Resource allocation method, device, computer equipment, storage medium and program product based on AI traffic perception

Abstract

The present application relates to a resource allocation method, apparatus, computer equipment, storage medium, and program product based on AI traffic perception. The method includes: obtaining AI traffic perception information of a user terminal and a corresponding demand resource table, clustering target tasks of multiple user terminals to obtain multiple target task groups, merging the demand resource tables of each target task group to obtain a target resource table, obtaining demand data and a demand model for each target task group according to the target resource table, and then obtaining the communication resources and computing resources corresponding to the target task group, as well as the corresponding resource allocation scheme, and establishing a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device transmits the corresponding demand data and demand model to the user terminal based on the resource mapping table, and allocates the corresponding computing resources to the user terminal based on the resource mapping table. The adoption of this method can effectively realize real-time resource allocation.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a resource allocation method, apparatus, computer equipment, storage medium, and program product based on AI traffic perception. Background Art

[0002] With the rapid development of information technology, 6G networks will not only significantly increase data transmission speeds but will also integrate cutting-edge technologies such as artificial intelligence, the Internet of Things, and edge computing to form a comprehensive intelligent ecosystem. This transformation will enable 6G networks to provide more efficient and flexible communication services to meet the needs of demanding application scenarios such as autonomous driving, smart cities, and virtual reality.

[0003] In traditional communication networks, traffic is primarily point-to-point, based on specific protocols and static resource allocation. However, in 6G networks, this traditional traffic model will undergo profound changes, gradually transforming into more dynamic and flexible "artificial intelligence (AI) traffic." AI traffic incorporates AI elements into traditional traffic models, including AI models, AI services, AI-related data, and AI resources. Traditional traffic perception and processing methods are no longer adaptable to this new form of AI traffic.

[0004] Based on this, how to effectively perceive the traffic of these AI elements, quickly obtain AI element information, and make decisions such as resource allocation, data transmission, and traffic control based on the traffic perception information has become an urgent problem to be solved. Summary of the Invention

[0005] Based on this, it is necessary to provide a resource allocation method, device, computer equipment, storage medium and program product based on AI traffic perception that can effectively realize real-time resource allocation in response to the above technical problems.

[0006] In a first aspect, the present application provides a resource allocation method based on AI traffic perception, comprising:

[0007] When a user terminal sends AI traffic to a wireless access network via a base station, AI traffic perception information corresponding to the user terminal is obtained, and based on the AI ​​traffic perception information, a required resource table corresponding to the user terminal is obtained; the AI ​​traffic perception information includes data information, model information, task information, resource information, and performance information; the task information includes target tasks;

[0008] Clustering the target tasks corresponding to the multiple user terminals to obtain multiple target task groups, merging the required resource tables corresponding to each target task group to obtain a target resource table corresponding to the target task group;

[0009] For each target task group, according to the target resource table corresponding to the target task group, obtain the demand data and demand model corresponding to the target task group, and according to the demand data and demand model, obtain the communication resources and computing resources corresponding to the target task group;

[0010] Obtain a resource allocation plan corresponding to communication resources and computing resources, and establish a resource mapping table based on the resource allocation plans corresponding to all target task groups, so that the base station device transmits corresponding demand data and demand models to the user terminal based on the resource mapping table, and allocates corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation plan.

[0011] In one embodiment, the demand resource table includes a data mapping table and a model mapping table; and the step of obtaining the demand resource table corresponding to the user terminal based on the AI ​​traffic perception information includes:

[0012] For each target task, obtain the task data and task model of the target task; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task;

[0013] According to all target tasks and corresponding task data, a data mapping table is established; the data mapping table is used to represent the mapping relationship between target tasks and task data;

[0014] A model mapping table is established based on the model IDs of all target tasks and corresponding task models; the model mapping table is used to represent the mapping relationship between the target tasks and the model IDs.

[0015] In one embodiment, the step of clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups includes:

[0016] For each user terminal, feature extraction is performed on the data mapping table and the model mapping table to obtain corresponding data feature information and model feature information, and based on the data feature information and the model feature information, a feature vector corresponding to the target task corresponding to the user terminal is obtained;

[0017] Clustering is performed on the feature vectors corresponding to the target tasks of all user terminals to obtain multiple target task groups.

[0018] In one embodiment, the step of obtaining the demand data and demand model corresponding to the target task group according to the target resource table corresponding to the target task group includes:

[0019] Get the data storage node corresponding to the data mapping table;

[0020] According to the data format and data type in the data mapping table, request the corresponding required data from the data storage node;

[0021] For the model ID in the model mapping table, execute the corresponding lifecycle management operation of the model ID and obtain the required model based on the operation result.

[0022] In one embodiment, the step of obtaining communication resources and computing resources corresponding to the target task group according to the demand data and the demand model includes:

[0023] For each user terminal corresponding to the target task group, obtain the user ID corresponding to the user terminal, and obtain the transmission path corresponding to the demand data and the demand model according to the user ID;

[0024] Obtaining communication resources corresponding to the transmission path;

[0025] Obtain model identification information corresponding to the target task group, and obtain computing resources corresponding to the target task group based on the model identification information.

[0026] In one embodiment, the intelligent control unit includes at least one computing resource pool; the method further includes:

[0027] In the event of sudden changes in the AI ​​traffic sent by the user terminal, the corresponding computing resource pool is allocated to the AI ​​traffic based on the current traffic status of the AI ​​traffic, so as to obtain the AI ​​traffic perception information corresponding to the user terminal through the computing resource pool.

