) corresponding to the next state S′;

[0141] Step 5: Store the five-tuple {φ(S), A, R, φ(S′), end} into the experience replay set M;

[0142] Step 6: Let S = S′;

[0143] Step 7: Sample m samples {φ(Sj), Aj, Rj, φ(S′j), endj}, j = 1, 2…, m from the experience replay set M and calculate the current target Q value:

[0144] Step 8: Based on the current target Q value, the action value function weight θ of the Q network is updated by backpropagating the gradient of the neural network using the mean square error loss function.

[0145] The online neural network updates the weights θ based on the experience replay set M, and the target neural network periodically resets the weights θ'=θ using the gradient descent algorithm. The mean square error loss function can be defined as:

[0146] Lθ=E[yQ(s,a;θ)] 2

[0147] The current target Q value y is defined as:

[0148] y=R(s,a)+γQ'(s',argmaxQ(s',a';θ);θ')

[0149] Step 9: If S′ is in the terminal state, the current round of iteration is completed, otherwise go to step 3;

[0150] Step 10: Iterate steps 2 to 9 until the overall task goal of the first network node is reached, and use the reward function R of each action A in the action set to determine the optimal selection decision for the target task on the second network node.

[0151] The optimal selection decision includes at least two second network nodes and task allocation weights of at least two second network nodes.

[0152] 204. Divide the target task into at least two subtasks according to the task allocation weights of the at least two second network nodes, and send the subtasks to the corresponding second network nodes respectively.

[0153] In an embodiment of the present disclosure, after determining the optimal selection decision for the AI ​​network function service corresponding to the second network node, the target task can be divided into at least two subtasks based on the task allocation weights of at least two second network nodes in the optimal selection decision, and each subtask can be sent to the corresponding second network node. For example, if it is determined that the optimal selection decision for the AI ​​network function service corresponding to the second network node includes second network nodes a, b, and c, and the task allocation weights corresponding to second network nodes a, b, and c are 20%, 50%, and 30%, respectively, the complete target task can be divided into subtask 1, subtask 2, and subtask 3 based on the task allocation weights, where subtask 1 accounts for 20% of the total target task, subtask 2 accounts for 50% of the total target task, and subtask 3 accounts for 30% of the total target task. Subtasks 1, 2, and 3 do not overlap and each corresponds to a portion of the target task. Subtask 1 can then be sent to second network node a, subtask 2 to second network node b, and subtask 3 to second network node c, so that the target task can be jointly completed by second network nodes a, b, and c.

[0154] In summary, according to the task scheduling method of the artificial intelligence (AI) network function service provided by the embodiment of the present disclosure, it is possible to consider refining the AI ​​network function and introducing a hierarchical relationship. By introducing a deep reinforcement learning algorithm in the task scheduling process of the network function service, and based on the reward mechanism of the deep reinforcement learning algorithm, the high-level first network node can find the low-level second network node strategy combination corresponding to the maximum reward value, thereby reasonably allocating task resources and maximizing task scheduling efficiency, so that the AI ​​service can be carried out efficiently and flexibly.

[0155] Figure 3 shows a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to a first network node, based on the embodiments shown in Figures 1 and 2, as shown in Figure 3, and may include the following steps.

[0156] 301. Receive the artificial intelligence (AI) network function service establishment request sent by the access and mobility management function (AMF).

[0157] In the embodiment of the present disclosure, the implementation process is the same as step 201 of the embodiment, and will not be repeated here.

[0158] 302. In response to the artificial intelligence (AI) network function service establishment request, select at least two second network nodes from multiple second network nodes.

[0159] In the embodiment of the present disclosure, the implementation process can be referred to steps 202 to 203 of the embodiment, which will not be repeated here.

[0160] 303. Send the target task corresponding to the AI ​​network function service establishment request to at least two second network nodes.

[0161] In the embodiment of the present disclosure, the implementation process can be referred to step 204 of the embodiment, which will not be described in detail here.

[0162] 304. Receive and aggregate task execution results sent by at least two second network nodes.

[0163] In the embodiments of the present disclosure, at least two second network nodes jointly complete the target task, with each second network node responsible for a portion of the subtask. Therefore, after the first network node receives the task execution results sent by the at least two second network nodes, it can aggregate the task execution results sent by the at least two second network nodes to obtain the complete task execution result for the target task. This aggregation process can structure the task execution results to meet integrity requirements.

[0164] 305. Send the aggregated task execution results to the access and mobility management function AMF.

[0165] In an embodiment of the present disclosure, the complete task execution result of the aggregated target task can be sent to the Access and Mobility Management Function (AMF), so that the task execution result can be transparently transmitted to the user equipment UE that sends the artificial intelligence AI network function service establishment request using AMF. Furthermore, after receiving the aggregated task execution result, the user equipment UE can also provide feedback based on the task execution result, and send the feedback to the AMF, which is then sent to the first network node using AMF. Accordingly, the first network node can also receive feedback on the aggregated task execution result, and further adjust and optimize the task scheduling strategy based on the feedback, so that task scheduling can better meet the personalized needs of users.

[0166] In summary, according to the task scheduling method for artificial intelligence (AI) network function services provided by the embodiments of the present disclosure, a first network node can respond to an AI network function service establishment request by selecting at least two second network nodes from multiple second network nodes to provide the corresponding AI network function service, send the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes, and utilize the at least two second network nodes to jointly execute the target task. This allows for reasonable task resource allocation, maximizes task scheduling efficiency, and enables AI services to be performed efficiently and flexibly.

