CN120166409B - Methods, apparatus, devices and storage media for orchestrating AI inference tasks in wireless access networks - Google Patents
Methods, apparatus, devices and storage media for orchestrating AI inference tasks in wireless access networksInfo
- Publication number
- CN120166409B CN120166409B CN202311733048.9A CN202311733048A CN120166409B CN 120166409 B CN120166409 B CN 120166409B CN 202311733048 A CN202311733048 A CN 202311733048A CN 120166409 B CN120166409 B CN 120166409B
- Authority
- CN
- China
- Prior art keywords
- reasoning
- information
- access network
- task
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/18—Network planning tools
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/22—Traffic simulation tools or models
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
本发明公开了一种无线接入网络的AI推理任务编排方法、装置、设备和存储介质，在RAN侧引入人工智能技术，首先接收非实时无线控制器发送的AI推理模型的AI推理任务保障策略，以及获取至少一个终端的终端信息，其中，AI推理任务保障策略中定义了无线接入网络的优化目标和AI推理模型的特征信息，终端信息包括计算资源信息和通信资源信息；然后根据所述AI推理任务保障策略和所述终端信息生成无线接入网络的AI推理任务编排方案，能根据实时的通信状态和计算能力，对RAN侧的计算节点做近实时的A I推理任务协作编排，提高了RAN侧资源分配和协作计算的效率，进而提升了系统性能。
This invention discloses an AI inference task orchestration method, apparatus, device, and storage medium for a wireless access network (RAN). By introducing artificial intelligence technology on the RAN side, the method first receives an AI inference task guarantee strategy for the AI inference model sent by a non-real-time wireless controller, and acquires terminal information of at least one terminal. The AI inference task guarantee strategy defines the optimization objectives of the RAN and the characteristic information of the AI inference model. The terminal information includes computing resource information and communication resource information. Then, based on the AI inference task guarantee strategy and the terminal information, an AI inference task orchestration scheme for the RAN is generated. This scheme enables near-real-time collaborative orchestration of AI inference tasks for computing nodes on the RAN side, based on real-time communication status and computing capabilities. This improves the efficiency of RAN-side resource allocation and collaborative computing, thereby enhancing system performance.
Description
Technical Field
The present invention relates to the field of wireless communications technologies, and in particular, to a method, an apparatus, a device, and a storage medium for scheduling AI reasoning tasks in a wireless access network.
Background
With the continuous and deep research on computing power in wireless local area networks, base stations are increasingly receiving attention from the industry as an edge computing platform, and wireless local area networks are developing towards communication and computing integration. While existing studies provide solutions for balancing accuracy and delay in device edge computing collaboration, these studies have focused mainly on edge servers and cannot be applied directly to computing collaboration between base stations and terminals where computing resources are shared by communication processing and computing tasks, which are in competing relationship. Therefore, in the case of limited computing resources and bandwidth resources for communication, the RAN (Radio Access Network ) side needs to perform near real-time adaptive scheduling on multi-base-station multi-terminal tasks according to dynamically changing channel states and different information of different tasks.
Near-Real-time Controller (Near-Real-TIME RAN INTELLIGENT Controller) in conventional O-RAN (Open-Radio Access Network, open wireless access network) can collect Near-Real-time communication related parameters through standardized E2 interface to perform intelligent optimization on system performance, however, this is merely collection optimization on communication related parameters of RAN side, and lack of uniform perceived scheduling capability on computing resources and communication resources results in low efficiency of resource allocation and cooperation computation of RAN side, which results in reduced system performance.
Disclosure of Invention
The embodiment of the invention aims to provide an AI reasoning task arrangement method of a wireless access network, which introduces an artificial intelligence (ARTIFICIAL INTELLIGENCE) technology at the RAN side, can carry out near-real-time AI reasoning task cooperation arrangement on a computing node at the RAN side according to a real-time communication state and computing capacity, improves the efficiency of resource allocation and cooperation computation at the RAN side, and further improves the system performance.
In order to achieve the above object, an embodiment of the present invention provides an AI reasoning task orchestration method for a wireless access network, applied to a near real-time wireless controller, where the method includes:
Receiving an AI (advanced technology attachment) reasoning task guarantee strategy of an AI reasoning model sent by a non-real-time wireless controller, wherein the AI reasoning model is deployed in a wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
Acquiring terminal information of at least one terminal, wherein the terminal information comprises computing resource information and communication resource information;
And generating an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
As an improvement of the above solution, the generating an AI reasoning task orchestration solution of a radio access network according to the AI reasoning task guarantee policy and the terminal information includes:
Determining a node to be allocated and a corresponding computing node according to the optimization target of the wireless access network, and selecting a corresponding dividing point and an exiting point from the characteristic information of the AI reasoning model according to the information of each terminal, wherein the dividing point and the exiting point are one layer of the AI reasoning model;
And forming the AI reasoning task arrangement scheme by the nodes to be distributed and the corresponding computing nodes, the dividing points and the exit points.
As an improvement of the above solution, after generating an AI reasoning task orchestration scheme of the radio access network according to the AI reasoning task guarantee policy and the terminal information, the method further includes:
and sending the AI reasoning task arrangement scheme to the nodes to be distributed and the corresponding computing nodes so that the nodes to be distributed and the corresponding computing nodes finish respective reasoning computing tasks.