[0028] In a second aspect, the present application also provides a resource allocation device based on AI traffic perception, including:

[0029] The traffic sensing module is used to obtain AI traffic sensing information corresponding to the user terminal when the user terminal sends AI traffic to the wireless access network through the base station device, and obtain the required resource table corresponding to the user terminal based on the AI ​​traffic sensing information; the AI ​​traffic sensing information includes data information, model information, task information, resource information and performance information;

[0030] The task clustering module is used to cluster the target tasks corresponding to multiple user terminals to obtain multiple target task groups, and merge the required resource tables corresponding to each target task group to obtain the target resource table corresponding to the target task group;

[0031] A resource acquisition module is used to obtain, for each target task group, the demand data and demand model corresponding to the target task group according to the target resource table corresponding to the target task group, and obtain the communication resources and computing resources corresponding to the target task group according to the demand data and demand model;

[0032] The resource allocation module is used to obtain resource allocation plans corresponding to communication resources and computing resources, and establish a resource mapping table based on the resource allocation plans corresponding to all target task groups, so that the base station equipment transmits corresponding demand data and demand models to the user terminal based on the resource mapping table, and allocates corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to represent the mapping relationship between the target task group and the resource allocation plan.

[0033] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements any one of the method steps in the first aspect when executing the computer program.

[0034] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements any one of the method steps in the first aspect when the computer program is executed by a processor.

[0035] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which implements any one of the method steps in the first aspect when executed by a processor.

[0036] The above-mentioned resource allocation method, device, computer equipment, storage medium and program product based on AI traffic perception, by deploying an intelligent control unit in the wireless access network to form an intelligent layer of the wireless access network, can efficiently perceive AI traffic, obtain the required resource table of the user terminal in real time, provide customized resource allocation solutions, and achieve rapid response to user needs. By merging and optimizing resources according to the clustering results of the target task group, it realizes the intelligent allocation of model resources, data resources, communication resources and computing resources, thereby effectively realizing real-time resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 This is a diagram of an application environment for an AI traffic-aware resource allocation method in one embodiment;

[0039] Figure 2 1. A flowchart of a resource allocation method for AI traffic awareness in one embodiment;

[0040] Figure 3 This is a flowchart of a resource allocation method for AI traffic awareness in another embodiment;

[0041] Figure 4 This is a structural block diagram of an AI traffic-aware resource allocation device in one embodiment;

[0042] Figure 5 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0044] The AI ​​traffic-aware resource allocation method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the user terminal 102 communicates with the intelligent control unit 106 through the base station device 104. The intelligent control unit 106 is configured to obtain AI traffic perception information corresponding to the user terminal 102 during the process of the user terminal 102 sending AI traffic to the wireless access network through the base station device 104, and obtain a required resource table corresponding to the user terminal 102 based on the AI ​​traffic perception information, cluster target tasks corresponding to multiple user terminals 102 to obtain multiple target task groups, merge the required resource tables corresponding to each target task group to obtain a target resource table corresponding to the target task group, obtain the required data and required model corresponding to each target task group based on the target resource table corresponding to the target task group, obtain the communication resources and computing resources corresponding to the target task group based on the required data and required model, obtain the resource allocation scheme corresponding to the communication resources and computing resources, and establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device 104 transmits the corresponding required data to the user terminal 102 based on the resource mapping table. User terminals 102 may include, but are not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices may include smart speakers, smart TVs, smart air conditioners, smart car devices, and projectors. Portable wearable devices may include smart watches, smart bracelets, and head-mounted devices. Head-mounted devices may include virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. Base station equipment 104 is deployed at the network edge of the wireless access network, and intelligent control unit 106 is deployed at the intelligent layer of the wireless access network.

[0045] In an exemplary embodiment, Figure 2As shown in the figure, a resource allocation method based on AI traffic perception is provided, which is applied to Figure 1 The intelligent control unit 106 in FIG. 1 is taken as an example to illustrate the method, which includes the following steps 202 to 208. Among them:

[0046] S202: When a user terminal sends AI traffic to a wireless access network via a base station device, obtain AI traffic perception information corresponding to the user terminal, and obtain a required resource table corresponding to the user terminal based on the AI ​​traffic perception information; the AI ​​traffic perception information includes data information, model information, task information, resource information, and performance information; and the task information includes a target task.

[0047] Optionally, user terminals are deployed at user nodes, and base stations are deployed at transmission nodes, located at the network edge of the radio access network. These base stations include a centralized unit (CU), a distributed unit (DU), and a radio unit (RU). The CU includes a user plane (CU-UP) and a control plane (CU-CP). Intelligent control units are deployed at network nodes within the radio access network architecture. By introducing intelligent functions into the radio access network, an intelligent layer of the radio access network is achieved.

[0048] Optionally, when a user terminal sends AI traffic to a wireless access network via a base station, the intelligent control unit parses the AI ​​traffic packets to obtain AI traffic perception information corresponding to the user terminal. Compared to traditional communication traffic, AI traffic contains more information in the IP header and data payload, but still adheres to the basic IP packet format of Header+Payload. AI traffic perception information includes data information, model information, task information, resource information, and performance information. In practical applications, because the information contained in AI traffic may be intertwined, key information may need to be extracted through methods such as data analysis and decoding. Furthermore, because each AI traffic packet contains different information, it may be necessary to perceive multiple AI traffic packets within a period of time, perceive each AI traffic packet individually, or set other perception rules. In some embodiments, the intelligent control unit generates an AI traffic perception information list based on the perceived AI traffic perception information of all user terminals. This list is formed by sequentially arranging the AI ​​traffic perception information for each user node. The order of arrangement may be based on the order of user IDs in the user terminals.