[0167] Figure 4 is a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to the second network node and may include the following steps.

[0168] 401. Receive subtasks related to a target task sent by a first network node, wherein the subtasks are obtained by dividing the target task by the first network node according to task allocation weights in an optimal selection decision.

[0169] In an embodiment of the present disclosure, before executing the steps of this embodiment, the second network node may also upload node status information and preset task types for AI network function services to the first network node in real time, so as to facilitate the first network node to allocate and call tasks corresponding to the current second network node. The node status information includes at least one of the following parameters: CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information. Furthermore, in response to the first network node screening out the current second network node for executing the target task together with other screened second network nodes, the current second network node may receive subtasks of the target task sent by the first network node.

[0170] 402. Execute subtasks based on the task model completed through local training.

[0171] In an embodiment of the present disclosure, before executing the steps of this embodiment, the local task model needs to be pre-trained. Specifically, in the process of the first network node determining the optimal selection decision based on the deep reinforcement learning model, when the selected action includes the current second network node, the second network node sends the local model parameters to the first network node, and further receives the global model parameters issued by the first network node, wherein the global model parameters are obtained by weightedly aggregating the local model parameters according to the local data size based on the federated average FedAvg algorithm after the first network node receives the local model parameters uploaded by the second network node included in the current action; finally, the task model can be iteratively trained based on the global model parameters and using a sample set matching the target task to obtain a task model that has been locally trained.

[0172] In a specific application scenario, when the user equipment UE registers and sends an artificial intelligence AI network function service establishment request, it will also store a structured data set of the target task corresponding to the AI ​​network function service establishment request in the user data register (User Data Repository, UDR), and store an unstructured data set of the target task corresponding to the AI ​​network function service establishment request in the unstructured data storage function (Unstructured Data Storage Network Function, UDSF). Accordingly, for this embodiment, when executing a subtask based on a task model completed by local training, the embodiment steps may specifically include: retrieving structured data that matches the subtask in the user data register UDR, and retrieving unstructured data that matches the subtask in the unstructured data storage function UDSF; inputting the structured data and unstructured data into the task model completed by local training, and using the task model completed by local training to output the task execution result of the subtask.

[0173] 403. Send the task execution result to the first network node.

[0174] In summary, according to the task scheduling method of the artificial intelligence AI network function service provided by the embodiment of the present disclosure, it is possible to consider refining the AI ​​network function, introducing a hierarchical relationship, and using the high-level first network node to be responsible for the signaling analysis of the AI ​​network function service to realize resource allocation and distribution scheduling of the low-level second network node. The second network node can receive the subtasks of the target task sent by the first network node, perform the subtasks using the task model completed by local local training, and jointly complete the target task with other second network nodes selected by the first network node. By subdividing the artificial intelligence AI network function, the task scheduling efficiency can be maximized, so that the AI ​​service can be carried out efficiently and flexibly.

[0175] Figure 5 is a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to a user equipment (UE), and the method may include the following steps.

[0176] 501. Send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF).

[0177] In an embodiment of the present disclosure, a user equipment (UE) may first perform a registration process to connect to the core network. After completing the registration, it may send an artificial intelligence (AI) network function service establishment request to an access and mobility management function (AMF) through a radio access network (RAN), so as to utilize the access and mobility management function AMF to send an artificial intelligence (AI) network function service establishment request to a first network node, i.e., the manager of the AI ​​service, to request the user equipment to provide the AI ​​network function service. In addition, when registering and sending an artificial intelligence (AI) network function service establishment request, the user equipment UE will also simultaneously store a structured data set of a target task corresponding to the AI ​​network function service establishment request in a user data register (UDR), and store an unstructured data set of a target task corresponding to the AI ​​network function service establishment request in an unstructured data storage function (UDSF). So that when the first network node subsequently schedules at least two second network nodes to respectively execute subtasks corresponding to the target task, each second network node can retrieve structured data matching the subtask in the user data register UDR and retrieve unstructured data matching the subtask in the unstructured data storage function UDSF.

[0178] 502. Receive the aggregated task execution result transparently transmitted by the access and mobility management function AMF.

[0179] In an embodiment of the present disclosure, when a first network node schedules at least two second network nodes to respectively execute subtasks corresponding to a target task and receives task execution results fed back by the at least two second network nodes, the task execution results sent by the at least two second network nodes may be aggregated to obtain a complete task execution result of the target task. The aggregated complete task execution result of the target task is then sent to the access and mobility management function (AMF). The AMF transparently transmits the complete task execution result to the UE via the radio access network (RAN). Furthermore, the user equipment (UE) that sends the artificial intelligence (AI) network function service establishment request may receive the task execution result transparently transmitted by the mobility management function (AMF).

[0180] As an optional method, after receiving the task execution result transparently transmitted by the mobility management function AMF, the user equipment UE can send a reception response message of the task execution result to the mobility management function AMF, and can also send feedback on the aggregated task execution result.

[0181] In summary, according to the task scheduling method for the artificial intelligence AI network function service provided by the embodiment of the present disclosure, the user equipment UE can use the mobility management function AMF to implement data interaction with the first network node, so that the first network node can determine the specific AI task type according to the artificial intelligence AI network function service establishment request of the user equipment UE, and select the corresponding AI network function, so that task resources can be reasonably allocated and fine-grained AI network function services can be provided.