As an improvement of the above solution, when the terminal is a user equipment, the acquiring terminal information of at least one terminal includes:
a request message is sent to a base station where the user equipment resides, so that the base station reports the terminal information of the user equipment according to the request message, or,
And subscribing the terminal information to a base station where the user equipment resides so that the base station reports the terminal information of the user equipment.
When the terminal is user equipment, the user equipment supports information interaction with the near real-time wireless controller, and the acquiring of the terminal information of at least one terminal comprises the following steps:
a request message is sent to the user equipment to enable the user equipment to report terminal information according to the request message, or,
And subscribing the terminal information to the user equipment so as to enable the user equipment to report the terminal information.
In order to achieve the above object, the embodiment of the present invention further provides an AI reasoning task orchestration method of a wireless access network, which is applied to a non-real-time wireless controller, and the method includes:
Obtaining model information of an AI reasoning model;
An AI reasoning task guarantee strategy is generated according to the model information, wherein the AI reasoning model is deployed in the wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
and sending the AI reasoning task guarantee strategy to a near-real-time wireless controller so that the near-real-time wireless controller generates an AI reasoning task arrangement scheme of a wireless access network according to the AI reasoning task guarantee strategy and terminal information, wherein the terminal information comprises computing resource information and communication resource information.
As an improvement of the scheme, the model information comprises characteristic information of the AI reasoning model and performance guarantee parameters of the AI reasoning task, wherein the performance guarantee parameters comprise, but are not limited to, model reasoning accuracy and reasoning calculation times of unit time.
In order to achieve the above object, an embodiment of the present invention further provides an AI reasoning task orchestration device of a wireless access network, including:
The system comprises an AI (advanced technology attachment) reasoning task guarantee strategy receiving module, an AI (advanced technology attachment) reasoning task guarantee strategy processing module and a non-real-time wireless controller, wherein the AI reasoning task guarantee strategy receiving module is used for receiving an AI reasoning task guarantee strategy of an AI reasoning model sent by the non-real-time wireless controller, the AI reasoning model is deployed in a wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
The terminal information acquisition module is used for acquiring terminal information of at least one terminal, wherein the terminal information comprises computing resource information and communication resource information;
And the AI reasoning task arrangement scheme generation module is used for generating an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
In order to achieve the above object, an embodiment of the present invention further provides an AI reasoning task orchestration device of a wireless access network, including:
the model information acquisition module is used for acquiring the model information of the AI reasoning model;
the system comprises an AI reasoning task guarantee strategy generation module, an AI reasoning task guarantee strategy generation module and a model information generation module, wherein the AI reasoning task guarantee strategy generation module is used for generating an AI reasoning task guarantee strategy according to the model information, the AI reasoning model is deployed in a wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
And the AI reasoning task guarantee strategy sending module is used for sending the AI reasoning task guarantee strategy to a near-real-time wireless controller so that the near-real-time wireless controller can generate an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and terminal information, wherein the terminal information comprises computing resource information and communication resource information.
To achieve the above object, an embodiment of the present invention further provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, where the processor executes the computer program to implement the AI inference task orchestration method of the radio access network according to any one of the embodiments.
To achieve the above object, an embodiment of the present invention further provides a computer readable storage medium, where the computer readable storage medium includes a stored computer program, and when the computer program runs, controls a device where the computer readable storage medium is located to execute the AI reasoning task orchestration method of the wireless access network according to any one of the embodiments.
Compared with the prior art, the AI reasoning task arrangement method, the device, the equipment and the storage medium of the wireless access network disclosed by the invention introduce an artificial intelligence technology at the RAN side, firstly receive the AI reasoning task guarantee strategy of the AI reasoning model sent by the non-real-time wireless controller, and acquire the terminal information of at least one terminal, wherein the AI reasoning task guarantee strategy defines the optimization target of the wireless access network and the characteristic information of the AI reasoning model, the terminal information comprises the computing resource information and the communication resource information, then generate an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information, and can carry out near-real-time AI reasoning task cooperation arrangement on the computing node at the RAN side according to the real-time communication state and the computing capability, thereby improving the efficiency of resource allocation and cooperation calculation at the RAN side and further improving the system performance.
Drawings
Fig. 1 is a flowchart of an AI reasoning task orchestration method for a wireless access network according to an embodiment of the present invention;
fig. 2 is a schematic diagram of information interaction between a near real-time wireless controller and a non-real-time wireless controller according to an embodiment of the present invention;
Fig. 3 is a flowchart of a terminal information acquisition manner of a first user equipment according to an embodiment of the present invention;
fig. 4 is a flowchart of a second method for obtaining terminal information of a user equipment according to an embodiment of the present invention;
Fig. 5 is a flowchart of another AI reasoning task orchestration method for a wireless access network according to an embodiment of the present invention;
Fig. 6 is a block diagram of a device for arranging AI reasoning tasks in a wireless access network according to an embodiment of the present invention;
fig. 7 is a block diagram of an AI reasoning task orchestration device of another wireless access network according to an embodiment of the present invention;
Fig. 8 is a block diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
With the advent of XR (Extended Reality), autopilot, industrial intelligent control, and other emerging applications, deep learning related algorithms such as image recognition have received more attention for computationally intensive tasks. The mobile device has limited capability and cannot meet strict calculation requirements and delay requirements, so that calculation tasks can be completed cooperatively by means of an AI edge calculation mode. The AI edge calculation mode can be realized by constructing a deep neural network model, and because the convolution layer in the deep neural network model structure is a feature map of an output tensor generated by performing dot multiplication operation on the spatial dimension of the input tensor by a kernel, the convolution layer can be used as a segmentation point of the model, and a plurality of calculation nodes can cooperatively complete the reasoning of the model. Meanwhile, the resource waste can be reduced by adding a branch classifier in the model structure and exiting inference in advance under the condition of sacrificing accuracy. Based on early exit mechanism and model segmentation technology, a plurality of researches on collaborative reasoning of edge computing nodes exist at present, and the precision requirements of time delay and reasoning are met under the limitation of computing resources and system bandwidth by selecting proper point of distinction and exit point for AI reasoning task. Therefore, in the embodiment of the invention, an AI technology is introduced at the RAN side to assist the terminal to complete the calculation task.