[0049] Exemplarily, AI traffic perception information includes multiple types of information as shown in Table 1.

[0050] Table 1 Example of AI traffic perception information

[0051]

[0052] Optionally, the intelligent control unit obtains a demand resource table corresponding to the user terminal based on the AI ​​traffic perception information, wherein the demand resource table is used to represent the resources required to achieve the target task requested by the user terminal, such as communication resources and computing resources.

[0053] S204: clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups, merging the required resource tables corresponding to each target task group to obtain a target resource table corresponding to the target task group.

[0054] Optionally, in actual applications, the wireless access network receives AI traffic from multiple user nodes. By clustering the target tasks corresponding to multiple user terminals, target tasks of the same type are combined to obtain multiple target task groups. For each target task group, the resource demand tables corresponding to each user terminal in the group are merged, and the resource requirements of all tasks in the group are comprehensively considered to obtain the target resource table corresponding to the target task group. Since the required resources for the same type of target tasks are generally similar, clustering can directly allocate resources to different target task groups, improving resource allocation efficiency.

[0055] S206: For each target task group, obtain the demand data and demand model corresponding to the target task group according to the target resource table corresponding to the target task group, and obtain the communication resources and computing resources corresponding to the target task group according to the demand data and demand model.

[0056] Optionally, for each target task group, corresponding demand data and demand model are obtained based on the demand resource data represented by the corresponding target resource table, where the demand data refers to the data resources that need to be allocated to the user terminal, and the demand model refers to the model resources that need to be allocated to the user terminal. Subsequently, communication resources and computing resources corresponding to the target task group are obtained based on the demand data and demand model, where the communication resources include resources for implementing demand data transmission and resources for implementing demand model transmission, and the computing resources include resources for implementing demand data processing and resources for implementing demand model processing.

[0057] S208: Obtain a resource allocation scheme corresponding to communication resources and computing resources, and establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device transmits corresponding demand data and demand models to the user terminal based on the resource mapping table, and allocates corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

[0058] Optionally, based on the communication resources and computing resources corresponding to the target task group, a resource allocation scheme for implementing the allocation of communication resources and computing resources is determined, that is, how to allocate the available communication resources and computing resources to each target task group is determined. Based on the resource allocation schemes corresponding to all target task groups, a resource mapping table is established, wherein the resource mapping table is used to characterize the mapping relationship between each target task group and the corresponding resource allocation scheme. After receiving the instruction from the intelligent control unit, the base station equipment transmits the corresponding demand data and demand model to the user terminal according to the resource mapping table, and allocates the corresponding computing resources to the user terminal based on the resource mapping table, to ensure that the AI ​​tasks of each user terminal can obtain appropriate resource support.

[0059] In the above-mentioned resource allocation method based on AI traffic perception, by deploying an intelligent control unit in the wireless access network to form an intelligent layer of the wireless access network, it can efficiently perceive AI traffic, obtain the required resource table of the user terminal in real time, provide customized resource allocation solutions, and achieve rapid response to user needs. By merging and optimizing resources according to the clustering results of the target task group, it can realize intelligent allocation of model resources, data resources, communication resources and computing resources, thereby effectively realizing real-time resource allocation.

[0060] In an exemplary embodiment, the demand resource table includes a data mapping table and a model mapping table; the step of obtaining the demand resource table corresponding to the user terminal based on the AI ​​traffic perception information includes: for each target task, obtaining the task data and task model of the target task; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task; a data mapping table is established based on all target tasks and corresponding task data; the data mapping table is used to characterize the mapping relationship between the target task and the task data; a model mapping table is established based on the model IDs of all target tasks and corresponding task models; the model mapping table is used to characterize the mapping relationship between the target task and the model ID.

[0061] Optionally, the demand resource table includes a data mapping table and a model mapping table, wherein the data mapping table is used to characterize the mapping relationship between the target task and the demand data, and the model mapping table is used to characterize the mapping relationship between the target task and the model ID. Specifically, for each target task, the required data content (i.e., demand data), model storage file, and configuration file (i.e., demand model) are extracted from the AI ​​traffic perception information, and all target tasks and their corresponding demand data are associated to form a data mapping table, and all target tasks and their corresponding model IDs are associated to form a model mapping table. Through the data mapping table, each target task is accurately associated with the data content it requires, ensuring efficient allocation and use of data. Through the model mapping table, each target task is accurately associated with the model it requires, ensuring rapid calling and deployment of the model.

[0062] For example, for the target task's requirement data, the data flow, data format, data storage location, data type, data usage specifications, and other information of the requirement data are obtained, and the resulting data mapping table can be expressed as: [Target task 1: Data requirement 1; ...; Target task k: Data requirement k]. For the target task's requirement model, the mapping relationship between the model lifecycle management operations that the target task needs to perform and the specific model ID is obtained, and the resulting model mapping table can be expressed as: [AI-related target task 1: lifecycle management operation p, model ID m ; ...; AI-related target task k: lifecycle management operation q, model ID n ].

[0063] In this embodiment, by accurately associating the target task with the required data, it is ensured that each task can quickly obtain the required data content, thereby avoiding the waste and repeated allocation of data resources. By accurately associating the target task with the model ID, it is ensured that each task can quickly call the required model, thereby improving the utilization efficiency of model resources. Furthermore, through the data mapping table and the model mapping table, intelligent matching of tasks and resources is achieved, thereby improving resource allocation efficiency.