[0182] Figure 6 is a flowchart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to the access and mobility management function (AMF), and the method may include the following steps.

[0183] 601. Receive an artificial intelligence (AI) network function service establishment request sent by a user equipment (UE).

[0184] 602. Send an AI network function service establishment request to the first network node.

[0185] In an embodiment of the present disclosure, after receiving an artificial intelligence (AI) network function service establishment request from a user equipment (UE), the access and mobility management function (AMF) may send the AI ​​network function service establishment request to a first network node so that the first network node can exchange information with the user equipment (UE). Specifically, in response to the AI ​​network function service establishment request, the first network node may select multiple second network nodes matching the AI ​​network function service type from subordinate second network nodes based on the AI ​​network function service type carried in the AI ​​network function service establishment request. Furthermore, the first network node may select at least two second network nodes from the multiple second network nodes to participate in task computation and communication during the current round of task execution. The first network node may further transmit the target task corresponding to the AI ​​network function service establishment request to at least two second network nodes. The two second network nodes jointly complete the overall target task, with each second network node performing a portion of the sub-AI network function of the target task. The target task may correspond to a fine-grained AI algorithm function service, including at least one of the following AI algorithm function services: classification, regression, clustering, etc. Personalized AI services such as image processing, speech recognition, machine translation, and business recommendations may also be provided based on user scenario requirements.

[0186] In summary, according to the task scheduling method for the artificial intelligence (AI) network function service provided by the embodiment of the present disclosure, it is possible to consider refining the AI ​​network function, introducing a hierarchical relationship, and using the high-level first network node to be responsible for signaling analysis of the AI ​​network function service to implement resource allocation and distribution scheduling for the low-level second network node. After receiving the artificial intelligence (AI) network function service establishment request sent by the user equipment (UE), the access and mobility management function (AMF) may send the AI ​​network function service establishment request to the first network node to further utilize the first network node to reasonably allocate task resources, thereby maximizing task scheduling efficiency and enabling AI services to be performed efficiently and flexibly.

[0187] Figure 7 is a flow chart of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to the access and mobility management function (AMF). Based on the embodiment shown in Figure 6, as shown in Figure 7, the method may include the following steps.

[0188] 701. Receive aggregated task execution results sent by a first network node.

[0189] 702. Transparently transmit the aggregated task execution results to the user equipment UE.

[0190] In an embodiment of the present disclosure, upon receiving task execution results for some subtasks of a target task from at least two scheduled second network nodes, the first network node may aggregate the task execution results of the subtasks and send the aggregated task execution results to the access and mobility management function (AMF). Furthermore, upon receiving the aggregated task execution results sent by the first network node, the access and mobility management function (AMF) may transparently transmit the aggregated task execution results to the user equipment (UE).

[0191] As an optional method, after transparently transmitting the aggregated task execution results to the user equipment UE, the access and mobility management function (AMF) may also receive a task execution result reception response message from the user equipment UE, as well as feedback regarding the aggregated task execution results. Furthermore, the access and mobility management function (AMF) may send the feedback to the first network node, so that the first network node can adjust and optimize its scheduling policy based on the feedback.

[0192] In summary, according to the task scheduling method of the artificial intelligence AI network function service provided by the embodiment of the present disclosure, after receiving the aggregated task execution result sent by the first network node, the mobility management function AMF can transparently transmit the aggregated task execution result to the user equipment UE, and send the feedback sent by the user equipment UE on the aggregated task execution result to the first network node, thereby realizing data interaction between the user equipment UE and the first network node regarding the task execution result, and facilitating the first network node to adjust and optimize its own scheduling strategy.

[0193] Figure 8 is a sequence diagram of a task scheduling method for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The method is applied to a communication system, comprising: a task scheduling device for an AI network function service applied to a first network node, a task scheduling device for an AI network function service applied to a second network node, and a task scheduling device applied to an access and mobility management function (AMF). In the communication system, a user equipment (UE) sends an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF); the task scheduling device applied to the access and mobility management function (AMF) sends the AI ​​network function service establishment request to the first network node; the task scheduling device applied to the AI ​​network function service applied to the first network node, in response to the artificial intelligence (AI) network function service establishment request, selects at least two second network nodes from a plurality of second network nodes, and sends a target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes; the task scheduling devices applied to the at least two second network nodes receive subtasks of the target task sent by the first network node, execute the subtasks according to a locally trained task model, and then send the task execution results to the first network node.

[0194] 8 , when the communication system is executed, it may include the following steps.

[0195] 801. The user equipment UE sends an AI network function service establishment request to the access and mobility management function AMF.

[0196] Among them, the AI ​​network function service establishment request may include the AI ​​network function service type, AI network function service identifier, user equipment UE information, etc.

[0197] 802. The access and mobility management function AMF sends an AI network function service establishment request to the first network node.

[0198] 803. The first network node determines, in response to the artificial intelligence (AI) network function service establishment request, multiple second network nodes that match the AI ​​network function service type, and receives node status information of the multiple second network nodes.

[0199] The node status information includes at least one of the following parameters: CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information.

[0200] 804. The first network node selects at least two second network nodes from the plurality of second network nodes based on the node status information and a preset deep reinforcement learning algorithm.