Referring to fig. 1, fig. 1 is a flowchart of an AI reasoning task orchestration method of a wireless access network according to an embodiment of the present invention, where the AI reasoning task orchestration method of the wireless access network is implemented by execution of a near real-time wireless controller, and the method includes:
s11, receiving an AI reasoning task guarantee strategy of an AI reasoning model sent by a non-real-time wireless controller;
s12, acquiring terminal information of at least one terminal;
And S13, generating an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
Illustratively, the O-RAN includes Near-Real-time controllers (Near-RT RICs) and Non-Real-time controllers (Non-Real-TIME RAN INTELLIGENT controllers, non-RT RICs), the terminals include, but are not limited to, base stations, centralized units including CU (Centralized Unit), CU-CPs (Centralized Unit-Control Plane), CU-UP (Centralized Unit-User Plane), distributed units DU (Distributed Unit), and User equipment UE (User Equipment). Referring to fig. 2, fig. 2 is a schematic information interaction diagram of a near-real-time wireless controller and a non-real-time wireless controller provided by an embodiment of the present invention, where the non-real-time controller is connected to the near-real-time controller through an open and standardized A1 interface, the purpose of the non-real-time controller is to provide a corresponding machine learning model to support RAN intelligence, and the non-real-time controller provides an AI reasoning model and data for the near-real-time controller, where the near-real-time controller performs a correlation operation by using an existing AI reasoning model when providing a corresponding function, where the AI reasoning model can perform one or more calculations of fault alarm analysis, coverage optimization, parameter optimization, spectrum analysis, inter-station coordination, mobility management, slice management, wireless positioning, and environment perception recognition. The near-real-time controller and the terminal perform information interaction through the E2 interface, the E2 interface and the A1 interface function in the O-RAN architecture are enhanced, the A1 interface is used for issuing an AI reasoning task guarantee strategy generated by the non-real-time controller according to model information of an AI reasoning model to the near-real-time controller, the E2 interface is used for collecting real-time terminal information, and near-real-time AI reasoning task collaborative arrangement is performed according to the terminal information and the AI reasoning task guarantee strategy, so that the resource allocation and collaborative calculation efficiency of the RAN side are improved, and the system performance is improved.
Specifically, in step S11, the non-real-time wireless controller generates the AI reasoning task security policy according to model information of an AI reasoning model, where the model information includes feature information of the AI reasoning model and performance security parameters of the AI reasoning task.
For example, the model information is obtained from the non-real-time wireless controller to SMO (SERVICE MANAGEMENT AND Orchestration, service management and flow orchestration framework), while for the source of the model information in SMO, the source may be directly configured by an operator or obtained by interaction with an external application, when the model information is obtained by interaction with an external application, a related interface or API (Application Programming Interface, application program interface) may be designed in advance, and a connection is established with the external application through an authorization and authentication mechanism, so as to obtain the model information.
Illustratively, the characteristic information of the AI inference model includes AI inference model ID, model layer number, type of each layer, output data amount of each layer, calculated amount of each layer, position of a segmentation point, position of an exit point, accuracy of an exit point, calculated amount of an exit point classifier, and the like. Wherein, each of the feature information may have a corresponding meaning with reference to table 1, and a specific example may have reference to table 2.
TABLE 1 characterization parameters and their corresponding meanings
Table 2ResNet-18 model characteristic information example
In table 2, conv represents a convolution layer, max pool represents a maximum pooling layer, avg pool represents an average pooling layer, fc represents a fully connected layer, exit represents an Exit point position, and the ResNet-18 model has four Exit points, namely Exit1 at layer 5, exit2 at layer 9, exit3 at layer 13, and Exit4 at layer 18, respectively, and the remaining layers can be used as dividing points. The exit point accuracy may evaluate the accuracy of its classification when this layer is selected as the exit point, the accuracy in the table being merely an example.
Illustratively, the performance guarantee parameters include, but are not limited to, model inference accuracy, number of inference calculations per unit time, model inference round trip delay, and model spectral efficiency. Wherein, the meaning corresponding to each performance guarantee parameter can refer to table 3.