[0064] In an exemplary embodiment, the steps of clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups include: for each user terminal, performing feature extraction on the data mapping table and the model mapping table respectively to obtain corresponding data feature information and model feature information, and obtaining the feature vector corresponding to the target task corresponding to the user terminal based on the data feature information and the model feature information; clustering the feature vectors corresponding to the target tasks corresponding to all user terminals to obtain multiple target task groups.

[0065] Optionally, based on the data and model mapping relationship table corresponding to each target task, the data and model feature information therein is extracted to generate a feature vector corresponding to the target task, wherein the dimension of the feature vector needs to be consistent in all target tasks, and the feature vector includes multiple features of factors such as data and model. By summarizing the feature vectors corresponding to all target tasks, a clustering algorithm is executed, such as the K-Means algorithm, the DBSCAN clustering algorithm, the hierarchical clustering algorithm, etc., and according to the output results of the clustering algorithm, all target tasks divided into the same category are formed into a target task group, and target tasks divided into different categories are formed into different target task groups. For each target task group, the information in the data mapping table and the model mapping table of each target task is summarized to form a target resource table corresponding to the target task group.

[0066] In this embodiment, similar tasks of multiple user terminals are divided into the same group through clustering processing, and unified resource allocation and scheduling can be performed for the same group of tasks, thereby achieving efficient sharing of resources. In addition, through task grouping, multiple target task groups can be processed in parallel, thereby improving the efficiency of task processing.

[0067] In an exemplary embodiment, the steps of obtaining the demand data and demand model corresponding to the target task group according to the target resource table corresponding to the target task group include: obtaining the data storage node corresponding to the data mapping table; requesting the corresponding demand data from the data storage node according to the data format and data type in the data mapping table; executing the lifecycle management operation corresponding to the model ID in the model mapping table, and obtaining the demand model according to the operation result.

[0068] Optionally, for the data mapping table, request the corresponding data resources from the data storage node corresponding to the data mapping table; for the model mapping table, execute the lifecycle management operation corresponding to the model ID, and obtain the required model based on the operation results, such as model training results, model inference results, or model update data.

[0069] In this embodiment, the delay in data acquisition is reduced through precise data storage node positioning and data request. Through model lifecycle management operations, high-quality demand models are generated to ensure the efficient execution of AI tasks, thereby realizing intelligent management of data resources.

[0070] In an exemplary embodiment, the steps of obtaining communication resources and computing resources corresponding to the target task group based on demand data and demand model include: obtaining the user ID corresponding to each user terminal corresponding to the target task group, and obtaining the transmission paths corresponding to the demand data and demand model according to the user ID; obtaining the communication resources corresponding to the transmission path; obtaining the model identification information corresponding to the target task group, and obtaining the computing resources corresponding to the target task group according to the model identification information.

[0071] Optionally, the user ID corresponding to each user terminal in the target task group is obtained separately, wherein the user ID can be determined based on the target task ID to ensure that each user terminal uniquely corresponds to the target task, wherein the user ID is used to identify the user terminal and provide a unique identifier for data transmission and model transmission. The transmission paths corresponding to the demand data and the demand model are obtained separately based on the user ID. Exemplarily, for the demand data, historical data can be obtained from the operation and maintenance management (OAM) or network data analysis function (NWDAF) through user plane signaling, or real-time data can be collected from the CU and DU to determine the transmission path of the demand data. For the demand model, it can be transmitted through user plane or control plane signaling, for example, through Media Access Control Control Element (MAC CE), Radio Resource Control (RRC) or Downlink Control Information (DCI), and model compression, encoding and other methods can be used to reduce transmission overhead.

[0072] Furthermore, based on the acquired demand data and demand model, the communication resources and computing resources corresponding to the target task group are obtained. In practical applications, the communication resource allocation scheme can be serial or parallel. The serial method means that all communication resources are allocated to a certain target task group at the same time for the transmission of data resources and model resources, and to another target task group at other times; the parallel scheme means that communication resources are allocated to different target task groups without duplication at the same time for the transmission of data resources and model resources.

[0073] Optionally, computing resources are used to implement processes such as model training and inference. The computing resource allocation scheme can adopt the principle of proportional fairness, allocating computing resources proportionally to all target task groups based on the computational complexity and other information in the model identification information. Furthermore, a resource mapping table is established based on the resource allocation schemes corresponding to all target task groups. The resource mapping table is used to represent the mapping relationship between target task groups and resource allocation schemes, so that the base station device allocates corresponding resources, such as data resources, communication resources, computing resources, etc., to each target task group based on the information in the resource mapping table.

[0074] In this embodiment, by respectively obtaining the transmission paths of the demand data and the demand model according to the user ID, and obtaining the corresponding demand model according to the transmission path, the transmission efficiency can be improved, resource utilization can be optimized, and real-time resource allocation can be effectively achieved.

[0075] In an exemplary embodiment, the intelligent control unit includes at least one computing power resource pool; the method further includes: when there is a sudden change in the AI ​​traffic sent by the user terminal, allocating a corresponding computing power resource pool to the AI ​​traffic based on the current traffic state of the AI ​​traffic, so as to obtain AI traffic perception information corresponding to the user terminal through the computing power resource pool.

[0076] Optionally, a computing resource pool is a collection of one or more computing resources within the intelligent control unit, used to support the execution of AI tasks. This computing resource pool can provide dynamic computing resource allocation to support tasks such as AI traffic perception, model training, and inference. The intelligent control unit monitors the state of AI traffic sent by user terminals in real time, identifies sudden changes in traffic (such as traffic surges or drops), and dynamically allocates the corresponding computing resource pool based on the current state of AI traffic, ensuring the efficient execution of AI tasks.