[0201] Among them, the preset deep reinforcement learning algorithm can be the Double Deep Q Learning (DDQN) algorithm. By designing a reward mechanism, the first network node can find the strategy combination of the second network node corresponding to the maximum reward value, thereby reasonably performing task scheduling and resource allocation, and providing users with flexible and efficient AI services.

[0202] 805. The first network node divides the target task into at least two subtasks according to a preset deep reinforcement learning algorithm, and sends the subtasks to corresponding second network nodes respectively.

[0203] 806. At least the second network node obtains a structured data set from a user data register UDR, and obtains an unstructured data set from an unstructured data storage function UDSF.

[0204] When registering and sending an artificial intelligence (AI) network function service establishment request, the user equipment (UE) will also simultaneously store a structured data set of the target task corresponding to the AI ​​network function service establishment request in the user data repository (UDR), and store an unstructured data set of the target task corresponding to the AI ​​network function service establishment request in the unstructured data storage function (UDSF). This allows at least two second network nodes to retrieve structured data matching the subtask from the user data register (UDR) and unstructured data matching the subtask from the unstructured data storage function (UDSF) when performing subtasks based on the locally trained task model.

[0205] 807. At least two second network nodes respectively train local task models, execute subtasks according to the locally trained task models, and send task execution results to the first network node.

[0206] 808. The first network node receives and aggregates task execution results sent by at least two second network nodes.

[0207] 809. The first network node sends the aggregated task execution result to the access and mobility management function AMF.

[0208] 810. The access and mobility management function AMF transparently transmits the aggregated task execution results to the user equipment UE.

[0209] 811. The user equipment UE sends a response message to the mobility management function AMF indicating that the task execution result has been received, and sends feedback on the task execution result.

[0210] 812. The access and mobility management function AMF sends feedback from the user equipment UE on the task execution result to the first network node.

[0211] By applying the task scheduling method for the artificial intelligence (AI) network function service provided in this embodiment, it is possible to consider refining the AI ​​network function, introducing a hierarchical relationship, and using the high-level first network node to be responsible for signaling analysis of the AI ​​network function service to achieve resource allocation and distribution scheduling for the low-level second network node. By having the first network node respond to the artificial intelligence (AI) network function service establishment request and screen out at least two second network nodes from multiple second network nodes for providing the corresponding AI network function service, resource allocation and distribution deployment of the AI ​​network function service can be achieved. By subdividing the artificial intelligence (AI) network function, the task scheduling efficiency can be maximized, allowing the AI ​​service to be carried out efficiently and flexibly.

[0212] In the embodiments provided above, the methods provided in the embodiments of the present application are described from the perspectives of the first network node, the second network node, the user equipment UE, and the mobility management function AMF. In order to implement the functions of the methods provided in the embodiments of the present application, the first network node, the second network node, the user equipment UE, and the mobility management function AMF may include hardware structures and software modules, and implement the above functions in the form of hardware structures, software modules, or hardware structures plus software modules. One of the above functions may be executed in the form of hardware structures, software modules, or hardware structures plus software modules.

[0213] Corresponding to the task scheduling of the artificial intelligence (AI) network function service provided in the above-mentioned embodiments, the present disclosure also provides a task scheduling device for the artificial intelligence (AI) network function service. Since the task scheduling device for the artificial intelligence (AI) network function service provided in the embodiment of the present disclosure corresponds to the task scheduling method for the artificial intelligence (AI) network function service provided in the above-mentioned embodiments, the implementation method of the task scheduling method for the artificial intelligence (AI) network function service is also applicable to the task scheduling device for the artificial intelligence (AI) network function service provided in this embodiment, and will not be described in detail in this embodiment.

[0214] FIG9 is a structural diagram of a task scheduling device 900 for an artificial intelligence (AI) network function service provided according to an embodiment of the present disclosure. The task scheduling device 900 for an artificial intelligence (AI) network function service can be used for a first network node.

[0215] As shown in FIG9 , the apparatus 900 may include:

[0216] The screening module 910 may be configured to screen out at least two second network nodes from a plurality of second network nodes in response to an artificial intelligence (AI) network function service establishment request, wherein the node level of the plurality of second network nodes is lower than the node level of the first network node;

[0217] The sending module 920 can be used to send the target task corresponding to the AI ​​network function service establishment request to at least two second network nodes.

[0218] In some embodiments of the present disclosure, as shown in FIG9 , the apparatus 900 further includes: a receiving module 930 ;

[0219] The receiving module 930 can be used to receive an artificial intelligence AI network function service establishment request sent by the access and mobility management function AMF.

[0220] In some embodiments of the present disclosure, the AI ​​network function service establishment request carries the AI ​​network function service type. When at least two second network nodes are screened out from multiple second network nodes, the screening module 910 can be used to determine multiple second network nodes that match the AI ​​network function service type, and receive node status information of the multiple second network nodes, where the node status information includes at least CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information; based on the node status information and a preset deep reinforcement learning algorithm, at least two second network nodes are screened out from the multiple second network nodes.

[0221] In some embodiments of the present disclosure, the receiving module 930 may be configured to receive a preset task type regarding an AI network function service uploaded by a second network node; and the screening module 910 may be configured to determine, based on the preset task type, multiple second network nodes that match the AI ​​network function service type.