TABLE 3 Performance guarantee parameters and their corresponding meanings
Further, after receiving the model information, the non-real-time wireless controller generates an AI reasoning task guarantee strategy according to the model information, wherein the AI reasoning task guarantee strategy defines an optimization target, optimizable parameters and characteristic information of an AI reasoning model of the wireless access network. The optimization targets of the wireless access network refer to the overall optimization targets of a protection strategy in the current system, such as maximizing model reasoning accuracy in the system, minimizing model reasoning round trip delay in the system and the like, and the optimizable parameters refer to parameters which can be regulated for AI reasoning task guarantee by the near-real-time wireless controller, such as segmentation point selection, exit point selection and the like of an AI reasoning model. The optimization objective of the wireless access network is to maximize the model reasoning accuracy in the system and minimize the model reasoning round trip delay in the system, namely, after the cut-off point and the exit point are selected, the maximum model reasoning accuracy and the minimum model reasoning round trip delay are required to be met, and the method can be obtained by carrying out polling calculation on each cut-off point and each exit point.
Specifically, in step S12, the terminal information includes, but is not limited to, computing resource information, communication resource information, current cell service user information, and preference information of the terminal for AI reasoning tasks. The computing resource information includes, but is not limited to, CPU (Central Processing Unit ) utilization rate, CPU frequency, CPU core binding state, residual CPU core number, CPU/GPU (Graphics Processing Unit, graphics processor)/NPU (Neural-network Processing Unit, embedded neural network processor) floating point operation number per second (FLOPS), GPU video memory capacity/residual video memory, GPU cuda core number/residual core number, etc. of the base station and the terminal, the communication resource information includes, but is not limited to, uplink/downlink total bandwidth, residual bandwidth, PRB (Physical Resource Block ) number, PRB utilization rate, etc. of the base station, and also includes, but is not limited to, channel condition of the terminal, communication link delay from the base station to the neighboring station, etc. such as SNR (Signal-to-Noise Ratio), RSSI (RECEIVED SIGNAL STRENGTH Indication of received Signal strength), etc., the current cell service user information includes, but is not limited to, user type identification (non-AI user or AI user), AI (e.g., resNet-18, AI), AI-based task model identification, AI-based on the terminal, etc., the terminal includes, and the task selection preference information includes, such as AI, and the task selection preference, such as AI, on each task selection, and the task selection.
It should be noted that, when the terminal is a base station or the base station where the centralized unit and the distributed unit are located, the terminal information may be directly sent to the near real-time wireless controller by the base station. When the terminal is a user equipment, two modes of acquiring terminal information of the user equipment are provided in the embodiment of the application, wherein the first mode is acquired through a base station, and the second mode is acquired directly through the user equipment. The base station according to the embodiments of the present application may take various forms, such as macro base station, micro base station, relay station, or access point. The base station may be an integrated base station or may be a base station comprising a centralized unit CU and a distributed unit DU.
In a first embodiment, when the terminal is a ue, the acquiring terminal information of at least one terminal includes sending a request message to a base station where the ue resides, so that the base station reports the terminal information of the ue according to the request message, or subscribing the terminal information to the base station where the ue resides, so that the base station reports the terminal information of the ue.
For example, referring to fig. 3, the near real-time wireless controller subscribes to or requests the terminal information from the base station where the user equipment resides through the E2 interface, and may be a periodic subscription or an event trigger report, where the event trigger includes, but is not limited to, a base station calculation power fluctuation, a terminal calculation power fluctuation, a base station bandwidth fluctuation, a terminal power fluctuation, a terminal segmentation mode preference change, and the like. And the base station collects air interface and terminal data of the user equipment according to the requirements, gathers terminal information and reports the terminal information of the user equipment to the near real-time wireless controller. In addition, the base station synchronously reports the terminal information of the base station to the near real-time wireless controller.
In a second embodiment, when the terminal is a user equipment, the user equipment supports information interaction with the near real-time wireless controller, and the acquiring the terminal information of at least one terminal includes sending a request message to the user equipment to enable the user equipment to report the terminal information according to the request message, or subscribing the terminal information to the user equipment to enable the user equipment to report the terminal information.
For example, referring to fig. 4, the user equipment and the near real-time wireless controller support corresponding interface application layer protocols on the basis of the existing protocol stack, so that information interaction between the two parties is realized. At this time, the near real-time wireless controller subscribes or requests the terminal information to the user equipment through the E2 interface, which may be periodical subscription or event triggering reporting, the user equipment collects air interface and terminal data according to the requirement, collects the terminal information, and reports the terminal information to the near real-time wireless controller. And the near real-time wireless controller synchronously sends a request message to the base station or subscribes to the terminal information of the base station so that the base station reports the terminal information of the base station.
Specifically, in step S13, the generating an AI reasoning task orchestration scheme of the radio access network according to the AI reasoning task guarantee policy and the terminal information includes:
S131, determining a node to be allocated and a corresponding computing node according to the optimization target of the wireless access network, and selecting a corresponding segmentation point and an exit point from the characteristic information of the AI reasoning model according to the information of each terminal, wherein the computing node is a base station, and the segmentation point and the exit point are one layer in the AI reasoning model;
S132, forming the AI reasoning task arrangement scheme by the nodes to be distributed, the corresponding computing nodes, the dividing points and the exiting points.