[0077] In this embodiment, by dynamically allocating computing resource pools based on sudden changes in AI traffic, AI traffic can be quickly processed and AI traffic perception information corresponding to the user terminal can be generated, thereby effectively realizing real-time resource allocation based on the AI ​​traffic perception information.

[0078] In an exemplary embodiment, Figure 3 As shown, a resource allocation method based on AI traffic perception is provided, which includes the following steps:

[0079] (1) Obtain AI traffic perception information corresponding to the user terminal through AI traffic perception: When the user terminal sends AI traffic to the wireless access network through the base station equipment, obtain the AI ​​traffic perception information corresponding to the user terminal. Among them, AI traffic perception information includes data information, model information, task information, resource information and performance information; task information includes target task.

[0080] (2) Generate a demand resource table based on AI traffic perception information: The demand resource table includes a data mapping table and a model mapping table. For each target task, the task data and task model of the target task are obtained; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task; a data mapping table is established based on all target tasks and corresponding task data; the data mapping table is used to represent the mapping relationship between the target task and the task data; a model mapping table is established based on the model IDs of all target tasks and corresponding task models; the model mapping table is used to represent the mapping relationship between the target task and the model ID.

[0081] (3) Clustering the target tasks to obtain multiple target task groups: For each user terminal, feature extraction is performed on the data mapping table and the model mapping table to obtain the corresponding data feature information and model feature information, and based on the data feature information and the model feature information, the feature vector corresponding to the target task corresponding to the user terminal is obtained; clustering is performed on the feature vectors corresponding to the target tasks corresponding to all user terminals to obtain multiple target task groups; the demand resource tables corresponding to each target task group are merged to obtain the target resource table corresponding to the target task group.

[0082] (4) Obtain the demand data and demand model corresponding to the target task group according to the target resource table: For each target task group, obtain the data storage node corresponding to the data mapping table; according to the data format and data type in the data mapping table, request the corresponding demand data from the data storage node; for the model ID in the model mapping table, perform the lifecycle management operation corresponding to the model ID, and obtain the demand model according to the operation result.

[0083] (5) Obtain the communication resources and computing resources corresponding to the target task group based on the transmission path of the resources: For each user terminal corresponding to the target task group, obtain the user ID corresponding to the user terminal, and obtain the transmission path corresponding to the demand data and the demand model respectively according to the user ID; obtain the communication resources corresponding to the transmission path; obtain the model identification information corresponding to the target task group, and obtain the computing resources corresponding to the target task group based on the model identification information.

[0084] (6) Establish a resource mapping table based on the resource allocation plan to achieve resource allocation for the user terminal: obtain the resource allocation plan corresponding to the communication resources and computing resources, and establish a resource mapping table based on the resource allocation plan corresponding to all target task groups, so that the base station equipment transmits the corresponding demand data and demand model to the user terminal based on the resource mapping table, and allocates the corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation plan. In the case of sudden changes in the AI ​​traffic sent by the user terminal, the corresponding computing resource pool is allocated to the AI ​​traffic according to the current traffic state of the AI ​​traffic, so as to obtain the AI ​​traffic perception information corresponding to the user terminal through the computing resource pool.

[0085] In this embodiment, by deploying an intelligent control unit in the wireless access network to form an intelligent layer of the wireless access network, it is possible to efficiently perceive AI traffic, obtain the required resource table of the user terminal in real time, provide customized resource allocation solutions, and achieve rapid response to user needs. By merging and optimizing resource allocation based on the clustering results of the target task group, intelligent allocation of model resources, data resources, communication resources and computing resources is achieved, thereby effectively realizing real-time resource allocation.

[0086] It should be understood that, although the steps in the flowcharts of the above embodiments are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the flowcharts of the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0087] Based on the same inventive concept, an embodiment of the present application also provides an AI traffic-aware resource allocation device for implementing the above-mentioned AI traffic-aware resource allocation method. The implementation solution provided by the device is similar to the implementation solution described in the above-mentioned method. Therefore, the specific limitations of one or more AI traffic-aware resource allocation device embodiments provided below can be found in the above-mentioned limitations on the AI ​​traffic-aware resource allocation method, and will not be repeated here.

[0088] In an exemplary embodiment, Figure 4 As shown, a resource allocation device based on AI traffic perception is provided, including: a traffic perception module 10, a task clustering module 20, a resource acquisition module 30 and a resource allocation module 40, wherein:

[0089] The traffic sensing module 10 is used to obtain AI traffic sensing information corresponding to the user terminal during the process of the user terminal sending AI traffic to the wireless access network through the base station device, and obtain the required resource table corresponding to the user terminal based on the AI ​​traffic sensing information; the AI ​​traffic sensing information includes data information, model information, task information, resource information and performance information; the task information includes the target task.

[0090] The task clustering module 20 is configured to cluster target tasks corresponding to multiple user terminals to obtain multiple target task groups, and merge the required resource tables corresponding to each target task group to obtain a target resource table corresponding to the target task group.

[0091] The resource acquisition module 30 is used to obtain the demand data and demand model corresponding to each target task group according to the target resource table corresponding to the target task group, and obtain the communication resources and computing resources corresponding to the target task group according to the demand data and demand model.