[0222] In some embodiments of the present disclosure, the preset deep reinforcement learning algorithm includes a dual deep Q learning DDQN algorithm. When at least two second network nodes are screened out from multiple second network nodes based on the node status information and the preset deep reinforcement learning algorithm, the screening module 910 can be used to define the state space as the node status information of the second network node, the action space as the second network node combination selected for the AI ​​network function service establishment request, with the purpose of minimizing system latency and energy consumption and maximizing the number of second network nodes participating in the service, and model a deep reinforcement learning model based on the DDQN algorithm; use the deep reinforcement learning model to determine the reward function of each action in the action set according to the node status information, and determine the optimal selection decision of the AI ​​network function service corresponding to the second network node based on the reward function, wherein the action is used to represent the selection decision of the AI ​​network function service corresponding to the second network node, and the optimal selection decision includes at least two second network nodes and the task allocation weights of at least two second network nodes.

[0223] In some embodiments of the present disclosure, when using a deep reinforcement learning model to determine the reward function of each action in an action set based on node state information, the screening module 910 can be used to use the deep reinforcement learning model to determine the action set based on the node state information, and repeatedly execute the reward function determination process until the current action is determined to be the last action in the action set: the reward function determination process includes: taking the next action as the current action according to the action selection order in the action set; receiving the local model parameters uploaded by the second network node included in the current action, and based on the federated average FedAvg algorithm, weightedly aggregating the local model parameters according to the local data size to obtain the global model parameters; sending the global model parameters to the second network node included in the current action to continue training the task model of the second network node based on the global model parameters, and obtaining the reward function of the current action after the training is completed.

[0224] In some embodiments of the present disclosure, when determining the optimal selection decision for the target task with respect to the second network node based on the reward function, the screening module 910 can be used to determine the target action with the largest corresponding reward function in the action set, and determine the selection strategy represented by the target action as the optimal selection decision for the target task with respect to the second network node.

[0225] In some embodiments of the present disclosure, the sending module 920 may be configured to divide the target task into at least two subtasks according to the task allocation weights of at least two second network nodes; and send the subtasks to the corresponding second network nodes respectively.

[0226] In some embodiments of the present disclosure, as shown in FIG9 , the apparatus 900 further includes: an aggregation module 940 ;

[0227] The aggregation module 940 may be configured to receive and aggregate task execution results sent by at least two second network nodes.

[0228] In some embodiments of the present disclosure, the sending module 920 may be configured to send the aggregated task execution results to the access and mobility management function AMF.

[0229] In some embodiments of the present disclosure, the receiving module 930 may be configured to receive feedback regarding aggregated task execution results.

[0230] Figure 10 is a schematic diagram of the structure of an artificial intelligence (AI) network function service task scheduling device 1000 provided by an embodiment of the present disclosure. The artificial intelligence (AI) network function service task scheduling device 1000 can be used for the second network node.

[0231] As shown in FIG10 , the apparatus 1000 may include:

[0232] The receiving module 1010 may be configured to receive subtasks of a target task sent by the first network node, wherein the subtasks are obtained by dividing the target task by the first network node according to the task allocation weights in the optimal selection decision;

[0233] An execution module 1020 may be used to execute subtasks based on the locally trained task model;

[0234] The sending module 1030 may be configured to send the task execution result to the first network node.

[0235] In some embodiments of the present disclosure, the sending module 1030 can also be used to send node status information and preset task types for AI network function services to the first network node. The node status information includes at least CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information.

[0236] In some embodiments of the present disclosure, as shown in FIG10 , the apparatus 1000 further includes: a training module 1040;

[0237] The sending module 1030 can also be used to send local model parameters to the first network node; the receiving module 1010 can also be used to receive global model parameters issued by the first network node, wherein the global model parameters are obtained by the first network node receiving the local model parameters uploaded by the second network node included in the current action, and aggregating the local model parameters weighted according to the local data size based on the federated average FedAvg algorithm; the training module 1040 can be used to iteratively train the task model based on the global model parameters and using a sample set matching the target task to obtain a task model that has been locally trained.

[0238] In some embodiments of the present disclosure, the execution module 1020 can be used to call structured data matching the subtask in the user data register UDR, and call unstructured data matching the subtask in the unstructured data storage function UDSF; input the structured data and unstructured data into the task model that has been trained locally, and use the task model that has been trained locally to output the task execution result of the subtask.

[0239] FIG11 is a schematic diagram of a task scheduling apparatus 1100 for an artificial intelligence (AI) network function service according to an embodiment of the present disclosure. The task scheduling apparatus 1100 for an artificial intelligence (AI) network function service may be used in a user equipment (UE).

[0240] As shown in FIG11 , the apparatus 1100 may include:

[0241] The sending module 1110 may be configured to send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF);

[0242] The receiving module 1120 can be used to receive the aggregated task execution results transparently transmitted by the access and mobility management function AMF.

[0243] In some embodiments of the present disclosure, as shown in FIG11 , the apparatus 1100 may further include: a storage module 1130 ;

[0244] The storage module 1130 can be used to store the structured data set of the target task corresponding to the AI ​​network function service establishment request in the user data register UDR, and to store the unstructured data set of the target task corresponding to the AI ​​network function service establishment request in the unstructured data storage function UDSF.

[0245] In some embodiments of the present disclosure, the sending module 1110 may also be used to send a reception response message of the task execution result and feedback on the aggregated task execution result to the access and mobility management function AMF.