The node to be distributed is a terminal which cannot meet the calculation requirement and needs to share the reasoning calculation task by means of the calculation node, and the calculation node is a terminal which can receive the reasoning calculation task of the node to be distributed. After receiving terminal information sent by at least one terminal, the near real-time wireless controller determines nodes to be distributed and computing nodes, applies an optimization algorithm (such as a differential algorithm), evaluates the computing nodes required by the nodes to be distributed, and selects segmentation points and exit points. If the calculation resource information determines that one node to be allocated needs a larger calculation amount, more layers (namely, the layers between the cut-off point and the exit point are more) can be allocated to the node to be allocated, otherwise, fewer layers can be allocated, if the calculation node with larger residual bandwidth is needed for the node to be allocated according to the communication resource information, the calculation node meeting the condition is preferentially allocated as the calculation node of the node to be allocated, if the AI inference task model identification is given in one of the terminal information, the corresponding AI inference model is selected for carrying out the inference task according to the AI inference task model identification when the AI inference task arrangement scheme is generated, if the cut-off point selection preference of each AI inference task is given in one of the terminal information, and if the AI inference task arrangement scheme is generated, the cut-off point is preferentially selected according to the cut-off point selection preference of the node to be allocated.
Specifically, after generating an AI reasoning task orchestration scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information, the method further includes:
And S14, sending the AI reasoning task arrangement scheme to the nodes to be distributed and the corresponding computing nodes so that the nodes to be distributed and the corresponding computing nodes can complete respective reasoning computing tasks.
For example, assuming that the node to be allocated is a ue, the computing node is a base station, and terminal information of three ues and two base stations is collected at this time, where the two base stations and the three ues satisfy the conditions that a base station a provides a communication connection service for a ue 1, a base station B provides a communication connection service for a ue 2 and a ue 3, the base stations a and B both provide computing services, and the AI reasoning model is ResNet-18 in table 2. Suppose that at this time, the arrangement scheme is to select a suitable computing node for the user devices 1 to 3, and select a dividing point and an exit point for processing the AI reasoning task, where the AI reasoning task arrangement scheme is shown in table 4 below.
Table 4 AI example inference task orchestration scheme
| User equipment | Computing node | Parting point | Exit point |
| User equipment 1 | Base station A | Layer ID=3 | Exit2 |
| User equipment 2 | Base station A | Layer ID=8 | Exit3 |
| User equipment 3 | Base station B | Layer ID=11 | Exit3 |
For example, the user equipment 1 is taken as an example according to table 4 to illustrate, at this time, the computing node selected for the user equipment 1 is the base station a, that is, the base station a and the user equipment 1 together complete the computing task, the dividing point is the 3 rd layer, the Exit point is Exit2, the computing amount completed by the base station a is the computing amount from the 3 rd layer to the 9 th layer, according to the selected dividing point, the user equipment 1 completes the inference computation of the previous layers, after the feature map generated by the dividing point is transmitted to the base station a, the base station a completes the inference computation from the 3 rd layer to the 9 th layer, and finally the generated result is transmitted back to the user equipment 1. Thus, the choice of the cut point and the exit point determines the respective calculation amounts of the node to be allocated and the calculation node, and the choice of the exit point also determines the accuracy of the model.
Compared with the prior art, the AI reasoning task arrangement method of the wireless access network disclosed by the invention is characterized in that an artificial intelligence technology is introduced at the RAN side, an AI reasoning task guarantee strategy of an AI reasoning model sent by a non-real-time wireless controller is received, terminal information of at least one terminal is acquired, wherein the AI reasoning task guarantee strategy defines an optimization target of the wireless access network and characteristic information of the AI reasoning model, the terminal information comprises calculation resource information and communication resource information, then an AI reasoning task arrangement scheme of the wireless access network is generated according to the AI reasoning task guarantee strategy and the terminal information, and the calculation nodes, task dividing points and exit points are selected, so that near-real-time AI reasoning task cooperation arrangement can be performed on the calculation nodes at the RAN side according to real-time communication states and calculation capabilities, the efficiency of resource allocation and cooperation calculation at the RAN side is improved, and the system performance is further improved.
Referring to fig. 5, fig. 5 is a flowchart of another AI reasoning task orchestration method of a wireless access network according to an embodiment of the present invention, where the AI reasoning task orchestration method of the wireless access network is implemented by execution of a non-real-time wireless controller, and the method includes:
s21, obtaining model information of an AI reasoning model;
s22, generating an AI reasoning task guarantee strategy according to the model information;
and S23, sending the AI reasoning task guarantee strategy to a near-real-time wireless controller, so that the near-real-time wireless controller generates an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
Specifically, the model information comprises characteristic information of the AI reasoning model and performance guarantee parameters of the AI reasoning task, wherein the performance guarantee parameters comprise, but are not limited to, model reasoning accuracy and reasoning calculation times of unit time.
It should be noted that, the specific working process of the AI reasoning task orchestration method for a wireless access network according to the embodiment of the present invention may refer to the above embodiment, and will not be described herein again.