[0092] The resource allocation module 40 is used to obtain resource allocation schemes corresponding to communication resources and computing resources, and to establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device transmits corresponding demand data and demand models to the user terminal based on the resource mapping table, and allocates corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

[0093] In an exemplary embodiment, the demand resource table includes a data mapping table and a model mapping table; the traffic perception module 10 is also used to obtain the task data and task model of the target task for each target task; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task; a data mapping table is established based on all target tasks and corresponding task data; the data mapping table is used to characterize the mapping relationship between the target task and the task data; a model mapping table is established based on the model ID of all target tasks and corresponding task models; the model mapping table is used to characterize the mapping relationship between the target task and the model ID.

[0094] In an exemplary embodiment, the task clustering module 20 is also used to perform feature extraction on the data mapping table and the model mapping table for each user terminal, respectively, to obtain corresponding data feature information and model feature information, and to obtain the feature vector corresponding to the target task corresponding to the user terminal based on the data feature information and the model feature information; cluster the feature vectors corresponding to the target tasks corresponding to all user terminals to obtain multiple target task groups.

[0095] In an exemplary embodiment, the resource acquisition module 30 is used to obtain the data storage node corresponding to the data mapping table; request the corresponding demand data from the data storage node according to the data format and data type in the data mapping table; execute the lifecycle management operation corresponding to the model ID in the model mapping table, and obtain the demand model according to the operation result.

[0096] In an exemplary embodiment, the resource acquisition module 30 is also used to obtain the user ID corresponding to each user terminal corresponding to the target task group, and obtain the transmission path corresponding to the demand data and the demand model according to the user ID; obtain the communication resources corresponding to the transmission path; obtain the model identification information corresponding to the target task group, and obtain the computing resources corresponding to the target task group according to the model identification information.

[0097] In an exemplary embodiment, the intelligent control unit includes at least one computing power resource pool; the resource allocation module 40 is also used to allocate a corresponding computing power resource pool to the AI ​​traffic according to the current traffic status of the AI ​​traffic when there is a sudden change in the AI ​​traffic sent by the user terminal, so as to obtain the AI ​​traffic perception information corresponding to the user terminal through the computing power resource pool.

[0098] Each module in the above-mentioned AI traffic-aware resource allocation device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0099] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means, and the wireless means can be implemented via Wi-Fi, mobile cellular networks, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a resource allocation method based on AI traffic perception. The display unit of the computer device is used to form a visually visible image, and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0100] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0101] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented: in the process of a user terminal sending AI traffic to a wireless access network through a base station device, AI traffic perception information corresponding to the user terminal is obtained, and based on the AI ​​traffic perception information, a demand resource table corresponding to the user terminal is obtained; the AI ​​traffic perception information includes data information, model information, task information, resource information and performance information; the task information includes a target task; clustering the target tasks corresponding to multiple user terminals to obtain multiple target task groups, and respectively performing clustering on the demand resource table corresponding to each target task group. Merge to obtain a target resource table corresponding to the target task group; for each target task group, obtain the demand data and demand model corresponding to the target task group according to the target resource table corresponding to the target task group, and obtain the communication resources and computing resources corresponding to the target task group according to the demand data and demand model; obtain the resource allocation scheme corresponding to the communication resources and computing resources, and establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station equipment transmits the corresponding demand data and demand model to the user terminal based on the resource mapping table, and allocates the corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

[0102] In one embodiment, the demand resource table includes a data mapping table and a model mapping table; when the processor executes the computer program, it obtains the demand resource table corresponding to the user terminal based on the AI ​​traffic perception information, including: for each target task, obtaining the task data and task model of the target task; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task; according to all target tasks and corresponding task data, a data mapping table is established; the data mapping table is used to characterize the mapping relationship between the target task and the task data; according to the model ID of all target tasks and the corresponding task model, a model mapping table is established; the model mapping table is used to characterize the mapping relationship between the target task and the model ID.

[0103] In one embodiment, when a processor executes a computer program, it involves clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups, including: for each user terminal, respectively extracting features from a data mapping table and a model mapping table to obtain corresponding data feature information and model feature information, and obtaining a feature vector corresponding to the target task corresponding to the user terminal based on the data feature information and the model feature information; clustering the feature vectors corresponding to the target tasks corresponding to all user terminals to obtain multiple target task groups.

[0104] In one embodiment, when a processor executes a computer program, it obtains the demand data and demand model corresponding to the target task group according to the target resource table corresponding to the target task group, including: obtaining the data storage node corresponding to the data mapping table; requesting the corresponding demand data from the data storage node according to the data format and data type in the data mapping table; executing the lifecycle management operation corresponding to the model ID in the model mapping table, and obtaining the demand model according to the operation result.

[0105] In one embodiment, when a processor executes a computer program, the processor obtains communication resources and computing resources corresponding to the target task group based on demand data and a demand model, including: obtaining a user ID corresponding to each user terminal corresponding to the target task group, and obtaining transmission paths corresponding to the demand data and the demand model according to the user ID; obtaining communication resources corresponding to the transmission path; obtaining model identification information corresponding to the target task group, and obtaining computing resources corresponding to the target task group according to the model identification information.

[0106] In one embodiment, the intelligent control unit includes at least one computing power resource pool; when the processor executes the computer program, it further implements the following steps: when there is a sudden change in the AI ​​traffic sent by the user terminal, the corresponding computing power resource pool is allocated to the AI ​​traffic based on the current traffic state of the AI ​​traffic, so as to obtain AI traffic perception information corresponding to the user terminal through the computing power resource pool.