[0246] Figure 12 is a schematic diagram of the structure of a task scheduling device 1100 of an artificial intelligence (AI) network function service provided by an embodiment of the present disclosure. The task scheduling device 1200 of the artificial intelligence (AI) network function service can be used for access and mobility management function (AMF).

[0247] As shown in FIG12 , the apparatus 1200 may include:

[0248] The receiving module 1210 may be configured to receive an artificial intelligence (AI) network function service establishment request sent by a user equipment (UE);

[0249] The sending module 1220 may be configured to send an AI network function service establishment request to the first network node.

[0250] In some embodiments of the present disclosure, as shown in FIG12 , the apparatus 1200 may further include: a transparent transmission module 1230 ;

[0251] The receiving module 1210 may be further configured to receive the aggregated task execution result sent by the first network node; and the transparent transmission module 1230 may be configured to transparently transmit the aggregated task execution result to the user equipment UE.

[0252] In some embodiments of the present disclosure, the receiving module 1210 can also be used to receive a reception response message of the task execution result and feedback on the aggregated task execution result; the sending module 1220 can also be used to send the feedback on the aggregated task execution result to the first network node.

[0253] Please refer to Figure 13, which is a schematic diagram of the structure of a communication device 1300 provided in an embodiment of the present application. Communication device 1300 can be a network device or a user device, or a chip, chip system, or processor that supports the network device to implement the above method. It can also be a chip, chip system, or processor that supports the user device to implement the above method. This device can be used to implement the method described in the above method embodiment. For details, please refer to the description of the above method embodiment.

[0254] The communication device 1300 may include one or more processors 1301. The processor 1301 may be a general-purpose processor or a dedicated processor. For example, it may be a baseband processor or a central processing unit. The baseband processor may be used to process communication protocols and communication data, and the central processing unit may be used to control the communication device (e.g., a base station, a baseband chip, a terminal device, a terminal device chip, a DU or CU, etc.), execute computer programs, and process computer program data.

[0255] Optionally, the communication device 1300 may further include one or more memories 1302, on which a computer program 1304 may be stored. The processor 1301 executes the computer program 1304 to cause the communication device 1300 to perform the method described in the above method embodiment. Optionally, the memory 1302 may also store data. The communication device 1300 and the memory 1302 may be provided separately or integrated together.

[0256] Optionally, the communication device 1300 may further include a transceiver 1305 and an antenna 1306. The transceiver 1305 may be referred to as a transceiver unit, a transceiver, or a transceiver circuit, and is configured to implement transceiver functions. The transceiver 1305 may include a receiver and a transmitter. The receiver may be referred to as a receiver or a receiving circuit, and is configured to implement a receiving function; the transmitter may be referred to as a transmitter or a transmitting circuit, and is configured to implement a transmitting function.

[0257] Optionally, the communication device 1300 may further include one or more interface circuits 1307. The interface circuit 1307 is configured to receive code instructions and transmit the instructions to the processor 1301. The processor 1301 executes the code instructions to enable the communication device 1300 to perform the method described in the above method embodiment.

[0258] In one implementation, processor 1301 may include a transceiver for implementing receiving and transmitting functions. For example, the transceiver may be a transceiver circuit, an interface, or an interface circuit. The transceiver circuit, interface, or interface circuit for implementing the receiving and transmitting functions may be separate or integrated. The transceiver circuit, interface, or interface circuit may be used for reading and writing code / data, or may be used for transmitting or delivering signals.

[0259] In one implementation, processor 1301 may store a computer program 1303. Computer program 1303, when executed on processor 1301, enables communication device 1300 to perform the method described in the above method embodiment. Computer program 1303 may be embedded in processor 1301, in which case processor 1301 may be implemented by hardware.

[0260] In one implementation, the communication device 1300 may include a circuit that can implement the functions of sending, receiving, or communicating in the aforementioned method embodiments. The processor and transceiver described in this application can be implemented on an integrated circuit (IC), an analog IC, a radio frequency integrated circuit RFIC, a mixed signal IC, an application specific integrated circuit (ASIC), a printed circuit board (PCB), an electronic device, etc. The processor and transceiver can also be manufactured using various IC process technologies, such as complementary metal oxide semiconductor (CMOS), N-type metal oxide semiconductor (nMetal-oxide-semiconductor, NMOS), P-type metal oxide semiconductor (positive channel metal oxide semiconductor, PMOS), bipolar junction transistor (bipolar junction transistor, BJT), bipolar CMOS (BiCMOS), silicon germanium (SiGe), gallium arsenide (GaAs), etc.

[0261] The communication device described in the above embodiments may be a network device or a user device, but the scope of the communication device described in this application is not limited thereto, and the structure of the communication device may not be limited to FIG13. The communication device may be an independent device or may be part of a larger device. For example, the communication device may be:

[0262] (1) An independent integrated circuit (IC), or chip, or chip system or subsystem;

[0263] (2) a collection of one or more ICs, optionally including a storage component for storing data and computer programs;

[0264] (3) ASIC, such as modem;

[0265] (4) Modules that can be embedded in other devices;

[0266] (5) Receivers, terminal devices, intelligent terminal devices, cellular phones, wireless devices, handheld devices, mobile units, vehicle-mounted devices, network devices, cloud devices, artificial intelligence devices, etc.;

[0267] (6)Others, etc.

[0268] If the communication device can be a chip or a chip system, please refer to the schematic diagram of the chip structure shown in Figure 14. The chip shown in Figure 14 includes a processor 1401 and an interface 1402. The number of processors 1401 can be one or more, and the number of interfaces 1402 can be multiple.

[0269] Optionally, the chip further includes a memory 1403, which is used to store necessary computer programs and data.

[0270] Those skilled in the art will also appreciate that the various illustrative logical blocks and steps listed in the embodiments of the present application can be implemented by electronic hardware, computer software, or a combination of both. Whether such functions are implemented by hardware or software depends on the specific application and the design requirements of the entire system. Those skilled in the art may use various methods to implement the functions for each specific application, but such implementation should not be construed as exceeding the scope of protection of the embodiments of the present application.

[0271] The present application also provides a readable storage medium having instructions stored thereon, which implement the functions of any of the above method embodiments when executed by a computer.

[0272] The present application also provides a computer program product, which implements the functions of any of the above method embodiments when executed by a computer.

[0273] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer programs. When the computer program is loaded and executed on a computer, all or part of the processes or functions according to the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer program can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer program can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated therein. Available media may be magnetic media (eg, floppy disks, hard disks, tapes), optical media (eg, high-density digital video discs (DVDs)), or semiconductor media (eg, solid state disks (SSDs)).

[0274] Those skilled in the art will understand that the various numerical numbers such as first and second involved in this application are only for the convenience of description and are not used to limit the scope of the embodiments of this application, and also indicate the order of precedence.

[0275] In this application, at least one can also be described as one or more, and multiple can be two, three, four or more, which is not limited in this application. In the embodiments of this application, for a technical feature, the technical features in the technical feature are distinguished by "first", "second", "third", "A", "B", "C" and "D", and there is no order of precedence or size between the technical features described by "first", "second", "third", "A", "B", "C" and "D".

[0276] As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and / or device (e.g., a magnetic disk, an optical disk, a memory, a programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0277] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0278] Computer systems may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The client and server relationship arises through computer programs running on the respective computers and having a client-server relationship to each other.

[0279] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0280] Furthermore, it should be understood that the various embodiments of the present application may be implemented individually or in combination with other embodiments where the solution permits.

[0281] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0282] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0283] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A task scheduling method for artificial intelligence (AI) network function services, characterized in that: The method is applied to a first network node, and includes: In response to the artificial intelligence (AI) network function service establishment request, screening at least two second network nodes from a plurality of second network nodes, wherein the node level of the plurality of second network nodes is lower than the node level of the first network node; Sending the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes.

2. The method according to claim 1, characterized in that The method further comprises: Receive the artificial intelligence AI network function service establishment request sent by the access and mobility management function AMF.

3. The method according to claim 1, characterized in that The AI ​​network function service establishment request carries the AI ​​network function service type, and the selecting at least two second network nodes from the plurality of second network nodes includes: Determine multiple second network nodes that match the AI ​​network function service type, and receive node status information of the multiple second network nodes, where the node status information includes at least CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information; Based on the node status information and a preset deep reinforcement learning algorithm, at least two second network nodes are selected from the plurality of second network nodes.

4. The method according to claim 3, characterized in that The method further comprises: Receiving a preset task type regarding an AI network function service uploaded by a second network node; The determining of multiple second network nodes matching the AI ​​network function service type includes: According to the preset task type, multiple second network nodes matching the AI ​​network function service type are determined.

5. The method according to claim 3, characterized in that The preset deep reinforcement learning algorithm includes a dual deep Q learning DDQN algorithm, and the selecting at least two second network nodes from the plurality of second network nodes based on the node state information and the preset deep reinforcement learning algorithm includes: With the goal of minimizing system latency and energy consumption and maximizing the number of second network nodes participating in the service, the state space is defined as the node state information of the second network node, the action space is defined as the second network node combination selected for the AI ​​network function service establishment request, and a deep reinforcement learning model is modeled based on the DDQN algorithm; The deep reinforcement learning model is used to determine the reward function of each action in the action set based on the node state information, and the optimal selection decision of the second network node corresponding to the AI ​​network function service is determined based on the reward function, wherein the action is used to represent the selection decision of the second network node corresponding to the AI ​​network function service, and the optimal selection decision includes at least two second network nodes and the task allocation weights of the at least two second network nodes.

6. The method according to claim 5, characterized in that Determining a reward function for each action in an action set according to the node state information using the deep reinforcement learning model includes: Determine an action set based on the node state information using the deep reinforcement learning model, and repeat the reward function determination process until the current action is determined to be the last action in the action set: The reward function determination process includes: taking the next action as the current action according to the action selection order in the action set; receiving the local model parameters uploaded by the second network node included in the current action, and based on the federated average FedAvg algorithm, weightedly aggregating the local model parameters according to the local data size to obtain global model parameters; sending the global model parameters to the second network node included in the current action to continue training the task model of the second network node based on the global model parameters, and obtaining the reward function of the current action after the training is completed.

7. The method according to claim 5, characterized in that Determining the optimal selection decision of the target task with respect to the second network node according to the reward function includes: A target action in the action set that corresponds to the maximum reward function is determined, and a selection strategy represented by the target action is determined as an optimal selection decision for the target task with respect to the second network node.

8. The method according to claim 5, characterized in that The sending the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes includes: Dividing the target task into at least two subtasks according to the task allocation weights of the at least two second network nodes; The subtasks are respectively sent to corresponding second network nodes.

