A model inference method, a model inference device, and a storage medium

Abstract

This disclosure relates to a model inference method, a model inference apparatus, and a storage medium. The model inference method, applied to an Operations, Maintenance, and Management (OAM) entity, includes: in response to receiving model subscription request information sent by a control radio access network (CNR) device, determining a first model corresponding to the model subscription request information; segmenting the first model to obtain a first number of model segment blocks; and distributing the first number of model segment blocks to a first number of CNR devices. This disclosure can effectively improve model inference efficiency, reduce inference latency, and help balance network load.

Description

Technical Field

[0001] This disclosure relates to the field of wireless communication technology, and in particular to a model inference method, a model inference device, and a storage medium. Background Technology

[0002] In next-generation communication technologies, decision-making regarding network intelligence and automation requires the use of artificial intelligence and machine learning to acquire a large amount of available data, including data collected by terminals and network-side devices. Based on this data, machine learning algorithms are used for inference and training to extract relevant models at different levels.

[0003] In related technologies, terminals request model subscriptions from Operation Administration and Maintenance (OAM) network elements via wireless access devices. The entire inference process for these models is handled by the OAM network element. During inference, all model inference data needs to be uploaded to the OAM, and the OAM also needs to train the model based on this data. Therefore, when the OAM receives multiple model subscription requests simultaneously, it cannot satisfy all requests and provide the necessary inference results, leading to increased latency in model inference feedback and reduced system efficiency. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this disclosure provides a model reasoning method, a model reasoning device, and a storage medium.

[0005] According to a first aspect of the present disclosure, a model reasoning method is provided, applied to an Operation and Maintenance Management (OAM) entity, the method comprising:

[0006] In response to receiving a model subscription request information sent by a control radio access network device, a first model corresponding to the model subscription request information is determined; the first model is segmented to obtain a first number of model segmentation blocks, and the first number of model segmentation blocks are distributed to a first number of control radio access network devices.

[0007] In one embodiment, each model segmentation block in the first number of model segmentation blocks corresponds to allocation information;

[0008] The allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block.

[0009] In one embodiment, the first number of control radio access network devices includes a first control radio access network device, which is a control radio access network device accessed by a terminal.

[0010] The step of distributing the first number of model segmentation blocks to the first number of control radio access network devices includes:

[0011] Among the control radio access network devices adjacent to the first control radio access network device, a plurality of auxiliary control radio access network devices are identified.

[0012] Among the plurality of auxiliary control wireless access network devices, a second number of control wireless access network devices is determined based on the computing power occupancy status and load of each auxiliary control wireless access network device; the second number of control wireless access network devices are the other control wireless access network devices in the first number besides the first control wireless access devices.

[0013] Based on the inference order of the first number of model segmentation blocks, the first model segmentation block is sent to the first control radio access network device, and the remaining number of model segmentation blocks are distributed to the second number of control radio access network devices.

[0014] In one embodiment, the model reasoning method further includes:

[0015] Receive model performance update data sent by the first control radio access network device; update the first model based on the model performance update data, determine the updated model parameters of the first model, and send the updated model parameters of the first model to the first control radio access network device.

[0016] In one embodiment, the model reasoning method further includes:

[0017] In response to receiving a first model analysis subscription update request, the distributed radio access network device accessed by the terminal is updated; wherein, the first model analysis subscription update request instructs the terminal to switch the distributed radio access network device, but does not switch the control radio access network device;

[0018] or

[0019] In response to receiving a second model analysis subscription update request, the distributed radio access network device accessed by the terminal is updated, and the first model is re-segmented; wherein, the second model analysis subscription update request instructs the terminal to switch the distributed radio access network device and switch the control radio access network device.

[0020] According to a second aspect of the present disclosure, a model inference method is provided for controlling a wireless access network device, the method comprising:

[0021] In response to receiving a model analysis subscription request sent by a distributed radio access network device, the system processes the model analysis subscription request to obtain model subscription request information and sends the model subscription request information to the OAM; it also receives a model segmentation block sent by the OAM; the model segmentation block is a model segmentation block determined by segmenting a first model; the first model is determined by the OAM based on the model subscription request information.

[0022] In one implementation, after sending the model subscription request information to OAM, the method further includes:

[0023] A model inference data request is sent to a distributed radio access network device to obtain model inference data; the model segmentation block is inferred based on the model inference data to obtain inference intermediate information of the model segmentation block.

[0024] In one embodiment, the model segmentation block corresponds to allocation information; the allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block;

[0025] The model inference method also includes:

[0026] In response to the fact that the control radio access network device is not the last control radio access network device, the intermediate inference information is sent to the next control radio access network device based on the inference order; in response to the fact that the control radio access network device is the last control radio access network device, after the model inference is completed, the first inference result corresponding to the first model is determined, and the first inference result is sent to the first control radio access network device, wherein the first control radio access network device is the control radio access network device accessed by the terminal.

[0027] In one embodiment, the method further includes:

[0028] In response to the fact that the control radio access network device is a first control radio access network device, the first inference result is received; the first inference result is sent to a first distributed radio access network device, wherein the first distributed radio access network device is a distributed radio access network device accessed by the terminal.

[0029] In one embodiment, after sending the first inference result to the first distributed radio access network device, the model inference method further includes:

[0030] The system receives performance data sent by a first distributed wireless access network device, wherein the performance data is the actual performance data of the terminal after adjusting the execution strategy based on a first model; the system processes the performance data to obtain model performance update data, and sends the model performance update data to the OAM.

[0031] In one implementation, sending a model subscription request message to OAM includes:

[0032] In response to the fact that the control radio access network device is a first control radio access network device, a model subscription request information is sent to OAM; wherein, the first control radio access network device is the control radio access network device corresponding to the first distributed wireless network device accessed by the terminal.

[0033] In one embodiment, the model reasoning method further includes:

[0034] In response to the fact that the control radio access network device is the first control radio access network device, if a model analysis subscription request is received again, a second distributed radio access network device is determined to resend the model analysis subscription request; the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching distributed radio access network devices; the first inference result is sent to the second distributed radio access network device, and a model subscription update request is sent to OAM.

[0035] In one embodiment, the model reasoning method further includes:

[0036] In response to the fact that the control radio access network device is the second control radio access network device, if a model analysis subscription request is received again, the system determines that the second distributed radio access network device resends the model analysis subscription request, and the second control radio access network device that received the model analysis subscription request again is also determined. The second control radio access network device is the control radio access network device corresponding to the second distributed radio access network device. The second distributed radio access network device is the distributed radio access network device that the terminal re-accesses after switching to the distributed radio access network device. The system then sends the first inference result to the second control radio access network device and sends a model subscription update request to the OAM.

[0037] According to a third aspect of the present disclosure, a model inference method is provided, applied to a distributed wireless access network device, the method comprising:

[0038] In response to receiving a model analysis subscription request from a terminal, the system sends the model analysis subscription request to the control radio access network device; wherein the model analysis subscription request is used to obtain a first model from the OAM; the first model includes a first number of model segmentation blocks.

[0039] In one embodiment, the method further includes:

[0040] The system receives a model inference data request sent by a control radio access network device, the model inference data request being used to obtain model inference data; obtains the model inference data from the terminal, and sends it to the control radio access network device.

[0041] In one embodiment, the method further includes:

[0042] In response to the distributed radio access network device being a first distributed radio access network device, the device receives a first inference result sent by a first control radio access network device and sends the first inference result to the terminal.

[0043] In one embodiment, after sending the first inference result to the terminal, the method further includes:

[0044] In response to the distributed radio access network device being a first distributed radio access network device, the device receives performance data sent by the terminal; the performance data is the actual performance data after the terminal adjusts the execution strategy based on the first model; and sends the performance data to the first control radio access network device.

[0045] In one embodiment, the method further includes:

[0046] In response to the fact that the radio access network device is a second distributed radio access network device, if a model analysis subscription request is received from the terminal, it is determined to send a model analysis subscription request to the control radio access network device corresponding to the second distributed radio access network device; wherein, the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching distributed radio access network devices.

[0047] According to a fourth aspect of the present disclosure, a model inference apparatus is provided, applied to an Operation and Maintenance Management (OAM) entity, the apparatus comprising:

[0048] The determining module is configured to, in response to receiving a model subscription request information sent by a control radio access network device, determine a first model corresponding to the model subscription request information; the sending module is configured to, divide the first model into a first number of model segmentation blocks, and distribute the first number of model segmentation blocks to a first number of control radio access network devices.

[0049] In one embodiment, each model segmentation block in the first number of model segmentation blocks corresponds to allocation information;

[0050] The allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block.

[0051] In one embodiment, the first number of control radio access network devices includes a first control radio access network device, which is a control radio access network device accessed by a terminal.

[0052] The sending module is used for:

[0053] Among the control radio access network devices adjacent to the first control radio access network device, a plurality of auxiliary control radio access network devices are identified.

[0054] Among the plurality of auxiliary control wireless access network devices, a second number of control wireless access network devices is determined based on the computing power occupancy status and load of each auxiliary control wireless access network device; the second number of control wireless access network devices are the other control wireless access network devices in the first number besides the first control wireless access devices.

[0055] Based on the inference order of the first number of model segmentation blocks, the first model segmentation block is sent to the first control radio access network device, and the remaining number of model segmentation blocks are distributed to the second number of control radio access network devices.

[0056] In one embodiment, the model inference device further includes: a receiving module;

[0057] The receiving module shown is used to receive model performance update data sent by the first control radio access network device; update the first model based on the model performance update data, determine the updated model parameters of the first model, and send the updated model parameters of the first model to the first control radio access network device.

[0058] In one embodiment, the receiving module is further configured to:

[0059] In response to receiving a first model analysis subscription update request, the distributed radio access network device accessed by the terminal is updated; wherein, the first model analysis subscription update request instructs the terminal to switch the distributed radio access network device, but does not switch the control radio access network device;

[0060] or

[0061] In response to receiving a second model analysis subscription update request, the distributed radio access network device accessed by the terminal is updated, and the first model is re-segmented; wherein, the second model analysis subscription update request instructs the terminal to switch the distributed radio access network device and switch the control radio access network device.

[0062] According to a fifth aspect of the present disclosure, a model inference apparatus is provided for controlling a wireless access network device, the apparatus comprising:

[0063] The sending module is configured to respond to a received model analysis subscription request sent by a distributed radio access network device, process the model analysis subscription request to obtain model subscription request information, and send the model subscription request information to the OAM; the receiving module is configured to receive a model segmentation block sent by the OAM; the model segmentation block is a model segmentation block determined by segmenting a first model; the first model is determined by the OAM based on the model subscription request information.

[0064] In one embodiment, the sending module is further configured to:

[0065] A model inference data request is sent to a distributed radio access network device to obtain model inference data; the model segmentation block is inferred based on the model inference data to obtain inference intermediate information of the model segmentation block.

[0066] In one embodiment, the model segmentation block corresponds to allocation information; the allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block;

[0067] The sending module is also used for:

[0068] In response to the fact that the control radio access network device is not the last control radio access network device, the intermediate inference information is sent to the next control radio access network device based on the inference order; in response to the fact that the control radio access network device is the last control radio access network device, after the model inference is completed, the first inference result corresponding to the first model is determined, and the first inference result is sent to the first control radio access network device, wherein the first control radio access network device is the control radio access network device accessed by the terminal.

[0069] In one embodiment, the sending module is further configured to:

[0070] In response to the fact that the control radio access network device is a first control radio access network device, the first inference result is received; the first inference result is sent to a first distributed radio access network device, wherein the first distributed radio access network device is a distributed radio access network device accessed by the terminal.

[0071] In one embodiment, after sending the first inference result to the first distributed radio access network device, the receiving module is further configured to:

[0072] The system receives performance data sent by a first distributed wireless access network device, wherein the performance data is the actual performance data of the terminal after adjusting the execution strategy based on a first model; the system processes the performance data to obtain model performance update data, and sends the model performance update data to the OAM.

[0073] In one embodiment, the sending module is further configured to:

[0074] In response to the fact that the control radio access network device is a first control radio access network device, a model subscription request information is sent to OAM; wherein, the first control radio access network device is the control radio access network device corresponding to the first distributed wireless network device accessed by the terminal.

[0075] In one embodiment, the sending module is further configured to:

[0076] In response to the fact that the control radio access network device is the first control radio access network device, if a model analysis subscription request is received again, a second distributed radio access network device is determined to resend the model analysis subscription request; the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching distributed radio access network devices; the first inference result is sent to the second distributed radio access network device, and a model subscription update request is sent to OAM.

[0077] In one embodiment, the sending module is further configured to:

[0078] In response to the fact that the control radio access network device is the second control radio access network device, if a model analysis subscription request is received again, the system determines that the second distributed radio access network device resends the model analysis subscription request, and the second control radio access network device that received the model analysis subscription request again is also determined. The second control radio access network device is the control radio access network device corresponding to the second distributed radio access network device. The second distributed radio access network device is the distributed radio access network device that the terminal re-accesses after switching to the distributed radio access network device. The system then sends the first inference result to the second control radio access network device and sends a model subscription update request to the OAM.

[0079] According to a sixth aspect of the present disclosure, a model inference apparatus is provided, applied to a distributed wireless access network device, the apparatus comprising:

[0080] The sending module is configured to send the model analysis subscription request to the control radio access network device in response to receiving the model analysis subscription request sent by the terminal; wherein the model analysis subscription request is used to obtain a first model from the OAM; the first model includes a first number of model segmentation blocks.

[0081] In one embodiment, the device further includes: a receiving module;

[0082] The receiving module is used to receive a model inference data request sent by the control radio access network device, wherein the model inference data request is used to obtain model inference data; obtain model inference data from the terminal, and send it to the control radio access network device.

[0083] In one embodiment, the receiving module is further configured to:

[0084] In response to the distributed radio access network device being a first distributed radio access network device, the device receives a first inference result sent by a first control radio access network device and sends the first inference result to the terminal.

[0085] In one embodiment, the receiving module is further configured to:

[0086] In response to the distributed radio access network device being a first distributed radio access network device, the device receives performance data sent by the terminal; the performance data is the actual performance data after the terminal adjusts the execution strategy based on the first model; and sends the performance data to the first control radio access network device.

[0087] In one embodiment, the receiving module is further configured to:

[0088] In response to the fact that the radio access network device is a second distributed radio access network device, if a model analysis subscription request is received from the terminal, it is determined to send a model analysis subscription request to the control radio access network device corresponding to the second distributed radio access network device; wherein, the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching distributed radio access network devices.

[0089] According to a seventh aspect of the present disclosure, a model inference apparatus is provided, comprising:

[0090] A processor; a memory for storing processor-executable instructions; wherein the processor is configured to: execute the model inference method described in the first aspect or any embodiment of the first aspect, or execute the model inference method described in the second aspect or any embodiment of the second aspect, or execute the model inference method described in the third aspect or any embodiment of the third aspect.

[0091] According to an eighth aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein when instructions in the storage medium are executed by a processor of a mobile terminal, the mobile terminal is enabled to execute the model inference method described in the first aspect or any embodiment of the first aspect, or to execute the model inference method described in the second aspect or any embodiment of the second aspect, or to execute the model inference method described in the third aspect or any embodiment of the third aspect.

[0092] The technical solutions provided by the embodiments of this disclosure can include the following beneficial effects: By segmenting the model using the OAM method of this disclosure and distributing the model segmentation blocks to different control radio access network devices, the AI ​​processing capabilities of the radio access network devices can be better developed, solving the problem of insufficient AI processing capabilities of the radio access network devices, and facilitating network load balancing. By fully utilizing local AI processing capabilities, model inference efficiency can be effectively improved, inference latency reduced, and network load balancing can be achieved, providing users with efficient and convenient AI analysis services.

[0093] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0094] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0095] Figure 1 A schematic diagram of a basic functional framework structure is shown according to an exemplary embodiment.

[0096] Figure 2 A schematic diagram of a network architecture is shown according to an exemplary embodiment.

[0097] Figure 3 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment.

[0098] Figure 4 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0099] Figure 5 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0100] Figure 6 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0101] Figure 7 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0102] Figure 8 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0103] Figure 9 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0104] Figure 10This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0105] Figure 11 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0106] Figure 12 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0107] Figure 13 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0108] Figure 14 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0109] Figure 15 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0110] Figure 16 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0111] Figure 17 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0112] Figure 18 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0113] Figure 19 This is a flowchart illustrating yet another model reasoning method according to an exemplary embodiment.

[0114] Figure 20 This is a flowchart illustrating terminal switching in a model inference method according to an exemplary embodiment.

[0115] Figure 21 This is a flowchart illustrating terminal switching in a model inference method according to an exemplary embodiment.

[0116] Figure 22 This is a schematic diagram illustrating the protocol and interface of a model reasoning method according to an exemplary embodiment.

[0117] Figure 23 This is a schematic diagram illustrating the protocol and interface principles for AI analysis task delivery when a terminal switches between the same gNB-CU, according to an exemplary embodiment of a model-free inference method.

[0118] Figure 24This is a schematic diagram illustrating the protocol and interface principles for AI analysis task delivery during terminal handover across gNB-CUs in a model-free inference method, according to an exemplary embodiment.

[0119] Figure 25 This is a block diagram of a model reasoning device according to an exemplary embodiment.

[0120] Figure 26 This is a block diagram of another model reasoning device according to an exemplary embodiment.

[0121] Figure 27 This is a block diagram of another model reasoning device according to an exemplary embodiment.

[0122] Figure 28 This is a block diagram illustrating a model inference apparatus according to an exemplary embodiment.

[0123] Figure 29 This is a block diagram illustrating yet another model reasoning apparatus according to an exemplary embodiment. Detailed Implementation

[0124] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0125] Decision-making in next-generation intelligent and automated behaviors requires the use of artificial intelligence and machine learning to acquire a large amount of available data, including data collected from terminals and the network side. Machine learning algorithms are then used to mine the input data and extract relevant models at different levels, which are then used to drive the process. To realize big data-enabled AI wireless networks, key technologies such as AI-supporting wireless network frameworks, the functions of AI modules / platforms, their inputs and outputs, and their relationship with wireless network elements are urgent research issues.

[0126] Therefore, this research project focuses on enhancing the intelligence of Radio Access Network (RAN) equipment: a study on NR and ENDC data acquisition. It discusses the design principles, basic concepts, applicable cases, and standard impacts. Figure 1 A schematic diagram of a basic functional framework structure is shown according to an exemplary embodiment, such as... Figure 1 As shown, this is a potential wireless network architecture that supports artificial intelligence, serving as an initial architecture.

[0127] The data collection and preparation unit includes data acquisition and data preprocessing functions. Data acquisition can be performed on multiple network elements, and the data provided includes measurement data, feedback performance data, and model performance data.

[0128] Model Training Unit: Iterates through computation and processing to obtain a better model for inference. Inputs include training data and model performance feedback.

[0129] Model inference unit: Uses a trained machine learning model to generate predictions or decisions.

[0130] Action: Utilizes model inference results to formulate and execute strategies, and feeds back relevant performance results to the Data collection after execution.

[0131] The above-described wireless network AI architecture injects intelligent power into enhancing the user experience of wireless network terminals. To maintain the continuity and accuracy of wireless network AI analysis services and improve the operational efficiency of wireless network AI, further standardization and optimization of the interactions between various AI functional units are needed to make the wireless network AI architecture more adaptable and scalable.

[0132] In related technologies, after the terminal initiates a model analysis subscription request, inference and training are all performed by OAM. Figure 2 A schematic diagram of a network architecture is shown according to an exemplary embodiment. For example... Figure 2 As shown, the system includes a terminal, a gNB-DU, a gNB-CU, and an OAM. The terminal accesses the gNB-DU via a wireless channel, multiple gNB-DUs access the gNB-CU via F1 interfaces, and the gNB-CUs are connected to each other via Xn interfaces. The OAM is mainly responsible for the model training function unit in the AI-supporting wireless network architecture, handling model training and model segmentation. The gNB-CU is responsible for the model inference function unit, performing model inference. The gNB-DU is mainly responsible for the data collection function unit, collecting real-time inference data and terminal performance feedback data. The terminal is responsible for the action execution function unit, making corresponding strategy adjustments based on the model inference results.

[0133] The terminal is responsible for the Action function; the Next Generation Node B Distributed Unit (gNB-DU) forwards the terminal's analysis requests and inference results, and performs the Data Collection function. The Next Generation Node B Control Unit (gNB-CU) forwards the terminal's analysis requests and inference results, and performs the Data Collection function. The OAM (Operational Modeling) unit performs the Model Training and Inference functions.

[0134] The execution process includes: the terminal initiates an analysis subscription request to the gNB-DU; the gNB-DU forwards the terminal's analysis subscription request to the gNB-CU; and the gNB-CU reports the terminal's analysis subscription request to the OAM. The OAM selects a suitable model based on the terminal's analysis subscription request and initiates model inference. The OAM initiates a model inference data request to the gNB-CU; each network element (gNB-CU, gNB-DU, and terminal) collects model inference data based on the inference data request information, processes the data, and sends it to the OAM. The OAM uses the model inference data to perform model inference and obtains the inference result, which is then sent to the gNB-CU. The gNB-CU sends the inference result to the gNB-DU, and the gNB-DU sends the inference result to the terminal. The terminal can then use the inference result to make corresponding policy adjustments.

[0135] However, the following problems exist in the related technologies:

[0136] (1) The entire model inference process is handled by the OAM network management system, requiring all model inference data to be sent to the OAM. This approach, which uploads real-time model inference data from the wireless side to the network management system, poses a challenge to data security, especially in scenarios where the model inference data includes terminal service data, where this approach will be limited.

[0137] (2) When completing the model inference work, all model inference data needs to be uploaded to OAM, which requires real-time data transmission. Under the condition of limited wireless communication resources, this solution will increase the network load.

[0138] (3) Model inference latency includes the transmission latency caused by uploading model inference data to OAM, the computation latency of model inference, and the transmission latency caused by OAM sending inference results to the terminal. The first part of the latency is relatively large, which will cause the inference results to be not fed back in time, affecting the terminal service experience.

[0139] (4) All model inference tasks are offloaded to OAM. At the same time, OAM also needs to complete model training. When the terminal's analysis subscription requests are intensive, OAM's computing power will be insufficient, which will reduce the system's working efficiency.

[0140] Based on this, this disclosure provides a model inference method that assigns model inference tasks to different gNB-CUs (i.e., the control wireless access network devices in this embodiment). Furthermore, based on the wireless network artificial intelligence architecture, the model is segmented according to the AI ​​processing capabilities of each network element. Multiple network elements with AI processing capabilities are selected to assist the model inference network element to which the terminal belongs in jointly completing the model inference work. The inference results are then fed back to the terminal, which performs corresponding strategy adjustments based on the inference results and provides performance feedback, thereby achieving continuous model optimization.

[0141] The specific process is as follows: First, the terminal initiates a model analysis subscription request. The gNB-CU connected to the terminal generates model subscription request information based on its own AI processing capabilities and the terminal's analysis subscription request information, and reports it to the OAM. The OAM performs model selection and model segmentation, allocation and distribution of model segmentation blocks based on the model subscription request, and sends the model segmentation block allocation information to all gNB-CUs participating in joint inference. The gNB-CU connected to the terminal initiates a model inference data request. Relevant network elements collect and process the data and send it to that gNB-CU. The gNB-CU connected to the terminal uses the model inference data to complete the inference of the first model segmentation block, and sends the intermediate inference results to the gNB-CU containing the next model segmentation block according to the model segmentation block allocation information, until the gNB-CU responsible for the inference task of the last model segmentation block receives the inference result and sends the inference result to the gNB-CU connected to the terminal according to the model segmentation block allocation information. The gNB-CU connected to the terminal sends the inference result to the terminal, and the terminal uses the inference result to make corresponding policy adjustments. gNB-CU collects model performance data and terminal performance feedback data and reports them to OAM. OAM trains and optimizes the model and sends the updated model parameters to gNB-CU.

[0142] In scenarios where terminals exhibit high-speed mobility, the redelivery of AI analysis tasks via wireless networks is implemented. Terminal access location information is added to the analysis request information. When the terminal access location changes, the access location information in the analysis request information is globally maintained by re-initiating the analysis request. Furthermore, processes such as forwarding inference results and re-selecting and segmenting models are incorporated to ensure the smooth delivery of model inference tasks after terminal switching, guaranteeing the continuity and accuracy of AI analysis services. Specifically, this can be divided into the following two scenarios:

[0143] 1) When a terminal switches to another gNB-DU (i.e., the distributed radio access network device in this embodiment) under the same gNB-CU, the terminal re-initiates the analysis subscription request, and the gNB-CU and OAM update the terminal's analysis subscription request information. If the current inference task is not completed when the terminal switches, the gNB-CU continues to complete the inference task. After obtaining the inference result, it sends the inference result to the gNB-DU currently accessed by the terminal according to the access location in the updated analysis subscription request message. The gNB-DU then sends the inference result to the terminal. After the terminal switches, the newly accessed gNB-DU is responsible for completing the relevant data collection and data forwarding tasks.

[0144] 2) When a terminal switches to another gNB-CU, the terminal resends the model analysis subscription request. The newly connected gNB-CU sends the model subscription request to the OAM. The OAM updates the terminal's analysis subscription request and sends the updated analysis subscription request information to the terminal's source gNB-CU. If the current inference task is not completed when the terminal switches, the source gNB-CU completes the inference task, obtains the inference result, and sends the inference result to the newly connected gNB-CU based on the access position in the updated analysis subscription request message. The source gNB-CU updates the terminal's model subscription analysis request information and is no longer responsible for the terminal's model subscription analysis request related tasks. The newly connected gNB-CU sends the inference result to the newly connected gNB-DU, which then sends the inference result to the terminal. The OAM re-selects and segments the model based on the model subscription analysis request sent by the newly connected gNB-CU and sends the model segmentation block allocation information to the gNB-CUs participating in joint inference. This implementation method can better develop the base station's AI processing capabilities, solve the problem of insufficient base station AI processing capabilities, and is beneficial for network load balancing. It also provides a delivery method for wireless network AI analysis tasks in high-speed terminal mobility scenarios, which solves the problem of discontinuity of AI analysis services caused by terminal switching, ensures the efficiency and continuity of wireless network AI analysis services, improves terminal service experience, and also helps to improve the operating efficiency of wireless network.

[0145] It is further understood that the wireless communication system of this disclosure is a network providing wireless communication functionality. The wireless communication system can employ different communication technologies, such as code division multiple access (CDMA), wideband code division multiple access (WCDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal frequency-division multiple access (OFDMA), single-carrier frequency division multiple access (SC-FDMA), and carrier sense multiple access with collision avoidance. Based on factors such as capacity, speed, and latency, networks can be categorized as 2G networks, 3G networks, 4G networks, or future evolution networks, such as 5G networks. 5G networks can also be referred to as New Radio (NR). For ease of description, this disclosure may sometimes simply refer to the wireless communication network as a network.

[0146] Furthermore, the network device involved in this disclosure can also be referred to as a wireless access network device. This wireless access network device can be: a base station, an evolved Node B (eB) base station, a home base station, an access point (AP) in a Wireless Fidelity (WIFI) system, a wireless relay node, a wireless backhaul node, a transmission point (TP), or a transmission and reception point (TRP), etc. It can also be a gNB in ​​an NR system, or a component or part of a base station. When it is a vehicle-to-everything (V2X) communication system, the network device can also be an in-vehicle device. It should be understood that the specific technologies and device forms used in the embodiments of this disclosure are not limited.

[0147] Furthermore, the terminal involved in this disclosure can also be referred to as a terminal device, user equipment (UE), mobile station (MS), mobile terminal (MT), etc., and is a device that provides voice and / or data connectivity to a user. For example, a terminal can be a handheld device with wireless connectivity, an in-vehicle device, etc. Currently, some examples of terminals include: smartphones (Mobile Phones), pocket personal computers (PPCs), handheld computers, personal digital assistants (PDAs), laptops, tablets, wearable devices, or in-vehicle devices, etc. In addition, when it is a vehicle-to-everything (V2X) communication system, the terminal device can also be an in-vehicle device. It should be understood that the embodiments of this disclosure do not limit the specific technology or specific device form adopted by the terminal.

[0148] Figure 3 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 3 As shown, the model reasoning method used in OAM entities includes the following steps.

[0149] In step S11, in response to receiving the model subscription request information sent by the control radio access network device, a first model corresponding to the model subscription request information is determined.

[0150] In this embodiment, the model subscription request information includes information on the AI ​​processing capabilities of the controlling radio access network device itself, and information on the terminal model analysis subscription request. The AI ​​processing capability information includes the base station server's computing speed and current surplus computing power. OAM selects a model that matches the terminal model analysis subscription request information based on the model subscription information, and further determines a model of appropriate size (i.e., the first model) from the qualified models based on the AI ​​processing capabilities of the controlling radio access network device. For ease of description, this disclosure refers to the qualified and appropriately sized model as the first model.

[0151] In step S12, the first model is divided to obtain a first number of model segmentation blocks, and the first number of model segmentation blocks are distributed to a first number of control wireless access network devices.

[0152] In this embodiment, OAM divides the first model into a first number of model segmentation blocks based on the AI ​​processing capability information of the controlled radio access network devices, and determines the same number of controlled radio access network devices based on the first number. The first number of model segmentation blocks are then distributed to the first number of controlled radio access network devices. The first number of controlled radio access network devices is determined by OAM from among the controlled radio access network devices adjacent to the one sending the model subscription request information. The determination can be based on factors such as the computing power occupancy and load of the controlled radio access network devices, selecting those with relatively low workload.

[0153] The model inference method provided in this disclosure can balance computing power among multiple control wireless access network devices by using a collaborative inference method among multiple control wireless access network devices, thereby making full use of the local AI processing capabilities of the control wireless access network devices and effectively improving model inference efficiency.

[0154] In some embodiments of this disclosure, each model segmentation block in the first number of model segmentation blocks corresponds to allocation information. The allocation information includes the inference order of the first number of model segmentation blocks and the control radio access network device corresponding to each model segmentation block. The control radio access network device corresponding to each model segmentation block is included in the allocation information with a corresponding identifier.

[0155] In some embodiments of this disclosure, the first number of control radio access network devices includes a first control radio access network device, wherein the first control radio access network device is a control radio access network device accessed by a terminal.

[0156] Figure 4 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 4 As shown, the model reasoning method used in OAM entities includes the following steps.

[0157] In step S21, among the control radio access network devices adjacent to the first control radio access network device, a plurality of auxiliary control radio access network devices are identified.

[0158] In this embodiment of the disclosure, OAM selects an auxiliary control radio access network device that can assist in model inference among the control radio access network devices adjacent to the first control radio access network device.

[0159] In step S22, among the multiple auxiliary control radio access network devices, a second number of control radio access network devices are determined based on the computing power idle state of each auxiliary control radio access network device.

[0160] In this embodiment of the disclosure, OAM determines a second number of control radio access network devices that can participate in the model inference based on the computing power occupancy status and load of each control radio access network device. The second number of control radio access network devices consists of the other control radio access network devices in the first number besides the first control radio access devices.

[0161] In step S23, based on the inference order of the first number of model segmentation blocks, the first model segmentation block is sent to the first control radio access network device, and the remaining number of model segmentation blocks are distributed to the second number of control radio access network devices.

[0162] In this embodiment of the disclosure, OAM sends the first model segmentation block to the first control radio access network device (e.g., gNB-CU1), sends the remaining model segmentation blocks to other control radio access network devices participating in joint inference, and sends the allocation information corresponding to the model segmentation blocks to all control radio access network devices participating in joint inference.

[0163] Figure 5 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 5 As shown, the model reasoning method used in OAM entities includes the following steps.

[0164] In step S31, model performance update data sent by the first control wireless access network device is received.

[0165] In this embodiment of the disclosure, the first control radio access network device compares the received performance data with the first inference result of the first model to determine model performance update data, and then sends the model performance update data to the OAM. The model performance update data may be model accuracy. The OAM may also receive the performance data sent by the first control radio access device.

[0166] In step S32, the first model is updated based on the model performance update data, the updated model parameters of the first model are determined, and the updated model parameters of the first model are sent to the first control wireless access network device.

[0167] In this embodiment of the disclosure, OAM trains and optimizes the first model based on performance data and model performance update data to obtain the updated model parameters of the first model, and sends the updated model parameters of the first model to the first control wireless access network device.

[0168] In some embodiments of this disclosure, in response to OAM receiving a first model analytics subscription update request, and the first model analytics subscription update request includes information about the terminal's model analytics subscription, the terminal's model analytics request information is updated based on the first model analytics subscription update request.

[0169] In some embodiments of this disclosure, in response to OAM receiving a first model analytics subscription update request, and the first model analytics subscription update request includes information such as the terminal's model analytics subscription information and the AI ​​processing capability information of the second control radio access network device, the first model is re-segmented based on the model analytics subscription information and the AI ​​processing capability information of the second control radio access network device, the first model is sent to the second control radio access network device, and the remaining model segmentation blocks are sent to other control radio access network devices participating in inference.

[0170] Figure 6 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 6 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0171] In step S41, in response to receiving a model analysis subscription request sent by a distributed radio access network device, the model analysis subscription request is processed to obtain model subscription request information, and the model subscription request information is sent to OAM.

[0172] In this embodiment of the disclosure, the model analyzes subscription requests, including the terminal's identifier, the analysis request type, and access location information. For example, the terminal accesses a first distributed radio access network device (e.g., gNB-DU1), and gNB-DU1 and gNB-DU2 access gNB-CU1. The terminal identifier is GUTI, and the analysis request type is represented by an analysis ID, such as Analysis ID 1: Location Prediction Analysis Service, Analysis ID 2: Load Prediction Analysis Service. The access location mainly includes information about the control radio access network device and the distributed radio access network device currently accessed by the terminal.

[0173] In response to the control radio access network device receiving a model analysis subscription request sent by the distributed radio access network device, the control radio access network device generates model subscription request information based on its own AI processing capabilities and the model analysis subscription request, and sends the model subscription request information to the OAM.

[0174] In step S42, the model segmentation block sent by OAM is received.

[0175] In this embodiment of the disclosure, the model segmentation block is the model segmentation block determined by segmenting the first model. The first model is determined by OAM based on model subscription request information.

[0176] The model inference method provided in this disclosure can balance computing power among multiple control wireless access network devices by using a collaborative inference method among multiple control wireless access network devices, thereby making full use of the local AI processing capabilities of the control wireless access network devices and effectively improving model inference efficiency.

[0177] Figure 7 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 7 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0178] In step S51, a model inference data request is sent to the distributed wireless access network device.

[0179] In this embodiment of the disclosure, the model inference data request is used to obtain model inference data. The controlling radio access network device sends the model inference data request to the distributed radio access network device. It should be noted that the controlling radio access network device can send the model inference data request to the distributed radio access network device to which the terminal is connected, or it can send the model inference data request to other distributed radio access network devices participating in inference within the range of the controlling radio access network device.

[0180] In step S52, the model segmentation block is inferred based on the model inference data to obtain the inference intermediate information of the model segmentation block.

[0181] In this embodiment of the disclosure, the model inference data includes model inference data collected by the distributed radio access network device and model inference data reported by the terminal. The control radio access network device performs inference on the model segment blocks based on the received model inference data to determine the inference intermediate information for each model segment block.

[0182] In some embodiments of this disclosure, each model segment block corresponds to allocation information. A control radio access network device receives model segment blocks and allocation information. The allocation information includes the inference order of a first number of model segment blocks, and the control radio access network device corresponding to each model segment block. The control radio access network device corresponding to each model segment block is included in the allocation information with a corresponding identifier.

[0183] Figure 8 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 8 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0184] In step S61, in response to the fact that the control radio access network device is not the last control radio access network device, the inference intermediate information is sent to the next control radio access network device based on the inference order.

[0185] In this embodiment of the disclosure, in response to the fact that the control radio access network device of the current inference model is not the control radio access network device where the last model segment block is located, the current control radio access network device sends the inference intermediate information to the control radio access network device where the next model segment block is located according to the inference order of the inference model segment blocks in the allocation information.

[0186] In step S62, in response to the control radio access network device being the last control radio access network device, after the model inference is completed, the first inference result corresponding to the first model is determined, and the first inference result is sent to the first control radio access network device, which is the control radio access network device accessed by the terminal.

[0187] In this embodiment of the disclosure, the control radio access network device responding to the current inference model is the control radio access network device where the last model segment block is located. The current control radio access network device completes the inference of the first model according to the inference order of the inference model segment blocks in the allocation information, determines the first inference result corresponding to the first model, and sends the first inference result to the control radio access network device where the first model segment block is located, i.e., the first control radio access network device. The first control radio access network device is the control radio access network device accessed by the terminal.

[0188] The model inference method disclosed herein ensures that the first model segmentation block is inferred within the control radio access network device currently connected to the terminal, and the raw inference data required for terminal inference is only provided to the currently connected radio access network device. Only intermediate inference information is transmitted between other control radio access network devices participating in joint inference. This intermediate inference information has undergone feature processing, has a small data volume, and is difficult to reverse infer terminal information from. This mechanism ensures the security of sensitive wireless network data while also saving data transmission overhead.

[0189] Figure 9 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 9 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0190] In step S71, in response to the control of the radio access network device being the first control radio access network device, the first inference result is received.

[0191] In this embodiment of the disclosure, the control radio access network device corresponding to the last model segmentation block completes the inference and determines the model inference result, i.e., the first inference result. This first inference result is then sent to the first control radio access network device, which obtains the first model inference result corresponding to the first model.

[0192] In step S72, the first inference result is sent to the first distributed wireless access network device.

[0193] In this embodiment of the disclosure, the first control radio access network device determines a first inference result of the first model based on the received inference result. The first inference result is then sent to the distributed radio access network device to which the terminal is connected.

[0194] Figure 10 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 10 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0195] In step S81, performance data sent by the first distributed wireless access network device is received.

[0196] In this embodiment, the performance data is the actual performance data after the terminal adjusts its execution strategy based on the first model. After adjusting the execution strategy based on the first model, the terminal reports the obtained actual performance data to the accessed distributed radio access network device. The distributed radio access network then sends the data to the control radio access network device. For example, the distributed radio access network device accessed by the terminal is gNB-DU1, and the control radio access network device corresponding to gNB-DU1 is gNB-CU1. After determining the actual performance data, the terminal sends it to gNB-DU1, and gNB-DU1 reports it to gNB-CU1.

[0197] Among them, performance data can be the quantification of performance improvement brought about by AI analysis services, such as the terminal subscribing to a certain analysis and implementing corresponding strategy adjustments based on the analysis results to achieve a 5% power saving.

[0198] In step S82, the performance data is processed to obtain model performance update data, and the model performance update data is sent to OAM.

[0199] In this embodiment, the performance data includes model performance data and performance feedback data from the terminal. The first control radio access network device processes the model performance data and performance feedback data from the terminal to obtain model performance update data, and then sends the model performance update data to the OAM.

[0200] Figure 11 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 11 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0201] In step S91, in response to the control radio access network device being the first control radio access network device, a model subscription request message is sent to OAM.

[0202] In this embodiment of the disclosure, the first control radio access network device is the control radio access network device corresponding to the first distributed wireless network device to which the terminal accesses. The first control radio access network device receives a model analysis subscription request sent by the distributed radio access network device, generates model subscription request information based on its own AI capabilities and the model analysis subscription request, and sends the model subscription request information to the OAM.

[0203] In some embodiments of this disclosure, due to the mobility of the terminal, a handover to the distributed radio access network device may occur. In one embodiment, the terminal switches to the distributed radio access network device but does not switch to the control radio access network device.

[0204] Figure 12 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 12 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0205] In step S101, in response to the first controlling radio access network device being the controlling radio access network device, if a model analysis subscription request is received again, a second distributed radio access network device is determined to resend the model analysis subscription request.

[0206] In this embodiment of the disclosure, the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching to the distributed radio access network device.

[0207] In response to the first control radio access network device receiving the model analysis subscription request again, it determines that the distributed radio access network device accessed by the terminal has switched, updates the terminal's analysis subscription information, and reports the analysis subscription information to OAM. In one embodiment, if the first control radio access network device has not completed the inference task of the first model, the first control radio access network device (e.g., gNB-CU1) completes the current model inference task, obtains the first inference result, and then sends the first inference result to the second distributed radio access network device (e.g., gNB-DU2) according to the access location in the update analysis request message, which then forwards it to the terminal.

[0208] In step S102, the first inference result is sent to the second distributed wireless access network device, and a model subscription update request is sent to OAM.

[0209] In this embodiment of the disclosure, the first control radio access network device sends the first inference result to the second distributed radio access network device and sends a model subscription update request to the OAM, requesting the OAM to update the terminal's analysis request information.

[0210] For example, taking gNB-CU1 as the first control radio access network device and gNB-DU2 as the second distributed radio access network device, gNB-CU1 updates the terminal's analysis request information. gNB-CU1 reports the terminal analysis request message to OAM, and OAM updates the terminal's analysis request information. In response to the current inference task not being completed during terminal handover, after gNB-CU1 completes the current model inference task and obtains the first inference result, it sends the first inference result to gNB-DU2 according to the access location in the updated analysis request message, and gNB-DU2 forwards it to the terminal.

[0211] In one embodiment, the terminal switches the accessed distributed radio access network device and controls the switching of the radio access network device. This embodiment includes:

[0212] Figure 13 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 13 As shown, the model inference method is used to control wireless access network devices and includes the following steps.

[0213] In step S111, in response to the control radio access network device being the second control radio access network device, if the model analysis subscription request is received again, the second distributed radio access network device that resends the model analysis subscription request and the second control radio access network device that receives the model analysis subscription request again are determined.

[0214] In this embodiment, the second control radio access network device is the control radio access network device corresponding to the second distributed radio access network device. The second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching to the distributed radio access network device.

[0215] In response to the control radio access network device receiving a model analysis subscription request again from the distributed radio access network device, and the control radio access network device receiving the request again being the second control radio access network device, the terminal determines the second distributed radio access network device to which it has reconnected.

[0216] In step S112, the first inference result is sent to the second control wireless access network device, and a model subscription update request is sent to OAM.

[0217] In this embodiment of the disclosure, the first control radio access network device sends the first inference result to the second control radio access network device and is no longer responsible for the analysis request of the terminal. It also sends a model subscription update request to the OAM.

[0218] For example, taking gNB-CU2 as the second control radio access network device and gNB-DU3 as the second distributed radio access network device, gNB-CU2 sends a model subscription request to OAM, including its own AI processing capability information and terminal analysis subscription request information. OAM updates the current terminal's analysis subscription request information according to the model subscription request and sends the updated analysis subscription request information to the source base station gNB-CU1. In response to the current inference task not being completed during terminal handover, gNB-CU1 completes the inference task, obtains the first inference result, and sends the first inference result to gNB-CU2 according to the access location in the updated analysis request message. gNB-CU1 updates the terminal analysis request information and is no longer responsible for the analysis request-related tasks of that terminal. gNB-CU2 sends the first inference result to gNB-DU3, and gNB-DU3 forwards it to the terminal. OAM re-segments the model according to the AI ​​processing capability information in gNB-CU2's model subscription request, sends the first block to the requesting gNB-CU2, sends the remaining blocks to other gNB-CUs, and sends the model segmentation block allocation information to the gNB-CUs participating in joint inference.

[0219] The model inference method provided in this disclosure ensures that the current first inference result of the source distributed radio access network device is successfully fed back to the terminal that has been switched over. The newly connected distributed radio access network device quickly takes over the model inference task, thereby avoiding the interruption of the AI ​​analysis service required by the terminal during the switching process and ensuring the continuity and accuracy of the mobile terminal AI analysis service.

[0220] Figure 14 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 14 As shown, the model inference method is used in distributed wireless access network devices and includes the following steps.

[0221] In step S121, in response to receiving the model analysis subscription request sent by the terminal, a model analysis subscription request is sent to the control radio access network device.

[0222] In this embodiment of the disclosure, a model analysis subscription request is used to obtain a first model from OAM. The first model includes a first number of model segmentation blocks.

[0223] The terminal initiates a model analysis subscription request to the access distributed radio access network (DRN) device. This request includes the terminal identifier, analysis request type, and access location information. Upon receiving the model analysis subscription request, the DRN device forwards it to the control DRN device.

[0224] Figure 15 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 15 As shown, the model inference method is used in distributed wireless access network devices and includes the following steps.

[0225] In step S131, a model inference data request sent by the control wireless access network device is received.

[0226] In step S132, model inference data is obtained from the terminal and sent to the control wireless access network device.

[0227] In this embodiment of the disclosure, a model inference data request is used to obtain model inference data. After receiving the model inference data request, the distributed radio access network device obtains the model inference data from the terminal and sends the model inference data to the control radio access network device.

[0228] Figure 16 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 16 As shown, the model inference method is used in distributed wireless access network devices and includes the following steps.

[0229] In step S141, in response to the distributed radio access network device being the first distributed radio access network device, the first inference result sent by the first control radio access network device is received.

[0230] In step S142, the first inference result is sent to the terminal.

[0231] In this embodiment of the disclosure, the first distributed radio access network device to which the terminal is connected receives a first inference result sent by the first control radio access network device and sends the first inference result to the terminal so that the terminal adjusts its execution strategy.

[0232] Figure 17 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 17 As shown, the model inference method is used in distributed wireless access network devices and includes the following steps.

[0233] In step S151, in response to the distributed radio access network device being the first distributed radio access network device, performance data sent by the terminal is received.

[0234] In this embodiment of the disclosure, the performance data is the actual performance data of the terminal after adjusting the execution strategy based on the first model;

[0235] In step S152, performance data is sent to the first control wireless access network device.

[0236] In this embodiment of the disclosure, the terminal sends actual performance data (also known as performance feedback data) obtained after adjusting the execution strategy to the first distributed radio access network device. This performance data is then sent to the first control radio access network device corresponding to the first distributed radio access network device.

[0237] Figure 18 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 18 As shown, the model inference method is used in distributed wireless access network devices and includes the following steps.

[0238] In step S161, in response to the fact that the radio access network device is the second distributed radio access network device, if a model analysis subscription request is received from the terminal, it is determined to send the model analysis subscription request to the control radio access network device corresponding to the second distributed radio access network device.

[0239] In this embodiment of the disclosure, the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching to the distributed radio access network device.

[0240] After the terminal switches to the new distributed radio access network (DRB) device, it resends the model analysis subscription request to that DRB. If the receiving DRB is the second DRB, i.e., the DRB the terminal has reconnected to, it then sends the model analysis subscription request to its corresponding control DRB.

[0241] In some embodiments of this disclosure, the interaction process between the control radio access network device and the distributed radio access network device is further illustrated using OAM as an example. The control radio access network device can be a gNB-CU, and the distributed radio access network device can be a gNB-DU. Figure 19 This is a flowchart illustrating a model reasoning method according to an exemplary embodiment. For example... Figure 19 As shown, it includes the following steps:

[0242] In step 1, the terminal initiates a model analysis subscription request.

[0243] In step 2, gNB-CU sends a model subscription request to OAM.

[0244] In step 3, OAM selects and segments models based on the model subscription request information, allocates and distributes model segmentation blocks, and sends the model segmentation block allocation information to the gNB-CU participating in joint inference.

[0245] In step 4, the gNB-CU accessed by the terminal initiates a model inference data request, the relevant network elements collect and process the data, and send the data to the gNB-CU.

[0246] In step 5, the gNB-CU accessed by the terminal uses model inference data to complete the inference of the first model segmentation block, and sends the intermediate inference results to the gNB-CU where the next model segmentation block is located according to the model segmentation block allocation information.

[0247] In step 6, after obtaining the first inference result, the gNB-CU responsible for the last model segmentation block inference task sends the first inference result to the gNB-CU accessed by the terminal according to the model segmentation block allocation information.

[0248] In step 7, the gNB-CU accessed by the terminal sends the first inference result to the terminal, and the terminal uses the first inference result to make corresponding policy adjustments.

[0249] In step 8, gNB-CU collects model performance data and terminal performance feedback data and reports them to OAM. OAM trains and optimizes the model and sends the updated model to gNB-CU.

[0250] In some embodiments of this disclosure, due to the mobility of the terminal, a handover to the distributed radio access network device may occur. In one embodiment, the terminal switches to the distributed radio access network device but does not switch to the control radio access network device. Figure 20 This is a flowchart illustrating terminal switching in a model inference method according to an exemplary embodiment. For example... Figure 20 As shown, it includes the following steps:

[0251] In step 1, the terminal re-initiates the analysis subscription request.

[0252] In step 2, gNB-CU and OAM update the model analysis subscription request information of the terminal.

[0253] In step 3, if the current inference task is not completed when the terminal switches, the gNB-CU completes the inference task and sends the first inference result to the gNB-DU currently accessed by the terminal, and the gNB-DU sends the first inference result to the terminal.

[0254] In step 4, after the terminal handover is completed, the newly connected gNB-DU is responsible for completing the relevant data collection and data forwarding tasks.

[0255] In some embodiments of this disclosure, due to the mobility of the terminal, a handover to a distributed radio access network device may occur. In one embodiment, the terminal switches to a distributed radio access network device, and the handover control radio access network device is also involved. Figure 21This is a flowchart illustrating terminal switching in a model inference method according to an exemplary embodiment. For example... Figure 21 As shown, it includes the following steps:

[0256] In step 1, the terminal re-initiates the model analysis subscription request.

[0257] In step 2, the newly connected gNB-CU of the terminal sends a model subscription request to OAM.

[0258] In step 3, OAM updates the terminal's analytics subscription request and sends the updated model analytics subscription request information to the terminal's source gNB-CU.

[0259] In step 4, if the current inference task is not completed when the terminal switches, the source gNB-CU completes the inference task and sends the first inference result to the gNB-CU that the terminal has newly connected to.

[0260] In step 5, the source gNB-CU updates the terminal analysis request information and is no longer responsible for the analysis request related tasks of that terminal.

[0261] In step 6, the newly connected gNB-CU sends the first inference result to the newly connected gNB-DU, which in turn sends the first inference result to the terminal.

[0262] In step 7, OAM re-selects and segments the model based on the model subscription request sent by the newly connected gNB-CU, and sends the model segmentation block allocation information to the gNB-CU participating in joint inference.

[0263] Figure 22 This is a schematic diagram illustrating the protocol and interface of a model reasoning method according to an exemplary embodiment. For example... Figure 22 As shown, this mainly involves the terminal provided in the embodiments of the present invention, the gNB-DU accessed by the terminal, the gNB-CU accessed by the terminal, other gNB-CUs (gNB-CU(1)~gNB-CU(N)) participating in joint inference, and OAM. Specifically, as follows:

[0264] 1a. The terminal sends a model analysis subscription request signaling to the gNB-DU, instructing it to initiate a model analysis subscription request to the receiver. 1b. The gNB-DU sends the analysis subscription request signaling to the gNB-CU. 2. The gNB-CU generates model subscription request information based on its AI processing capabilities and the analysis subscription request information. 3. The gNB-CU sends the model subscription request signaling to the OAM, instructing it to initiate a model subscription request to the receiver. 4. The OAM selects the first model that matches the analysis request based on the model subscription information and divides the model into several blocks according to the AI ​​processing capability information. 5a. The OAM sends the first model block and its allocation information to the gNB-CU connected to the terminal. 5b. The OAM sends the remaining model blocks and their allocation information to the other gNB-CUs participating in joint inference, instructing them to send the model block allocation information. 6. The gNB-CU sends a model inference data collection request signaling to the gNB-DU, instructing it to initiate a model inference data collection request to the receiver. 7. gNB-DU, terminal, and gNB-CU collect data according to the model inference data collection request and send it to gNB-CU. 8. gNB-CU uses the collected inference data to perform partial model inference on the first model segment block to obtain inference intermediate information. 9. gNB-CU sends the inference intermediate information to gNB-CU(1) where the next model segment block is located. 10. NB-CU(1) performs partial model inference on the model segment block to obtain the inference intermediate information result. 11. gNB-CU(1) sends the inference intermediate information to gNB-CU where the next model segment block is located until gNB-CU(N) where the last model segment block is located receives all the inference intermediate information. 12. gNB-CU(N) performs partial model inference on the last model segment block to obtain the first inference result. 13a. gNB-CU(N) sends the first inference result to the gNB-CU accessed by the terminal. 13b. gNB-CU sends the first inference result to gNB-DU accessed by the terminal. 13c. gNB-DU sends the first inference result to the terminal. 14. The terminal adjusts its strategy accordingly based on the first inference result. 15a. The terminal sends performance feedback data to gNB-DU. 15b. gNB-DU sends performance feedback data to gNB-CU. 16. gNB-CU compares the first inference result with the real data to obtain model performance data. 17. gNB-CU processes the model performance data and the terminal performance feedback data. 18. gNB-CU sends the model performance data and the terminal performance feedback data to OAM. 19. OAM trains and optimizes the model based on the model performance data and performance feedback data. 20. OAM sends the updated model parameters to gNB-CU.

[0265] Figure 23This is a schematic diagram illustrating the protocol and interface principles for AI analysis task delivery when a terminal switches between the same gNB-CU, according to an exemplary embodiment of a model-free inference method. Figure 23 As shown, this mainly involves the terminal provided in this embodiment, the terminal's source gNB-DU (gNB-DU 1), the newly accessed gNB-DU (gNB-DU 2), the gNB-CU accessed by the terminal, and OAM. Specifically:

[0266] 1a. The terminal sends the model analysis subscription request signaling to gNB-DU 2. 1b. gNB-DU 2 sends the model analysis subscription request signaling to gNB-CU. 2. gNB-CU updates the terminal's analysis request information. 3. gNB-CU sends the analysis subscription update request to OAM. 4. OAM updates the terminal's analysis request information. 5. If the current inference task is not completed during terminal handover, gNB-CU continues to complete the current inference task and obtains the first inference result. 6a. gNB-CU sends the first inference result to gNB-DU 2 according to the access location in the analysis subscription update request. 6b. gNB-DU 2 sends the first inference result to the terminal. 7. After the terminal handover is completed, gNB-DU 2 is responsible for the data collection and forwarding tasks related to the terminal's analysis request.

[0267] Figure 24 This is a schematic diagram illustrating the protocol and interface principles for AI analysis task delivery during terminal handover across gNB-CUs in a model-free inference method, as shown in an exemplary embodiment. Figure 24 As shown, this mainly involves the terminal provided in this embodiment, the terminal's source gNB-DU (gNB-DU 1), the newly accessed gNB-DU (gNB-DU 3), the terminal's source gNB-CU (gNB-CU 1), the newly accessed gNB-CU (gNB-CU 2), other gNB-CUs (gNB-CU(1)~gNB-CU(N)) participating in joint inference, and OAM. Specifically:

[0268] 1a. The terminal sends the model analysis subscription request signaling to gNB-DU 3. 1b. gNB-DU 3 sends the model analysis subscription request signaling to gNB-CU 2. 2. gNB-CU 2 generates model subscription request information based on its AI processing capabilities and the analysis subscription request information. 3. gNB-CU 2 sends the model subscription request signaling to OAM. 4. OAM updates the current terminal's analysis subscription request information based on the model subscription request. 5. OAM sends an analysis subscription update request to gNB-CU 1, instructing it to initiate an analysis subscription update request to the recipient. 6. If the current inference task is not completed when the terminal switches, gNB-CU 1 continues to complete the current inference task and obtains the first inference result. 7. gNB-CU 1 updates the terminal's analysis request information and is no longer responsible for the analysis request-related tasks of that terminal. 8a. gNB-CU 1 sends the first inference result to gNB-CU 2 based on the access location in the analysis subscription update request. 8b. gNB-CU 2 sends the first inference result to gNB-DU 3. 8c. gNB-DU 3 sends the first inference result to the terminal. 9. OAM re-selects and segments the model based on the AI ​​processing capability information in the model subscription request. 10a. OAM sends the first model segment and its allocation information to gNB-CU 2. 10b. OAM sends the remaining model segments and their allocation information to the gNB-CU for auxiliary inference.

[0269] Based on the same concept, embodiments of this disclosure also provide a model reasoning apparatus.

[0270] It is understood that the model inference apparatus provided in this disclosure includes hardware structures and / or software modules corresponding to each function in order to achieve the above-mentioned functions. In conjunction with the units and algorithm steps of the various examples disclosed in this disclosure, this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the technical solutions of this disclosure.

[0271] Figure 25 This is a block diagram illustrating a model inference apparatus according to an exemplary embodiment. (Refer to...) Figure 25 The model inference device 100 is applied to the Operation and Maintenance Management (OAM) entity and includes a determination module 101 and a transmission module 102.

[0272] The determining module 101 is configured to, in response to receiving a model subscription request information sent by a control radio access network device, determine a first model corresponding to the model subscription request information. The sending module 102 is configured to divide the first model into a first number of model segment blocks, and distribute the first number of model segment blocks to a first number of control radio access network devices.

[0273] In this embodiment of the disclosure, each model segmentation block in the first number of model segmentation blocks has corresponding allocation information.

[0274] The allocation information includes the inference order of a first number of model segment blocks, and the control radio access network device corresponding to each model segment block.

[0275] In this embodiment of the disclosure, the first number of control wireless access network devices includes a first control wireless access network device, which is a control wireless access network device accessed by the terminal.

[0276] The sending module 102 is configured to: identify multiple auxiliary control wireless access network devices among the control wireless access network devices adjacent to the first control wireless access network device; determine a second number of control wireless access network devices among the multiple auxiliary control wireless access network devices based on the computing power occupancy status and load of each auxiliary control wireless access network device; the second number of control wireless access network devices are the other control wireless access network devices in the first number besides the first control wireless access network device; and send the first model segmentation block to the first control wireless access network device based on the inference order of the first number of model segmentation blocks, and distribute the remaining number of model segmentation blocks to the second number of control wireless access network devices.

[0277] In this embodiment of the disclosure, the model inference device further includes a receiving module 103.

[0278] The receiving module 103 shown is used to receive model performance update data sent by the first control radio access network device. Based on the model performance update data, it updates the first model, determines the updated model parameters, and sends the updated model parameters to the first control radio access network device.

[0279] In this embodiment of the disclosure, the receiving module 103 is further configured to update the distributed radio access network (DRN) device accessed by the terminal in response to receiving a first model analysis subscription update request. The first model analysis subscription update request instructs the terminal to switch the DRF device without switching the control DRF device. Alternatively, in response to receiving a second model analysis subscription update request, the receiving module 103 updates the DRF device accessed by the terminal and re-segments the first model. The second model analysis subscription update request instructs the terminal to switch the DRF device and switch the control DRF device.

[0280] Figure 26 This is a block diagram illustrating a model inference apparatus according to an exemplary embodiment. (Refer to...) Figure 26 The model inference device 200 is used to control wireless access network equipment and includes a transmitting module 201 and a receiving module 202.

[0281] The sending module 201 is configured to, in response to receiving a model analysis subscription request sent by a distributed radio access network device, process the model analysis subscription request to obtain model subscription request information, and send the model subscription request information to the OAM. The receiving module 202 is configured to receive model segmentation blocks sent by the OAM. The model segmentation blocks are model segmentation blocks determined by segmenting the first model. The first model is determined by the OAM based on the model subscription request information.

[0282] In this embodiment of the disclosure, the sending module 201 is further configured to send a model inference data request to the distributed radio access network device, the model inference data request being used to obtain model inference data. Based on the model inference data, inference is performed on the model segmentation blocks to obtain intermediate inference information for the model segmentation blocks.

[0283] In this embodiment of the disclosure, each model segmentation block corresponds to allocation information. The allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block.

[0284] The sending module 201 is further configured to, in response to the fact that the controlled radio access network device is not the last controlled radio access network device, send intermediate inference information to the next controlled radio access network device based on the inference order. In response to the fact that the controlled radio access network device is the last controlled radio access network device, after the model inference is completed, determine the first inference result corresponding to the first model, and send the first inference result to the first controlled radio access network device, where the first controlled radio access network device is the controlled radio access network device accessed by the terminal.

[0285] In this embodiment of the disclosure, the sending module 201 is further configured to receive the first inference result in response to the control radio access network device being the first control radio access network device. The first inference result is then sent to the first distributed radio access network device, which is the distributed radio access network device to which the terminal is accessed.

[0286] In this embodiment of the disclosure, after sending the first inference result to the first distributed radio access network device, the receiving module 202 is further configured to receive performance data sent by the first distributed radio access network device. The performance data is the actual performance data of the terminal after adjusting the execution strategy based on the first model. The performance data is processed to obtain model performance update data, and the model performance update data is sent to OAM.

[0287] In this embodiment of the disclosure, the sending module 201 is further configured to send a model subscription request information to OAM in response to the control radio access network device being a first control radio access network device. The first control radio access network device is the control radio access network device corresponding to the first distributed wireless network device accessed by the terminal.

[0288] In this embodiment of the disclosure, the sending module 201 is further configured to, in response to the control radio access network device being the first control radio access network device, determine a second distributed radio access network device to resend the model analysis subscription request if a model analysis subscription request is received again. The second distributed radio access network device is the distributed radio access network device that the terminal re-accesses after switching distributed radio access network devices. The first inference result is sent to the second distributed radio access network device, and a model subscription update request is sent to the OAM.

[0289] In this embodiment, the sending module 201 is further configured to, in response to the control radio access network device being the second control radio access network device, if a model analysis subscription request is received again, determine the second distributed radio access network device that resent the model analysis subscription request, and the second control radio access network device that received the model analysis subscription request again, wherein the second control radio access network device is the control radio access network device corresponding to the second distributed radio access network device. The second distributed radio access network device is the distributed radio access network device that the terminal re-accesses after switching distributed radio access network devices. The first inference result is sent to the second control radio access network device, and a model subscription update request is sent to OAM.

[0290] Figure 27 This is a block diagram illustrating a model inference apparatus according to an exemplary embodiment. (Refer to...) Figure 27 The model inference device 300 is applied to a distributed wireless access network device, including a transmission module 301.

[0291] The sending module 301 is configured to send a model analysis subscription request to the control radio access network device in response to receiving a model analysis subscription request from the terminal. The model analysis subscription request is used to obtain a first model from the OAM (Optical Modeling and Access Network). The first model includes a first number of model segmentation blocks.

[0292] In this embodiment of the disclosure, the device further includes a receiving module 302.

[0293] The receiving module 302 is used to receive a model inference data request sent by the control radio access network device. The model inference data request is used to obtain model inference data. It obtains the model inference data from the terminal and sends it to the control radio access network device.

[0294] In this embodiment of the disclosure, the receiving module 302 is further configured to, in response to the distributed radio access network device being a first distributed radio access network device, receive a first inference result sent by the first control radio access network device, and send the first inference result to the terminal.

[0295] In this embodiment, the receiving module 302 is further configured to receive performance data sent by the terminal in response to the distributed radio access network device being a first distributed radio access network device. The performance data is the actual performance data after the terminal adjusts the execution strategy based on the first model. The performance data is then sent to the first control radio access network device.

[0296] In this embodiment of the disclosure, the receiving module 302 is further configured to, in response to the radio access network device being a second distributed radio access network device, determine to send a model analysis subscription request to the control radio access network device corresponding to the second distributed radio access network device if it receives a model analysis subscription request resent by the terminal. The second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching distributed radio access network devices.

[0297] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0298] Figure 28 This is a block diagram illustrating an apparatus 400 for model reasoning according to an exemplary embodiment. For example, apparatus 400 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0299] Reference Figure 28 The device 400 may include one or more of the following components: processing component 402, memory 404, power component 406, multimedia component 408, audio component 410, input / output (I / O) interface 412, sensor component 414, and communication component 416.

[0300] Processing component 402 typically controls the overall operation of device 400, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 402 may include one or more processors 420 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 402 may include one or more modules to facilitate interaction between processing component 402 and other components. For example, processing component 402 may include a multimedia module to facilitate interaction between multimedia component 408 and processing component 402.

[0301] Memory 404 is configured to store various types of data to support the operation of device 400. Examples of such data include instructions for any application or method operating on device 400, contact data, phonebook data, messages, pictures, videos, etc. Memory 404 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0302] The power supply component 406 provides power to the various components of the device 400. The power supply component 406 may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device 400.

[0303] Multimedia component 408 includes a screen that provides an output interface between the device 400 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 408 includes a front-facing camera and / or a rear-facing camera. When the device 400 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0304] Audio component 410 is configured to output and / or input audio signals. For example, audio component 410 includes a microphone (MIC) configured to receive external audio signals when device 400 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 404 or transmitted via communication component 416. In some embodiments, audio component 410 also includes a speaker for outputting audio signals.

[0305] I / O interface 412 provides an interface between processing component 402 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0306] Sensor assembly 414 includes one or more sensors for providing status assessments of various aspects of device 400. For example, sensor assembly 414 may detect the on / off state of device 400, the relative positioning of components such as the display and keypad of device 400, changes in the position of device 400 or a component of device 400, the presence or absence of user contact with device 400, the orientation or acceleration / deceleration of device 400, and temperature changes of device 400. Sensor assembly 414 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 414 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 414 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0307] Communication component 416 is configured to facilitate wired or wireless communication between device 400 and other devices. Device 400 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 416 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 416 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0308] In an exemplary embodiment, the apparatus 400 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0309] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 404 including instructions, which can be executed by a processor 420 of the device 400 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0310] Figure 29 This is a block diagram illustrating an apparatus 500 for model inference according to an exemplary embodiment. For example, apparatus 500 may be provided as a server. (Refer to...) Figure 29The apparatus 500 includes a processing component 522, which further includes one or more processors, and memory resources represented by memory 532 for storing instructions, such as application programs, that can be executed by the processing component 522. The application programs stored in memory 532 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 522 is configured to execute instructions to perform the methods described above.

[0311] Device 500 may also include a power supply component 526 configured to perform power management of device 500, a wired or wireless network interface 550 configured to connect device 500 to a network, and an input / output (I / O) interface 558. Device 500 may operate on an operating system stored in memory 532, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0312] It can be further understood that in this disclosure, "multiple" refers to two or more, and other quantifiers are similar. "And / or" describes the relationship between related objects, indicating that three relationships can exist; for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. The singular forms "a," "the," and "the" are also intended to include the plural forms unless the context clearly indicates otherwise.

[0313] It is further understood that the terms "first," "second," etc., are used to describe various types of information, but this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another, and do not indicate a specific order or degree of importance. In fact, the expressions "first," "second," etc., are completely interchangeable. For example, without departing from the scope of this disclosure, first information can also be referred to as second information, and similarly, second information can also be referred to as first information.

[0314] It is further understood that although operations are described in a specific order in the accompanying drawings in the embodiments of this disclosure, this should not be construed as requiring these operations to be performed in the specific order or serial order shown, or requiring all of the shown operations to be performed to obtain the desired result. In certain environments, multitasking and parallel processing may be advantageous.

[0315] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0316] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A model reasoning method, characterized in that, The method, applied to an Operations, Maintenance, and Administration (OAM) entity, includes: In response to receiving a model subscription request information sent by a control radio access network device, a first model corresponding to the model subscription request information is determined; The first model is divided into a first number of model segmentation blocks, and the first number of model segmentation blocks are distributed to a first number of control wireless access network devices. In this context, each of the first number of model segmentation blocks has corresponding allocation information; The allocation information includes the inference order of a first number of model segment blocks, and the control radio access network device corresponding to each model segment block.

2. The model reasoning method according to claim 1, characterized in that; The first number of control radio access network devices includes a first control radio access network device, which is a control radio access network device accessed by a terminal. The step of distributing the first number of model segmentation blocks to the first number of control radio access network devices includes: Among the control radio access network devices adjacent to the first control radio access network device, a plurality of auxiliary control radio access network devices are identified. Among the plurality of auxiliary control wireless access network devices, a second number of control wireless access network devices is determined based on the computing power occupancy status and load of each auxiliary control wireless access network device; the second number of control wireless access network devices are the other control wireless access network devices in the first number besides the first control wireless access devices. Based on the inference order of the first number of model segmentation blocks, the first model segmentation block is sent to the first control radio access network device, and the remaining number of model segmentation blocks are distributed to the second number of control radio access network devices.

3. The model reasoning method according to claim 2, characterized in that, The model inference method also includes: Receive model performance update data sent by the first control wireless access network device; The first model is updated based on the model performance update data, the updated model parameters of the first model are determined, and the updated model parameters of the first model are sent to the first control wireless access network device.

4. The model reasoning method according to claim 2, characterized in that, The model inference method also includes: In response to receiving a first model analysis subscription update request, the distributed radio access network device accessed by the terminal is updated; wherein, the first model analysis subscription update request instructs the terminal to switch the distributed radio access network device, but does not switch the control radio access network device; or In response to receiving a second model analysis subscription update request, the distributed radio access network device accessed by the terminal is updated, and the first model is re-segmented; wherein, the second model analysis subscription update request instructs the terminal to switch the distributed radio access network device and switch the control radio access network device.

5. A model reasoning method, characterized in that, The method, applied to controlling wireless access network devices, includes: In response to receiving a model analysis subscription request sent by a distributed radio access network device, the model analysis subscription request is processed to obtain model subscription request information, and the model subscription request information is sent to OAM; Receive model segmentation blocks sent by OAM; the model segmentation blocks are model segmentation blocks determined by segmenting the first model; the first model is determined by OAM based on the model subscription request information; The model segmentation blocks are associated with allocation information; the allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block.

6. The model reasoning method according to claim 5, characterized in that, After sending the model subscription request information to OAM, the method further includes: Send a model inference data request to the distributed wireless access network device, wherein the model inference data request is used to obtain model inference data; Based on the model inference data, inference is performed on the model segmentation block to obtain the inference intermediate information of the model segmentation block.

7. The model reasoning method according to claim 6, characterized in that, The model inference method also includes: In response to the fact that the control radio access network device is not the last control radio access network device, the inference intermediate information is sent to the next control radio access network device based on the inference order. In response to the fact that the control radio access network device is the last control radio access network device, after the model inference is completed, the first inference result corresponding to the first model is determined, and the first inference result is sent to the first control radio access network device, where the first control radio access network device is the control radio access network device accessed by the terminal.

8. The model reasoning method according to claim 7, characterized in that, The method further includes: In response that the control radio access network device is a first control radio access network device, the first inference result is received; The first inference result is sent to a first distributed wireless access network device, which is the distributed wireless access network device accessed by the terminal.

9. The model reasoning method according to claim 8, characterized in that, After sending the first inference result to the first distributed wireless access network device, the model inference method further includes: Receive performance data sent by the first distributed wireless access network device, wherein the performance data is the actual performance data of the terminal after adjusting the execution strategy based on the first model; The performance data is processed to obtain model performance update data, and the model performance update data is sent to OAM.

10. The model reasoning method according to claim 5, characterized in that, Send a model subscription request message to OAM, including: In response that the control radio access network device is the first control radio access network device, a model subscription request information is sent to OAM; Wherein, the first control wireless access network device is the control wireless access network device corresponding to the first distributed wireless network device to which the terminal accesses.

11. The model reasoning method according to claim 10, characterized in that, The model inference method also includes: In response to the fact that the control radio access network device is a first control radio access network device, if a model analysis subscription request is received again, a second distributed radio access network device is determined to resend the model analysis subscription request; the second distributed radio access network device is the distributed radio access network device that the terminal reconnects to after switching to the distributed radio access network device. The first inference result is sent to the second distributed wireless access network device, and a model subscription update request is sent to OAM.

12. The model reasoning method according to claim 11, characterized in that, The model inference method also includes: In response to the fact that the control radio access network device is the second control radio access network device, if a model analysis subscription request is received again, it is determined that the second distributed radio access network device retransmits the model analysis subscription request, and the second control radio access network device that receives the model analysis subscription request again is also determined. The second control radio access network device is the control radio access network device corresponding to the second distributed radio access network device; the second distributed radio access network device is the distributed radio access network device that the terminal re-accesses after switching to the distributed radio access network device. The first inference result is sent to the second control wireless access network device, and a model subscription update request is sent to OAM.

13. A model reasoning method, characterized in that, Applied to distributed wireless access network devices, the method includes: In response to receiving a model analysis subscription request from a terminal, the model analysis subscription request is sent to the control radio access network device; The model analysis subscription request is used to obtain a first model from OAM; the first model includes a first number of model segments, each of the first number of model segments has corresponding allocation information, the allocation information includes the inference order of the first number of model segments, and the control radio access network device corresponding to each model segment.

14. The model reasoning method according to claim 13, characterized in that, The method further includes: Receive a model inference data request sent by a wireless access network device, the model inference data request being used to obtain model inference data; Obtain model inference data from the terminal and send it to the control wireless access network device.

15. The model reasoning method according to claim 13, characterized in that, The method further includes: In response to the distributed radio access network device being a first distributed radio access network device, the device receives a first inference result sent by the first control radio access network device. The first inference result is sent to the terminal.

16. The model reasoning method according to claim 15, characterized in that, After sending the first inference result to the terminal, the method further includes: In response to the distributed wireless access network device being a first distributed wireless access network device, the device receives performance data sent by the terminal; the performance data is the actual performance data of the terminal after adjusting the execution strategy based on the first model. The performance data is sent to the first control wireless access network device.

17. The model reasoning method according to claim 13, characterized in that, The method further includes: In response to the fact that the wireless access network device is a second distributed wireless access network device, if a model analysis subscription request is received resent by the terminal, it is determined to send a model analysis subscription request to the control wireless access network device corresponding to the second distributed wireless access network device. The second distributed wireless access network device is the distributed wireless access network device that the terminal reconnects to after switching to the distributed wireless access network device.

18. A model reasoning device, characterized in that, The device is used in Operations, Maintenance and Management (OAM) entities and includes: The determination module is configured to, in response to receiving model subscription request information sent by the control radio access network device, determine a first model corresponding to the model subscription request information; The sending module is configured to segment the first model to obtain a first number of model segment blocks, and distribute the first number of model segment blocks to a first number of control radio access network devices. Each model segment block in the first number of model segment blocks has corresponding allocation information. The allocation information includes the inference order of the first number of model segment blocks and the control radio access network device corresponding to each model segment block.

19. A model reasoning device, characterized in that, The apparatus is used to control wireless access network devices, and includes: The sending module is configured to respond to receiving a model analysis subscription request sent by a distributed radio access network device, process the model analysis subscription request to obtain model subscription request information, and send the model subscription request information to the OAM; A receiving module is used to receive model segmentation blocks sent by OAM; the model segmentation blocks are model segmentation blocks determined by segmenting a first model; the first model is determined by OAM based on the model subscription request information; wherein, the model segmentation blocks correspond to allocation information; the allocation information includes the inference order of a first number of model segmentation blocks, and the control radio access network device corresponding to each model segmentation block.

20. A model reasoning device, characterized in that, Applied to distributed wireless access network equipment, the device includes: The sending module is configured to send the model analysis subscription request to the control radio access network device in response to receiving the model analysis subscription request sent by the terminal; The model analysis subscription request is used to obtain a first model from OAM; the first model includes a first number of model segments, each of the first number of model segments has corresponding allocation information, the allocation information includes the inference order of the first number of model segments, and the control radio access network device corresponding to each model segment.

21. A model reasoning device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to: execute the model inference method according to any one of claims 1-4, or execute the model inference method according to any one of claims 5-12, or execute the model inference method according to any one of claims 13-17.

22. A non-transitory computer-readable storage medium, wherein when instructions in the storage medium are executed by a processor of a mobile terminal, the mobile terminal is enabled to perform the model inference method of any one of claims 1-4, or to perform the model inference method of any one of claims 5-12, or to perform the model inference method of any one of claims 13-17.

Images