[0107] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: in the process of a user terminal sending AI traffic to a wireless access network through a base station device, AI traffic perception information corresponding to the user terminal is obtained, and based on the AI ​​traffic perception information, a demand resource table corresponding to the user terminal is obtained; the AI ​​traffic perception information includes data information, model information, task information, resource information and performance information; the task information includes a target task; clustering the target tasks corresponding to multiple user terminals to obtain multiple target task groups, and merging the demand resource tables corresponding to each target task group to obtain to the target resource table corresponding to the target task group; for each target task group, according to the target resource table corresponding to the target task group, obtain the demand data and demand model corresponding to the target task group, and according to the demand data and demand model, obtain the communication resources and computing resources corresponding to the target task group; obtain the resource allocation scheme corresponding to the communication resources and computing resources, and establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device transmits the corresponding demand data and demand model to the user terminal based on the resource mapping table, and allocates the corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

[0108] In one embodiment, the demand resource table includes a data mapping table and a model mapping table; when the computer program is executed by the processor, it involves obtaining the demand resource table corresponding to the user terminal based on the AI ​​traffic perception information, including: for each target task, obtaining the task data and task model of the target task; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task; according to all target tasks and corresponding task data, a data mapping table is established; the data mapping table is used to characterize the mapping relationship between the target task and the task data; according to the model ID of all target tasks and the corresponding task model, a model mapping table is established; the model mapping table is used to characterize the mapping relationship between the target task and the model ID.

[0109] In one embodiment, when a computer program is executed by a processor, it involves clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups, including: for each user terminal, feature extraction is performed on the data mapping table and the model mapping table respectively to obtain corresponding data feature information and model feature information, and based on the data feature information and the model feature information, a feature vector corresponding to the target task corresponding to the user terminal is obtained; and feature vectors corresponding to the target tasks corresponding to all user terminals are clustered to obtain multiple target task groups.

[0110] In one embodiment, when a computer program is executed by a processor, it involves obtaining demand data and demand models corresponding to the target task group based on the target resource table corresponding to the target task group, including: obtaining a data storage node corresponding to a data mapping table; requesting corresponding demand data from the data storage node based on the data format and data type in the data mapping table; executing a lifecycle management operation corresponding to the model ID in the model mapping table, and obtaining the demand model based on the operation result.

[0111] In one embodiment, when a computer program is executed by a processor, the computer program involves obtaining communication resources and computing resources corresponding to the target task group based on demand data and a demand model, including: obtaining a user ID corresponding to each user terminal corresponding to the target task group, and obtaining transmission paths corresponding to the demand data and the demand model based on the user ID; obtaining communication resources corresponding to the transmission path; obtaining model identification information corresponding to the target task group, and obtaining computing resources corresponding to the target task group based on the model identification information.

[0112] In one embodiment, the intelligent control unit includes at least one computing power resource pool; when the computer program is executed by the processor, the computer program further implements the following steps: when there is a sudden change in the AI ​​traffic sent by the user terminal, the corresponding computing power resource pool is allocated to the AI ​​traffic based on the current traffic state of the AI ​​traffic, so as to obtain AI traffic perception information corresponding to the user terminal through the computing power resource pool.

[0113] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps: in a process in which a user terminal sends AI traffic to a wireless access network via a base station device, obtaining AI traffic perception information corresponding to the user terminal, and obtaining a demand resource table corresponding to the user terminal based on the AI ​​traffic perception information; the AI ​​traffic perception information includes data information, model information, task information, resource information, and performance information; the task information includes a target task; clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups, merging the demand resource tables corresponding to each target task group, and obtaining a target task group; a target resource table corresponding to the target task group; for each target task group, according to the target resource table corresponding to the target task group, obtain the demand data and demand model corresponding to the target task group, and according to the demand data and demand model, obtain the communication resources and computing resources corresponding to the target task group; obtain the resource allocation scheme corresponding to the communication resources and computing resources, and establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station equipment transmits the corresponding demand data and demand model to the user terminal based on the resource mapping table, and allocates the corresponding computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

[0114] In one embodiment, the demand resource table includes a data mapping table and a model mapping table; when the computer program is executed by the processor, it involves obtaining the demand resource table corresponding to the user terminal based on the AI ​​traffic perception information, including: for each target task, obtaining the task data and task model of the target task; the task data includes the data content of the target task; the task model includes the model storage file and configuration file of the target task; according to all target tasks and corresponding task data, a data mapping table is established; the data mapping table is used to characterize the mapping relationship between the target task and the task data; according to the model ID of all target tasks and the corresponding task model, a model mapping table is established; the model mapping table is used to characterize the mapping relationship between the target task and the model ID.

[0115] In one embodiment, when a computer program is executed by a processor, it involves clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups, including: for each user terminal, feature extraction is performed on the data mapping table and the model mapping table respectively to obtain corresponding data feature information and model feature information, and based on the data feature information and the model feature information, a feature vector corresponding to the target task corresponding to the user terminal is obtained; and feature vectors corresponding to the target tasks corresponding to all user terminals are clustered to obtain multiple target task groups.

[0116] In one embodiment, when a computer program is executed by a processor, it involves obtaining demand data and demand models corresponding to the target task group based on the target resource table corresponding to the target task group, including: obtaining a data storage node corresponding to a data mapping table; requesting corresponding demand data from the data storage node based on the data format and data type in the data mapping table; executing a lifecycle management operation corresponding to the model ID in the model mapping table, and obtaining the demand model based on the operation result.

[0117] In one embodiment, when a computer program is executed by a processor, the computer program involves obtaining communication resources and computing resources corresponding to the target task group based on demand data and a demand model, including: obtaining a user ID corresponding to each user terminal corresponding to the target task group, and obtaining transmission paths corresponding to the demand data and the demand model based on the user ID; obtaining communication resources corresponding to the transmission path; obtaining model identification information corresponding to the target task group, and obtaining computing resources corresponding to the target task group based on the model identification information.

[0118] In one embodiment, the intelligent control unit includes at least one computing power resource pool; when the computer program is executed by the processor, the computer program further implements the following steps: when there is a sudden change in the AI ​​traffic sent by the user terminal, the corresponding computing power resource pool is allocated to the AI ​​traffic based on the current traffic state of the AI ​​traffic, so as to obtain AI traffic perception information corresponding to the user terminal through the computing power resource pool.

[0119] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0120] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0121] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A resource allocation method based on AI traffic perception, characterized in that: Applied to intelligent control units; The intelligent control unit is deployed in a wireless access network architecture; the method includes: During the process of a user terminal sending AI traffic to a wireless access network via a base station device, obtaining AI traffic perception information corresponding to the user terminal, and obtaining a required resource table corresponding to the user terminal based on the AI ​​traffic perception information; the AI ​​traffic perception information includes data information, model information, task information, resource information, and performance information; and the task information includes a target task; Clustering target tasks corresponding to multiple user terminals to obtain multiple target task groups, and merging the required resource tables corresponding to each target task group to obtain a target resource table corresponding to the target task group; For each target task group, obtaining demand data and a demand model corresponding to the target task group according to the target resource table corresponding to the target task group, and obtaining communication resources and computing resources corresponding to the target task group according to the demand data and the demand model; Obtain a resource allocation scheme corresponding to the communication resources and the computing resources, and establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device transmits corresponding demand data and demand model to the user terminal based on the resource mapping table, and allocates corresponding communication resources and computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

2. The method according to claim 1, characterized in that The demand resource table includes a data mapping table and a model mapping table; obtaining the demand resource table corresponding to the user terminal according to the AI ​​traffic perception information includes: For each target task, obtain the task data and task model of the target task; the task data includes the data content required by the target task; the task model includes the model storage file and configuration file required by the target task; Establish a data mapping table based on all target tasks and corresponding task data; the data mapping table is used to represent the mapping relationship between the target tasks and the task data; A model mapping table is established based on all target tasks and the model IDs of the corresponding task models; the model mapping table is used to represent the mapping relationship between the target tasks and the model IDs.

3. The method according to claim 2, characterized in that The target tasks corresponding to the multiple user terminals are clustered to obtain multiple target task groups, including: For each user terminal, feature extraction is performed on the data mapping table and the model mapping table to obtain corresponding data feature information and model feature information, and a feature vector corresponding to the target task corresponding to the user terminal is obtained based on the data feature information and the model feature information; Clustering is performed on the feature vectors corresponding to the target tasks of all user terminals to obtain multiple target task groups.

4. The method according to claim 2, characterized in that The acquiring, according to the target resource table corresponding to the target task group, demand data and demand model corresponding to the target task group includes: Acquire the data storage node corresponding to the data mapping table; Requesting corresponding required data from the data storage node according to the data format and data type in the data mapping table; For the model ID in the model mapping table, perform the lifecycle management operation corresponding to the model ID, and obtain the demand model according to the operation result.

5. The method according to claim 1, wherein The acquiring, according to the demand data and the demand model, communication resources and computing resources corresponding to the target task group includes: For each user terminal corresponding to the target task group, obtaining a user ID corresponding to the user terminal, and obtaining the demand data and the transmission path corresponding to the demand model according to the user ID; Acquiring communication resources corresponding to the transmission path; Obtain model identification information corresponding to the target task group, and obtain computing resources corresponding to the target task group according to the model identification information.

6. The method according to claim 1, characterized in that The intelligent control unit includes at least one computing resource pool; the method further includes: In the event that there is a sudden change in the AI ​​traffic sent by the user terminal, a corresponding computing resource pool is allocated to the AI ​​traffic based on the current traffic state of the AI ​​traffic, so as to obtain AI traffic perception information corresponding to the user terminal through the computing resource pool.

7. A resource allocation device based on AI traffic perception, characterized in that: The device comprises: A traffic sensing module is configured to obtain AI traffic sensing information corresponding to a user terminal during the process of the user terminal sending AI traffic to a wireless access network via a base station device, and obtain a required resource table corresponding to the user terminal based on the AI ​​traffic sensing information; the AI ​​traffic sensing information includes data information, model information, task information, resource information, and performance information; and the task information includes a target task; A task clustering module is used to cluster target tasks corresponding to multiple user terminals to obtain multiple target task groups, and merge the required resource tables corresponding to each target task group to obtain a target resource table corresponding to the target task group; a resource acquisition module, configured to acquire, for each target task group, demand data and a demand model corresponding to the target task group according to a target resource table corresponding to the target task group, and acquire, based on the demand data and the demand model, communication resources and computing resources corresponding to the target task group; A resource allocation module is used to obtain a resource allocation scheme corresponding to the communication resources and the computing resources, and to establish a resource mapping table based on the resource allocation schemes corresponding to all target task groups, so that the base station device transmits corresponding demand data and demand models to the user terminal based on the resource mapping table, and allocates corresponding communication resources and computing resources to the user terminal based on the resource mapping table; the resource mapping table is used to characterize the mapping relationship between the target task group and the resource allocation scheme.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images