9. The method according to claim 1, characterized in that The method further comprises: Receive and aggregate the task execution results sent by the at least two second network nodes.

10. The method according to claim 9, characterized in that The method further comprises: The aggregated task execution results are sent to the Access and Mobility Management Function (AMF).

11. The method according to claim 1, wherein The method further comprises: Receive feedback on aggregated task execution results.

12. A task scheduling method for artificial intelligence (AI) network function services, characterized in that: The method is applied to a second network node, and includes: Receiving subtasks of a target task sent by a first network node, wherein the subtasks are obtained by dividing the target task by the first network node according to task allocation weights in an optimal selection decision; Execute the subtask according to the task model completed by local training; Sending the task execution result to the first network node.

13. The method according to claim 12, characterized in that The method further comprises: Send node status information and a preset task type for the AI ​​network function service to the first network node, where the node status information includes at least CPU computing frequency, energy consumption information, wireless bandwidth information, and channel status information.

14. The method according to claim 12, characterized in that Before executing the subtask based on the task model completed by local training, the method further includes: sending local model parameters to the first network node; Receiving global model parameters sent by the first network node, wherein the global model parameters are obtained by weightedly aggregating the local model parameters according to local data size based on a federated average FedAvg algorithm after the first network node receives the local model parameters uploaded by the second network node included in the current action; Based on the global model parameters and using a sample set matching the target task to iteratively train the task model, a locally trained task model is obtained.

15. The method according to claim 12, characterized in that The subtask is performed according to the task model completed by local training, including: Retrieving structured data matching the subtask in a user data register UDR, and retrieving unstructured data matching the subtask in an unstructured data storage function UDSF; The structured data and the unstructured data are input into a task model that has been trained locally, and the task execution result of the subtask is output using the task model that has been trained locally.

16. A task scheduling method for artificial intelligence (AI) network function services, characterized in that: The method is applied to user equipment UE, and the method includes: Send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF); Receive the aggregated task execution result transparently transmitted by the access and mobility management function AMF.

17. The method according to claim 16, characterized in that The method further comprises: The structured data set of the target task corresponding to the AI ​​network function service establishment request is stored in the user data register UDR, and the unstructured data set of the target task corresponding to the AI ​​network function service establishment request is stored in the unstructured data storage function UDSF.

18. The method according to claim 16, characterized in that The method further comprises: Send a reception response message of the task execution result and feedback on the aggregated task execution result to the access and mobility management function AMF.

19. A task scheduling method for artificial intelligence (AI) network function services, characterized in that: The method is applied to an access and mobility management function (AMF), and the method includes: Receiving an artificial intelligence (AI) network function service establishment request sent by a user equipment (UE); Send the AI ​​network function service establishment request to the first network node.

20. The method according to claim 19, wherein The method further comprises: receiving the aggregated task execution result sent by the first network node; The aggregated task execution result is transparently transmitted to the user equipment UE.

21. The method according to claim 20, characterized in that The method further comprises: Receive response messages for task execution results and feedback on aggregated task execution results; The feedback on the aggregated task execution results is sent to the first network node.

22. A task scheduling device for artificial intelligence (AI) network function services, characterized in that: The device is applied to a first network node, and includes: a screening module, configured to, in response to the artificial intelligence (AI) network function service establishment request, screen out at least two second network nodes from a plurality of second network nodes, wherein the node level of the plurality of second network nodes is lower than the node level of the first network node; A sending module is used to send the target task corresponding to the AI ​​network function service establishment request to the at least two second network nodes.

23. A task scheduling device for artificial intelligence (AI) network function services, characterized in that: The device is applied to a second network node, and includes: a receiving module, configured to receive subtasks of a target task sent by a first network node, wherein the subtasks are obtained by dividing the target task by the first network node according to task allocation weights in an optimal selection decision; An execution module, configured to execute the subtask according to the task model completed through local training; A sending module is used to send the task execution result to the first network node.

24. A task scheduling device for artificial intelligence (AI) network function services, characterized in that: The device is applied to a user equipment UE, and includes: A sending module is used to send an artificial intelligence (AI) network function service establishment request to the access and mobility management function (AMF); The receiving module is used to receive the aggregated task execution results transparently transmitted by the access and mobility management function AMF.

25. A task scheduling device for artificial intelligence (AI) network function services, characterized in that: The device is applied to the access and mobility management function AMF, and the device includes: A receiving module, configured to receive an artificial intelligence (AI) network function service establishment request sent by a user equipment (UE); A sending module is used to send the AI ​​network function service establishment request to the first network node.

26. A communication device, wherein: include: transceiver; Memory; A processor is connected to the transceiver and the memory respectively, and is configured to control the wireless signal reception and transmission of the transceiver by executing computer-executable instructions on the memory, and can implement the method according to any one of claims 1 to 21.

27. A computer storage medium, wherein: The computer storage medium stores computer-executable instructions; after the computer-executable instructions are executed by the processor, the method according to any one of claims 1 to 21 can be implemented.

28. A communication system comprising at least one of the following network elements: the task scheduling device for an AI network function service applied to a first network node according to claim 22, and the task scheduling device for an AI network function service applied to a second network node according to claim 23.

29. The communication system according to claim 28 further includes an access and mobility management function AMF, and the AMF includes the task scheduling device applied to the access and mobility management function AMF according to claim 25.

Images