Referring to fig. 6, fig. 6 is a block diagram of a structure of an AI reasoning task orchestration device 100 of a wireless access network according to an embodiment of the present invention, where the AI reasoning task orchestration device 100 of the wireless access network includes:
The AI reasoning task guarantee strategy receiving module 11 is used for receiving an AI reasoning task guarantee strategy of an AI reasoning model sent by the non-real-time wireless controller, wherein the AI reasoning model is deployed in the wireless access network, and the AI reasoning task guarantee strategy defines the optimization target of the wireless access network and the characteristic information of the AI reasoning model;
A terminal information acquisition module 12, configured to acquire terminal information of at least one terminal, where the terminal information includes computing resource information and communication resource information;
And the AI reasoning task orchestration scheme generation module 13 is used for generating an AI reasoning task orchestration scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
Specifically, the AI reasoning task orchestration scheme generation module 13 is specifically configured to:
Determining a node to be allocated and a corresponding computing node according to the optimization target of the wireless access network, and selecting a corresponding segmentation point and an exit point from the characteristic information of the AI reasoning model according to the information of each terminal, wherein the computing node is a base station, and the segmentation point and the exit point are one layer of the AI reasoning model;
And forming the AI reasoning task arrangement scheme by the nodes to be distributed and the corresponding computing nodes, the dividing points and the exit points.
Specifically, the AI reasoning task orchestration device 100 of the radio access network further comprises:
and the AI reasoning task arrangement scheme method sending module is used for sending the AI reasoning task arrangement scheme to the nodes to be distributed and the corresponding computing nodes so as to enable the nodes to be distributed and the corresponding computing nodes to complete respective reasoning computing tasks.
Specifically, when the terminal is a user equipment, the terminal information obtaining module 12 is specifically configured to send a request message to a base station where the user equipment resides, so that the base station reports terminal information of the user equipment according to the request message, or subscribe the base station where the user equipment resides with the terminal information, so that the base station reports the terminal information of the user equipment.
Specifically, when the terminal is a user equipment, the user equipment supports information interaction with the near real-time wireless controller, and the terminal information acquisition module 12 is specifically configured to send a request message to the user equipment so that the user equipment reports terminal information according to the request message, or subscribe the user equipment with the terminal information so that the user equipment reports the terminal information.
Referring to fig. 7, fig. 7 is a block diagram illustrating a structure of an AI reasoning task orchestration device 200 of another wireless access network according to an embodiment of the present invention, where the AI reasoning task orchestration device 200 of the wireless access network includes:
A model information acquisition module 21 for acquiring model information of an AI inference model;
The AI reasoning task guarantee strategy generation module 22 is used for generating an AI reasoning task guarantee strategy according to the model information, wherein the AI reasoning model is deployed in the wireless access network, and the AI reasoning task guarantee strategy defines the optimization target of the wireless access network and the characteristic information of the AI reasoning model;
And the AI reasoning task guarantee strategy sending module 23 is used for sending the AI reasoning task guarantee strategy to a near-real-time wireless controller so that the near-real-time wireless controller can generate an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and terminal information, wherein the terminal information comprises computing resource information and communication resource information.
Specifically, the model information comprises characteristic information of the AI reasoning model and performance guarantee parameters of the AI reasoning task, wherein the performance guarantee parameters comprise, but are not limited to, model reasoning accuracy and reasoning calculation times of unit time.
Compared with the prior art, the AI reasoning task arrangement device of the wireless access network disclosed by the invention introduces an artificial intelligence technology at the RAN side, firstly receives an AI reasoning task guarantee strategy of an AI reasoning model sent by a non-real-time wireless controller, and acquires terminal information of at least one terminal, wherein the AI reasoning task guarantee strategy defines an optimization target of the wireless access network and characteristic information of the AI reasoning model, the terminal information comprises computing resource information and communication resource information, and then generates an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
Referring to fig. 8, fig. 8 is a block diagram of an electronic device 300 according to an embodiment of the present invention, where the electronic device 300 includes a processor 31, a memory 32, and a computer program stored in the memory 32 and executable on the processor 31. The processor 31 executes the computer program to implement the steps in the embodiments of the AI reasoning task scheduling method of each radio access network, such as steps S11 to S13 and S21 to S23.
Illustratively, the computer program may be partitioned into one or more modules/units that are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing the specified functions, which instruction segments are used to describe the execution of the computer program in the electronic device 300.
The electronic device 300 may include, but is not limited to, a processor 31, a memory 32. It will be appreciated by those skilled in the art that the schematic diagram is merely an example of the electronic device 300 and is not limiting of the electronic device 300, and may include more or fewer components than shown, or may combine certain components, or different components, e.g., the electronic device 300 may also include input-output devices, network access devices, buses, etc.
The Processor 31 may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (DIGITAL SIGNAL Processor, DSP), application SPECIFIC INTEGRATED Circuit (ASIC), off-the-shelf Programmable gate array (Field-Programmable GATE ARRAY, FPGA) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, and the processor 31 is a control center of the electronic device 300, and connects various parts of the entire electronic device 300 using various interfaces and lines.
The memory 32 may be used to store the computer programs and/or modules, and the processor 31 may implement various functions of the electronic device 300 by executing or executing the computer programs and/or modules stored in the memory 32, and invoking data stored in the memory 32. The memory 32 may mainly include a storage program area that may store an operating system, an application program required for at least one function (such as a sound playing function, an image playing function, etc.), etc., and a storage data area that may store data created according to the use of the cellular phone (such as audio data, a phonebook, etc.), etc. In addition, the memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart memory card (SMART MEDIA CARD, SMC), secure Digital (SD) card, flash memory card (FLASH CARD), at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device.
Wherein the integrated modules/units of the electronic device 300 may be stored in a computer readable storage medium if implemented in the form of software functional units and sold or used as a stand alone product. Based on such understanding, the present invention may implement all or part of the flow of the method of the above embodiment, or may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and the computer program may implement the steps of each of the method embodiments described above when executed by the processor 31. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth.
While the foregoing is directed to the preferred embodiments of the present invention, it will be appreciated by those skilled in the art that changes and modifications may be made without departing from the principles of the invention, such changes and modifications are also intended to be within the scope of the invention.
Claims (11)
1. An AI reasoning task orchestration method for a wireless access network, applied to a near real-time wireless controller, the method comprising:
Receiving an AI (advanced technology attachment) reasoning task guarantee strategy of an AI reasoning model sent by a non-real-time wireless controller, wherein the AI reasoning model is deployed in a wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
Acquiring terminal information of at least one terminal, wherein the terminal information comprises computing resource information and communication resource information;
And generating an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
2. The AI-inference task orchestration method for a wireless access network according to claim 1, wherein the generating an AI-inference task orchestration scheme for a wireless access network according to the AI-inference task provisioning policy and the terminal information comprises:
Determining a node to be allocated and a corresponding computing node according to the optimization target of the wireless access network, and selecting a corresponding dividing point and an exiting point from the characteristic information of the AI reasoning model according to the information of each terminal, wherein the dividing point and the exiting point are one layer of the AI reasoning model;
And forming the AI reasoning task arrangement scheme by the nodes to be distributed and the corresponding computing nodes, the dividing points and the exit points.
3. The AI-inference task orchestration method for a wireless access network according to claim 2, wherein after generating an AI-inference task orchestration scheme for a wireless access network according to the AI-inference task provisioning policies and the terminal information, the method further comprises:
and sending the AI reasoning task arrangement scheme to the nodes to be distributed and the corresponding computing nodes so that the nodes to be distributed and the corresponding computing nodes finish respective reasoning computing tasks.
4. The AI-inference task orchestration method of a radio access network according to claim 1, wherein when the terminal is a user equipment, the obtaining terminal information of at least one terminal comprises:
a request message is sent to a base station where the user equipment resides, so that the base station reports the terminal information of the user equipment according to the request message, or,
And subscribing the terminal information to a base station where the user equipment resides so that the base station reports the terminal information of the user equipment.
5. The AI reasoning task orchestration method of the wireless access network of claim 1, wherein when the terminal is a user equipment, the user equipment supports information interaction with a near real-time wireless controller, and wherein the obtaining terminal information of at least one terminal comprises:
a request message is sent to the user equipment to enable the user equipment to report terminal information according to the request message, or,
And subscribing the terminal information to the user equipment so as to enable the user equipment to report the terminal information.
6. An AI reasoning task orchestration method for a wireless access network, applied to a non-real-time wireless controller, the method comprising:
Obtaining model information of an AI reasoning model;
An AI reasoning task guarantee strategy is generated according to the model information, wherein the AI reasoning model is deployed in the wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
and sending the AI reasoning task guarantee strategy to a near-real-time wireless controller so that the near-real-time wireless controller generates an AI reasoning task arrangement scheme of a wireless access network according to the AI reasoning task guarantee strategy and terminal information, wherein the terminal information comprises computing resource information and communication resource information.
7. The AI reasoning task orchestration method of claim 6, wherein the model information includes characteristic information of the AI reasoning model and performance guarantee parameters of the AI reasoning task, wherein the performance guarantee parameters include, but are not limited to, model reasoning accuracy and number of reasoning calculations per unit time.
8. An AI reasoning task orchestration device of a radio access network, comprising:
The system comprises an AI (advanced technology attachment) reasoning task guarantee strategy receiving module, an AI (advanced technology attachment) reasoning task guarantee strategy processing module and a non-real-time wireless controller, wherein the AI reasoning task guarantee strategy receiving module is used for receiving an AI reasoning task guarantee strategy of an AI reasoning model sent by the non-real-time wireless controller, the AI reasoning model is deployed in a wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
The terminal information acquisition module is used for acquiring terminal information of at least one terminal, wherein the terminal information comprises computing resource information and communication resource information;
And the AI reasoning task arrangement scheme generation module is used for generating an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and the terminal information.
9. An AI reasoning task orchestration device of a radio access network, comprising:
the model information acquisition module is used for acquiring the model information of the AI reasoning model;
the system comprises an AI reasoning task guarantee strategy generation module, an AI reasoning task guarantee strategy generation module and a model information generation module, wherein the AI reasoning task guarantee strategy generation module is used for generating an AI reasoning task guarantee strategy according to the model information, the AI reasoning model is deployed in a wireless access network, and an optimization target of the wireless access network and characteristic information of the AI reasoning model are defined in the AI reasoning task guarantee strategy;
And the AI reasoning task guarantee strategy sending module is used for sending the AI reasoning task guarantee strategy to a near-real-time wireless controller so that the near-real-time wireless controller can generate an AI reasoning task arrangement scheme of the wireless access network according to the AI reasoning task guarantee strategy and terminal information, wherein the terminal information comprises computing resource information and communication resource information.
10. An electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the AI inference task orchestration method of the radio access network according to any one of claims 1-7 when the computer program is executed.
11. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored computer program, wherein the computer program when run controls a device in which the computer readable storage medium is located to perform the AI reasoning task orchestration method of the radio access network according to any one of claims 1-7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311733048.9A CN120166409B (en) | 2023-12-15 | 2023-12-15 | Methods, apparatus, devices and storage media for orchestrating AI inference tasks in wireless access networks |
| PCT/CN2024/137406 WO2025124306A1 (en) | 2023-12-15 | 2024-12-06 | Ai inference task orchestration method for radio access network, apparatus, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311733048.9A CN120166409B (en) | 2023-12-15 | 2023-12-15 | Methods, apparatus, devices and storage media for orchestrating AI inference tasks in wireless access networks |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN120166409A CN120166409A (en) | 2025-06-17 |
| CN120166409B true CN120166409B (en) | 2026-01-16 |
Family
ID=96002545
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311733048.9A Active CN120166409B (en) | 2023-12-15 | 2023-12-15 | Methods, apparatus, devices and storage media for orchestrating AI inference tasks in wireless access networks |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN120166409B (en) |
| WO (1) | WO2025124306A1 (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120529342B (en) * | 2025-07-25 | 2025-11-07 | 中国电信股份有限公司 | Model inference methods, devices, network equipment, readable storage media, and program products |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022060777A1 (en) * | 2020-09-17 | 2022-03-24 | Intel Corporation | Online reinforcement learning |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113875278B (en) * | 2019-05-24 | 2025-03-11 | 苹果公司 | 5G New Radio Load Balancing and Mobility Robustness |
| US11902104B2 (en) * | 2020-03-04 | 2024-02-13 | Intel Corporation | Data-centric service-based network architecture |
| WO2023283192A1 (en) * | 2021-07-06 | 2023-01-12 | Intel Corporation | A1 policy functions for open radio access network (o-ran) systems |
| WO2023172292A2 (en) * | 2021-08-25 | 2023-09-14 | Northeastern University | Zero-touch deployment and orchestration of network intelligence in open ran systems |
| CN113923694B (en) * | 2021-12-14 | 2022-05-03 | 网络通信与安全紫金山实验室 | Network resource arranging method, system, device and storage medium |
| CN116301912A (en) * | 2022-12-14 | 2023-06-23 | 平安银行股份有限公司 | Inference service deployment method, device, equipment and storage medium |
| CN115988580A (en) * | 2022-12-28 | 2023-04-18 | 中国联合网络通信集团有限公司 | Radio resource control method, radio resource control device, electronic device, and storage medium |
| CN116708443A (en) * | 2023-07-24 | 2023-09-05 | 中国电信股份有限公司 | Multi-level computing power network task scheduling method and device |
-
2023
- 2023-12-15 CN CN202311733048.9A patent/CN120166409B/en active Active
-
2024
- 2024-12-06 WO PCT/CN2024/137406 patent/WO2025124306A1/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022060777A1 (en) * | 2020-09-17 | 2022-03-24 | Intel Corporation | Online reinforcement learning |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2025124306A1 (en) | 2025-06-19 |
| CN120166409A (en) | 2025-06-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Multi-tenant cross-slice resource orchestration: A deep reinforcement learning approach | |
| CN114007225B (en) | BWP distribution methods, devices, electronic equipment, and computer-readable storage media | |
| CN111246586B (en) | Method and system for distributing smart grid resources based on genetic algorithm | |
| WO2023172292A9 (en) | Zero-touch deployment and orchestration of network intelligence in open ran systems | |
| Donatti et al. | Survey on machine learning-enabled network slicing: Covering the entire life cycle | |
| Perveen et al. | Dynamic traffic forecasting and fuzzy-based optimized admission control in federated 5G-open RAN networks | |
| He et al. | Edge-aided computing and transmission scheduling for LTE-U-enabled IoT | |
| US20230325258A1 (en) | Method and apparatus for autoscaling containers in a cloud-native core network | |
| Wang et al. | Optimal QoS-aware channel assignment in D2D communications with partial CSI | |
| CN113159145B (en) | A feature engineering arrangement method and device | |
| D’Oro et al. | OrchestRAN: Orchestrating network intelligence in the open RAN | |
| CN113994732B (en) | Network data analysis function NWDAF entity and its execution method | |
| JP2021083058A (en) | Control device, control method, and program | |
| Nadeem et al. | Efficient resource allocation using distributed edge computing in D2D based 5G-HCN with network slicing | |
| Bajracharya et al. | Dynamic pricing for intelligent transportation system in the 6G unlicensed band | |
| CN120166409B (en) | Methods, apparatus, devices and storage media for orchestrating AI inference tasks in wireless access networks | |
| WO2023203356A1 (en) | Distributed mobile network traffic data decomposition and forecasting method and apparatus | |
| Gupta et al. | Resource orchestration in network slicing using GAN-based distributional deep Q-network for industrial applications: RK Gupta et al. | |
| Pratap | Cyclic stable matching inspired resource provisioning for IoT-enabled 5G networks | |
| Staffolani et al. | PRORL: Proactive resource orchestrator for open RANs using deep reinforcement learning | |
| Cao et al. | Learning-based multitier split computing for efficient convergence of communication and computation | |
| Yang et al. | Performance evaluation of priority based adaptive multiguard channel call admission control for multiclass services in mobile networks | |
| WO2024114772A1 (en) | Slice configuration method and apparatus, and storage medium | |
| Han et al. | Cell-less offloading of distributed learning tasks in multi-access edge computing | |
| Alsaket et al. | Exploring evolutionary multi-objective techniques in self-organizing networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |