A model management method and apparatus, a communication device, a storage medium, and a program product

Abstract

This application discloses a model management method and apparatus, communication equipment, storage medium, and program product; the method includes: a first network element sending a user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select a user equipment (UE) for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

Description

Technical Field

[0001] This application relates to the field of wireless technology, and in particular to a model management method and apparatus, communication equipment, storage medium, and program product. Background Technology

[0002] For physical layer use cases of Artificial Intelligence (AI), scenario generalization and the overhead of real-world data acquisition are the main challenges currently facing the industry.

[0003] If AI models are managed separately at the base station level, each base station stores its own model for its specific scenario. This results in significant overhead for model storage. Furthermore, frequent model switching is required when user configurations change or when cell handover occurs, leading to poor generalization ability of the AI ​​models. Summary of the Invention

[0004] To address the aforementioned technical problems, embodiments of this application provide a model management method and apparatus, communication equipment, storage medium, and program product.

[0005] The model management method provided in this application includes:

[0006] The first network element sends a user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select a user equipment (UE) for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

[0007] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0008] In some embodiments, the method further includes:

[0009] The first network element receives training datasets sent by the N base stations; the training datasets are used for training M AI models, where M is a positive integer;

[0010] The first network element distributes the M AI models to the N base stations.

[0011] In some implementations, after the first network element receives the training dataset sent by the N base stations, the method further includes:

[0012] The first network element clusters the training dataset and generates M datasets based on the clustering results, where M is a positive integer;

[0013] The first network element trains the M AI models based on the M datasets.

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

[0015] The first network element establishes a first connection with each of the base stations, and the first connection is used by the base station to send a training dataset to the first network element.

[0016] In some implementations, each of the AI ​​models is associated with one or more identifiers, including base station identifiers and / or cell identifiers;

[0017] The step of distributing the M AI models to the N base stations includes:

[0018] Based on the identifiers associated with the M AI models, the M AI models are distributed to the N base stations.

[0019] In some implementations, the training dataset is a channel-dependent dataset.

[0020] The model management method provided in this application includes:

[0021] The base station receives a user selection configuration and / or a first address sent by a first network element, wherein the user selection configuration is used to select a UE for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

[0022] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0023] In some embodiments, the method further includes:

[0024] The base station configures the first address to the UE;

[0025] After receiving the data packet sent by the UE, the base station routes the data packet to the first network element according to the first address. The data packet carries a training dataset and the destination address of the data packet is the first address.

[0026] In some embodiments, the method further includes:

[0027] The base station selects a UE based on the user selection configuration to perform channel measurements and obtain a training dataset.

[0028] The base station receives the training dataset sent by the UE and sends the training dataset to the first network element; the training dataset is used by the first network element to train an AI model.

[0029] The base station receives the AI ​​model distributed by the first network element.

[0030] In some embodiments, the method further includes:

[0031] The base station configures the AI ​​model to the UE.

[0032] In some embodiments, the method further includes:

[0033] The base station establishes a second connection with the UE and a first connection with the first network element. The second connection is used for the UE to send a training dataset to the base station, and the first connection is used for the base station to send a training dataset to the first network element.

[0034] In some implementations, the training dataset is a channel-dependent dataset.

[0035] The model management method provided in this application includes:

[0036] The UE receives the first address configured by the base station, where the first address is the address of the first network element;

[0037] The UE sends a data packet to the base station. The data packet carries a training dataset and the destination address of the data packet is the first address. The first address is used to route the data packet to the first network element.

[0038] In some implementations, the training dataset is used to train an AI model for the first network element; the method further includes:

[0039] The UE receives the AI ​​model configured by the base station.

[0040] In some implementations, the training dataset is a channel-dependent dataset.

[0041] The model management device provided in this application embodiment is applied to a first network element, and the device includes:

[0042] The first communication unit is used to send user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select user equipment (UE) for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

[0043] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0044] In some implementations, the first communication unit is further configured to receive training datasets sent by the N base stations; the training datasets are used for training M AI models, where M is a positive integer; and to distribute the M AI models to the N base stations.

[0045] In some implementations, the first network element further includes:

[0046] The first processing unit is used to cluster the training dataset and generate M datasets based on the clustering results, where M is a positive integer; and to train the M AI models based on the M datasets.

[0047] In some implementations, the first communication unit is further configured to establish a first connection with each of the base stations, the first connection being used by the base station to send a training dataset to the first network element.

[0048] In some implementations, each of the AI ​​models is associated with one or more identifiers, including base station identifiers and / or cell identifiers;

[0049] The first communication unit is used to distribute the M AI models to the N base stations based on the identifiers associated with the M AI models.

[0050] In some implementations, the training dataset is a channel-dependent dataset.

[0051] The model management device provided in this application embodiment is applied to a base station, and the device includes:

[0052] The second communication unit is used to receive a user selection configuration and / or a first address sent by the first network element, wherein the user selection configuration is used to select the UE for channel measurement; the first address is the address of the first network element, and the first address is used to route the UE's data packets to the first network element.

[0053] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0054] In some implementations, the second communication unit is further configured to configure the first address to the UE; after receiving a data packet sent by the UE, it routes the data packet to the first network element according to the first address, wherein the data packet carries a training dataset and the destination address of the data packet is the first address.

[0055] In some implementations, the second communication unit is further configured to select a UE for channel measurement based on the user selection configuration to obtain a training dataset; receive the training dataset sent by the UE and send the training dataset to the first network element; the training dataset is used by the first network element to train an AI model; and receive the AI ​​model distributed by the first network element.

[0056] In some implementations, the second communication unit is further configured to configure the AI ​​model to the UE.

[0057] In some implementations, the second communication unit is further configured to establish a second connection with the UE and a first connection with the first network element, wherein the second connection is used for the UE to send a training dataset to the base station, and the first connection is used for the base station to send a training dataset to the first network element.

[0058] In some implementations, the training dataset is a channel-dependent dataset.

[0059] The model management device provided in this application embodiment is applied to a UE, and the device includes:

[0060] The third communication unit is used to receive a first address configured by the base station, the first address being the address of the first network element; and to send a data packet to the base station, the data packet carrying a training dataset and the destination address of the data packet being the first address, the first address being used to route the data packet to the first network element.

[0061] In some implementations, the training dataset is used by the first network element to train an AI model; the third communication unit is also used to receive the AI ​​model configured by the base station.

[0062] In some implementations, the training dataset is a channel-dependent dataset.

[0063] The communication device provided in this application includes a processor and a memory. The memory is used to store computer programs, and the processor is used to call and run the computer programs stored in the memory to execute any of the above-described model management methods.

[0064] The computer-readable storage medium provided in this application embodiment is used to store a computer program that causes a computer to execute any of the above-described model management methods.

[0065] The computer program product provided in this application includes computer program instructions that cause a computer to execute any of the above-described model management methods.

[0066] In the technical solution of this application embodiment, a first network element is introduced, through which user selection configuration and / or first address are sent to N base stations. On the one hand, the user selection configuration can assist the base station in selecting the UE for channel measurement to obtain a training dataset for training the AI ​​model. On the other hand, the first address can guide the data packets (carrying the training dataset) from the UE to be correctly routed to the first network element, so that the first network element can train the AI ​​model based on the training dataset, thereby effectively improving the generalization performance of the AI ​​model. Attached Figure Description

[0067] Figure 1 This is a schematic diagram illustrating an application scenario of an embodiment of this application;

[0068] Figure 2 This is a schematic diagram illustrating another application scenario of an embodiment of this application;

[0069] Figure 3 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 1 ;

[0070] Figure 4 This is a schematic diagram of channel data clustering provided in an embodiment of this application;

[0071] Figure 5 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 2 ;

[0072] Figure 6 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 3 ;

[0073] Figure 7 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 4 ;

[0074] Figure 8 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 5 ;

[0075] Figure 9 This is a schematic diagram of the structural composition of the model management device provided in the embodiments of this application. Figure 1 ;

[0076] Figure 10 This is a schematic diagram of the structural composition of the model management device provided in the embodiments of this application. Figure 2 ;

[0077] Figure 11 This is a schematic diagram of the structural composition of the model management device provided in the embodiments of this application. Figure 3 ;

[0078] Figure 12This is a schematic structural diagram of a communication device provided in an embodiment of this application;

[0079] Figure 13 This is a schematic structural diagram of the chip according to an embodiment of this application. Detailed Implementation

[0080] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0081] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of this application.

[0082] like Figure 1 As shown, the communication system may include User Equipment (UE) 110 and Network Equipment 120. Network Equipment 120 can communicate with UE 110 via an air interface. Network Equipment 120 may be an access network device that communicates with UE 110. The access network device can provide communication coverage for a specific area and can communicate with UEs located within that coverage area. The access network device may be a Base Station (BS). UE may be any terminal, such as an access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication equipment, user agent, or user device, etc.

[0083] Figure 2 This is a schematic diagram of another application scenario of this application embodiment.

[0084] like Figure 2 As shown, the communication system includes a UE, a Radio Access Network (RAN), and a Core Network (CN). Specifically, for AI scenarios, the RAN enhances the following functions: task control, computation control, data control, and model management. Model management is implemented by constructing a dedicated network element on the RAN side. For ease of explanation, this element is referred to as the first network element. The first network element handles dataset construction, AI model training, and AI model distribution, effectively improving the generalization ability of the AI ​​model and reducing data acquisition overhead. Of course, the first network element can have other naming conventions, and this application does not impose any restrictions on this; for example, the first network element could be called the model management function.

[0085] It should be noted that the terms "system" and "network" are often used interchangeably in this document. The term "and / or" in this document merely describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship. It should also be understood that "instruction" mentioned in the embodiments of this application can be a direct instruction, an indirect instruction, or an indication of a related relationship. For example, A instructing B can mean that A directly instructs B, for example, B can be obtained through A; it can also mean that A indirectly instructs B, for example, A instructs C, B can be obtained through C; or it can mean that there is a related relationship between A and B. It should also be understood that "correspondence" mentioned in the embodiments of this application can indicate a direct or indirect correspondence between two things, or an related relationship between two things, or a relationship of instruction and being instructed, configuration and being configured, etc. It should also be understood that the "predefined" or "predefined rules" mentioned in the embodiments of this application can be implemented by pre-storing corresponding codes, tables, or other means that can be used to indicate relevant information in the device (e.g., including UE and network devices), and this application does not limit the specific implementation method. For example, predefined can refer to those defined in the protocol.

[0086] It should be noted that the first network element described in this embodiment may also have other names, such as model management function, AI model management function, management network element, etc. This embodiment does not limit the name of the first network element.

[0087] It should be noted that the functions of the AI ​​models described in the embodiments of this application include, but are not limited to, CSI compressed feedback, channel estimation, and channel prediction. Taking the CSI compressed feedback function as an example, the corresponding AI model can also be called a CSI compressed feedback model. Similarly, taking the channel estimation function as an example, the corresponding AI model can also be called a channel estimation model. And taking the channel prediction function as an example, the corresponding AI model can also be called a channel prediction model.

[0088] Figure 3 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 1 ,like Figure 3 As shown, the model management method includes:

[0089] Step 301: The first network element sends a user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select a UE for channel measurement; the first address is the address of the first network element, and the first address is used to route the UE's data packets to the first network element.

[0090] In some implementations, the first network element sends a user selection configuration to each of the N base stations. The user selection configuration is used by the base station to select the UE for channel measurement.

[0091] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0092] Regarding the user mobility speed selection criteria, the criteria include a specific speed range, and the base station selects UEs whose mobility speed falls within that speed range.

[0093] In terms of user signal quality selection conditions, user signal quality selection conditions include a specific signal quality range (such as a specific SINR range), and the base station selects UEs whose signal quality is within that SINR range.

[0094] In terms of user configuration selection conditions, the user configuration selection conditions include specific user configurations, and the base station selects UEs with those specific user configurations.

[0095] Generally, different base stations have different user selection configurations. Thus, different base stations can select different ranges of UEs based on their respective user selection configurations.

[0096] The base station can select the UE based on the user's configuration and complete the channel measurement, thereby enabling targeted channel measurement based on the dataset requirements.

[0097] In some implementations, the first network element receives training datasets sent by N base stations, where N is a positive integer; the training datasets are used to train M AI models, where M is a positive integer; the first network element distributes the M AI models to the N base stations.

[0098] For any one of the N base stations, the base station selects the UE for channel measurement based on the user selection configuration; after the UE performs channel measurement and obtains the training dataset, it sends the training dataset to the base station, which then sends the training dataset to the first network element. Accordingly, the first network element receives the training datasets sent by the N base stations.

[0099] In some implementations, the training dataset is a channel-related dataset, which may be referred to as channel data.

[0100] In some implementations, the training dataset is a dataset associated with the following functions: CSI compression feedback, channel estimation, and channel prediction.

[0101] In some implementations, each channel data is associated with / carries an identifier, which includes a base station identifier and / or a cell identifier. For example, four base stations (N=4) are designated as base station #1, base station #2, base station #3, and base station #4. Base station #1 selects UE #1 and UE #2 for channel measurement based on user selection configuration #1. UE #1 obtains channel data #1, and UE #2 obtains channel data #2. Both channel data #1 and channel data #2 are associated with / carry identifier #1 (i.e., the identifier of base station #1 / cell #1). Base station #2 selects UE #3 and UE #4 for channel measurement based on user selection configuration #2. UE #3 obtains channel data #3, and UE #4 obtains channel data #4. Both channel data #3 and channel data #4 are associated with / carry identifier #2 (i.e., the identifier of base station #2 / cell #2). Base station #3 selects UE #5 and UE #6 for channel measurement based on user-selected configuration #3. UE #5 performs channel measurement to obtain channel data #5, and UE #6 performs channel measurement to obtain channel data #6. Both channel data #5 and channel data #6 are associated with / carry identifier #3 (i.e., the identifier of base station #3 / cell #3). Base station #4 selects UE #7 and UE #8 for channel measurement based on user-selected configuration #4. UE #7 performs channel measurement to obtain channel data #7, and UE #8 performs channel measurement to obtain channel data #8. Both channel data #7 and channel data #8 are associated with / carry identifier #4 (i.e., the identifier of base station #4 / cell #4).

[0102] In some implementations, the channel data includes at least one of the following: a channel matrix, channel characteristic parameters, and channel characteristic statistical parameters.

[0103] In some implementations, the UE performs channel measurements for a specific scenario to obtain a channel matrix. Optionally, the UE extracts features from the channel matrix to obtain channel feature parameters. Optionally, the UE performs fitting (or statistical analysis) on the channel feature parameters to obtain channel feature statistical parameters.

[0104] In some implementations, the UE performs channel measurements based on Minimization of Drive Test (MDT) to obtain the channel matrix. It should be noted that the channel matrix obtained by the UE may differ depending on the scenario.

[0105] In some implementations, the channel matrix includes a frequency-domain channel matrix and / or a time-domain channel matrix. The frequency-domain channel matrix can be represented as H(f) = Nr * Nt * Nrb, where Nr is the receiver port dimension (i.e., the number of receiver ports), Nt is the transmitter port dimension (i.e., the number of transmitter ports), and Nrb is the frequency-domain dimension (i.e., the number of rbs). The time-domain channel matrix can be represented as H(t) = Nr * Nt * Ntau, where Nr is the receiver port dimension (i.e., the number of receiver ports), Nt is the transmitter port dimension (i.e., the number of transmitter ports), and Ntau represents the time-domain dimension (i.e., the number of taus).

[0106] In some implementations, specific transformations of the channel matrix can yield channel characteristic parameters. These parameters include, but are not limited to, channel multipath delay, horizontal angle value, vertical angle value, Rice factor (KF), intra-cluster delay, and intra-cluster angle.

[0107] In some implementations, channel characteristic parameters are fitted to multiple measurement points in a scenario to obtain channel characteristic statistical parameters. These parameters include, but are not limited to: channel multipath delay spread, horizontal angle spread, vertical angle spread, Rice factor (KF), cluster delay spread, and cluster angle spread.

[0108] In some implementations, a first connection is established between the first network element and each base station, wherein a second connection is established between the base station and the UE, the second connection is used for the UE to send a training dataset to the base station, and the first connection is used for the base station to send a training dataset to the first network element.

[0109] In some implementations, the second connection is a session connection between the UE and the base station, and the first connection is a session connection between the base station and the first network element.

[0110] In some implementations, the first network element sends a first address to each of the N base stations. The first address is the address of the first network element (such as an IP address). The first address is used to route data packets from the UE to the first network element. Specifically, the first address is used by the base station to route data packets received from the UE to the first network element. The data packets carry a training dataset and the destination address of the data packets is the first address.

[0111] In the above scheme, the first network element sends the user selection configuration to the base station, allowing the base station to select a suitable UE for channel measurement to ensure the effective collection of the training dataset. Optionally, the first network element also configures its IP address to the base station, and the training dataset collected by the UE can be transmitted to the first network element via session connection or IP routing, effectively ensuring data transmission between the UE, the base station, and the first network element.

[0112] In some implementations, the method further includes: the first network element clusters the training dataset and generates M datasets based on the clustering results, where M is a positive integer; the first network element trains M AI models based on the M datasets.

[0113] Suppose that the first network element acquires K training datasets, which are obtained from channel measurements in K scenarios. The K training datasets are then clustered in L dimensions (L is a positive integer) to obtain M sets of training datasets. Based on these M sets, M datasets are constructed, each containing a set of training datasets.

[0114] In some implementations, the M training datasets correspond to the M scene categories, or in other words, different training datasets correspond to different scene categories.

[0115] For example, such as Figure 4 As shown, the channel data is clustered in 3D (L=3), where the three dimensions are Rice factor (KF), delay spread (DS) / angle spread (AS), and variance of cluster number. Specifically, according to the KF dimension, the channel data can be clustered into m groups, denoted as A1,…,Am; according to the DS / AS dimension, the channel data can be clustered into n groups, denoted as B1,…,Bn; and according to the variance of cluster number dimension, the channel data can be clustered into q groups, denoted as C1,…,Cq. Therefore, 3D clustering of the channel data results in m*n*q groups. Each group of channel data corresponds to a scene category, and each scene category corresponds to an AI model.

[0116] For example, taking the UMA scenario, the channel data is clustered in two dimensions (L=2), with KF and DS / AS representing the two dimensions respectively. According to the KF dimension, the channel data can be clustered into m groups; according to the DS / AS dimension, the channel data can be clustered into n groups. Taking m=2 as an example, according to the KF dimension, the channel data can be clustered into non-direct scattering scenarios (NLoS) and direct scattering scenarios (LoS). Taking n=2 as an example, according to the DS / AS dimension, the channel data can be clustered into complex scattering scenarios and simple scattering scenarios. Therefore, clustering the channel data into m*n=4 groups can also be understood as dividing the UMA scenario into m*n=4 sub-scenarios, each sub-scenarios corresponding to a scenario category, and each scenario category corresponding to an AI model.

[0117] In some implementations, unsupervised learning methods (such as weighted k-means) can be used to perform L-dimensional (L is a positive integer) clustering on K channel data to obtain M groups of channel data.

[0118] In some implementations, different weighting coefficients can be assigned to different scenario categories according to clustering requirements.

[0119] In some implementations, each AI model is associated with one or more identifiers, including base station identifiers and / or cell identifiers; the first network element distributes the M AI models to N base stations based on the identifiers associated with the M AI models.

[0120] For an AI model, it is trained on a dataset, which contains a set of training datasets. Each training dataset in the set of training datasets is associated with / carries an identifier. Thus, the AI ​​model is associated with all the identifiers associated with / carried by the training datasets in the set of training datasets.

[0121] For example, consider 8 channel data sets (K=8): Channel Data #1 (Identifier #1), Channel Data #2 (Identifier #1), Channel Data #3 (Identifier #2), Channel Data #4 (Identifier #2), Channel Data #5 (Identifier #3), Channel Data #6 (Identifier #3), Channel Data #7 (Identifier #4), and Channel Data #8 (Identifier #4). Channel Data #1, Channel Data #2, Channel Data #3, and Channel Data #4 are clustered into Data Set #1. Data Set #1 is used to train AI Model #1. Therefore, AI Model #1 is associated with Identifier #1 and Identifier #2. Channel Data #5, Channel Data #6, Channel Data #7, and Channel Data #8 are clustered into Data Set #2. Data Set #2 is used to train AI Model #2. Therefore, AI Model #2 is associated with Identifier #3 and Identifier #4. The model management function distributes AI Model #1 to Base Station #1 (Identifier #1) and Base Station #2 (Identifier #2), and distributes AI Model #2 to Base Station #3 (Identifier #3) and Base Station #4 (Identifier #4).

[0122] In the technical solution of this application embodiment, the first network element clusters training datasets with similar characteristics and generates corresponding datasets for training the AI ​​model. On the one hand, it eliminates the need for each base station to train the AI ​​model separately, thereby reducing the overhead of AI model training and training dataset collection; on the other hand, it can effectively improve the generalization performance of the AI ​​model.

[0123] Figure 5 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 2 ,like Figure 5 As shown, the model management method includes:

[0124] Step 501: The base station receives a user selection configuration and / or a first address sent by a first network element, wherein the user selection configuration is used to select a UE for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

[0125] The base station receives the user selection configuration sent by the first network element, and selects the UE to perform channel measurements to obtain a training dataset based on the user selection configuration. After the UE performs channel measurements and obtains the training dataset, it sends the training dataset to the base station, which then sends the training dataset to the first network element.

[0126] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0127] Regarding the user mobility speed selection criteria, the criteria include a specific speed range, and the base station selects UEs whose mobility speed falls within that speed range.

[0128] In terms of user signal quality selection conditions, user signal quality selection conditions include a specific signal quality range (such as a specific SINR range), and the base station selects UEs whose signal quality is within that SINR range.

[0129] In terms of user configuration selection conditions, the user configuration selection conditions include specific user configurations, and the base station selects UEs with those specific user configurations.

[0130] In some implementations, the base station selects a UE based on the user selection configuration to perform channel measurements to obtain a training dataset; the base station receives the training dataset sent by the UE and sends the training dataset to a first network element; the training dataset is used by the first network element to train an AI model; the base station receives the AI ​​model distributed by the first network element.

[0131] Here, the training dataset is used for clustering of the first network element, and M datasets are generated based on the clustering results. M AI models are then trained based on the M datasets, where M is a positive integer.

[0132] In some implementations, each training dataset is associated with / carries an identifier, which includes a base station identifier and / or a cell identifier. For example, base station #1 selects UE #1 and UE #2 for channel measurement based on user selection configuration #1. UE #1 performs channel measurement to obtain training dataset #1, and UE #2 performs channel measurement to obtain training dataset #2. Both training dataset #1 and training dataset #2 are associated with / carry identifier #1 (i.e., the identifier of base station #1 / cell #1).

[0133] In some implementations, the training dataset is a channel-related dataset, which may be referred to as channel data.

[0134] In some implementations, the training dataset is a dataset associated with the following functions: CSI compression feedback, channel estimation, and channel prediction.

[0135] In some implementations, the channel data includes at least one of the following: a channel matrix, channel feature parameters, and channel feature statistical parameters, which are channel-related datasets.

[0136] In some implementations, the UE performs channel measurements for a specific scenario to obtain a channel matrix. Optionally, the UE extracts features from the channel matrix to obtain channel feature parameters. Optionally, the UE performs fitting (or statistical analysis) on the channel feature parameters to obtain channel feature statistical parameters.

[0137] In some implementations, the UE performs channel measurements based on the MDT to obtain the channel matrix. It should be noted that the channel matrix obtained by the UE from channel measurements may differ for different scenarios.

[0138] In some implementations, the channel matrix includes a frequency-domain channel matrix and / or a time-domain channel matrix. The frequency-domain channel matrix can be represented as H(f) = Nr * Nt * Nrb, where Nr is the receiver port dimension (i.e., the number of receiver ports), Nt is the transmitter port dimension (i.e., the number of transmitter ports), and Nrb is the frequency-domain dimension (i.e., the number of rbs). The time-domain channel matrix can be represented as H(t) = Nr * Nt * Ntau, where Nr is the receiver port dimension (i.e., the number of receiver ports), Nt is the transmitter port dimension (i.e., the number of transmitter ports), and Ntau represents the time-domain dimension (i.e., the number of taus).

[0139] In some implementations, specific transformations of the channel matrix can yield channel characteristic parameters. These parameters include, but are not limited to, channel multipath delay, horizontal angle value, vertical angle value, Rice factor (KF), intra-cluster delay, and intra-cluster angle.

[0140] In some implementations, channel characteristic parameters are fitted to multiple measurement points in a scenario to obtain channel characteristic statistical parameters. These parameters include, but are not limited to: channel multipath delay spread, horizontal angle spread, vertical angle spread, Rice factor (KF), cluster delay spread, and cluster angle spread.

[0141] In some implementations, a second connection is established between the base station and the UE, and a first connection is established between the base station and the first network element. The second connection is used for the UE to send a training dataset to the base station, and the first connection is used for the base station to send a training dataset to the first network element.

[0142] In some implementations, the second connection is a session connection between the UE and the base station, and the first connection is a session connection between the base station and the first network element.

[0143] In some implementations, the base station receives a first address sent by a first network element and configures the first address to the UE. The first address is the address of the first network element (such as an IP address). After receiving a data packet sent by the UE, the base station routes the data packet to the first network element according to the first address. The data packet carries a training dataset and the destination address of the data packet is the first address.

[0144] In the above scheme, the first network element sends the user selection configuration to the base station, allowing the base station to select a suitable UE for channel measurement to ensure the effective collection of the training dataset. Optionally, the first network element also configures its IP address to the base station, and the training dataset collected by the UE can be transmitted to the first network element via session connection or IP routing, effectively ensuring data transmission between the UE, the base station, and the first network element.

[0145] After receiving the training dataset, the first network element clusters the training dataset, generates M datasets based on the clustering results, and trains M AI models based on the M datasets. For specific implementation details, please refer to the aforementioned... Figure 3 Description of the relevant solutions.

[0146] In some implementations, each AI model is associated with one or more identifiers, including base station identifiers and / or cell identifiers; the first network element distributes the M AI models to N base stations based on the identifiers associated with the M AI models.

[0147] For an AI model, it is trained on a dataset, which contains a set of training datasets. Each training dataset in the set of training datasets is associated with / carries an identifier. Thus, the AI ​​model is associated with all the identifiers associated with / carried by the training datasets in the set of training datasets.

[0148] For example, consider 8 channel data sets (K=8): Channel Data #1 (Identifier #1), Channel Data #2 (Identifier #1), Channel Data #3 (Identifier #2), Channel Data #4 (Identifier #2), Channel Data #5 (Identifier #3), Channel Data #6 (Identifier #3), Channel Data #7 (Identifier #4), and Channel Data #8 (Identifier #4). Channel Data #1, Channel Data #2, Channel Data #3, and Channel Data #4 are clustered into Data Set #1. Data Set #1 is used to train AI Model #1. Therefore, AI Model #1 is associated with Identifier #1 and Identifier #2. Channel Data #5, Channel Data #6, Channel Data #7, and Channel Data #8 are clustered into Data Set #2. Data Set #2 is used to train AI Model #2. Therefore, AI Model #2 is associated with Identifier #3 and Identifier #4. The model management function distributes AI Model #1 to Base Station #1 (Identifier #1) and Base Station #2 (Identifier #2), and distributes AI Model #2 to Base Station #3 (Identifier #3) and Base Station #4 (Identifier #4).

[0149] In some implementations, after receiving the AI ​​model distributed by the first network element, the base station configures the AI ​​model to the UE.

[0150] The technical solution of this application embodiment has several advantages. First, the AI ​​model exhibits strong generalization performance. Aggregating training datasets with similar characteristics enhances the generalization ability of the trained AI model. Second, because a corresponding AI model is trained for each scene category, the size of the AI ​​model is effectively reduced, allowing a smaller model to fit training datasets across multiple scenes and configurations. Third, the cost of acquiring training datasets is effectively reduced. Since training datasets for the same type of scene can be shared / fused, the data acquisition cost for each scene category can be distributed.

[0151] Figure 6 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 3 ,like Figure 6 As shown, the model management method includes:

[0152] Step 601: The UE receives the first address configured by the base station, where the first address is the address of the first network element.

[0153] Step 602: The UE sends a data packet to the base station. The data packet carries a training dataset and the destination address of the data packet is the first address. The first address is used to route the data packet to the first network element.

[0154] In some implementations, the UE sends a training dataset to the base station; the training dataset is used by the first network element to train an AI model.

[0155] Here, after the UE performs channel measurements and obtains the training dataset, it sends the training dataset to the base station, which then sends the training dataset to the first network element. The first network element trains M AI models based on the training datasets from N base stations, as detailed above. Figure 3 , Figure 4 Related descriptions.

[0156] In some implementations, the training dataset is a channel-related dataset, which may be referred to as channel data.

[0157] In some implementations, the training dataset is a dataset associated with the following functions: CSI compression feedback, channel estimation, and channel prediction.

[0158] In some implementations, the channel data includes at least one of the following: a channel matrix, channel characteristic parameters, and channel characteristic statistical parameters.

[0159] In some implementations, the UE performs channel measurements for a specific scenario to obtain a channel matrix. Optionally, the UE extracts features from the channel matrix to obtain channel feature parameters. Optionally, the UE performs fitting (or statistical analysis) on the channel feature parameters to obtain channel feature statistical parameters.

[0160] In some implementations, the UE performs channel measurements based on the MDT to obtain the channel matrix. It should be noted that the channel matrix obtained by the UE from channel measurements may differ for different scenarios.

[0161] In some implementations, the channel matrix includes a frequency-domain channel matrix and / or a time-domain channel matrix. The frequency-domain channel matrix can be represented as H(f) = Nr * Nt * Nrb, where Nr is the receiver port dimension (i.e., the number of receiver ports), Nt is the transmitter port dimension (i.e., the number of transmitter ports), and Nrb is the frequency-domain dimension (i.e., the number of rbs). The time-domain channel matrix can be represented as H(t) = Nr * Nt * Ntau, where Nr is the receiver port dimension (i.e., the number of receiver ports), Nt is the transmitter port dimension (i.e., the number of transmitter ports), and Ntau represents the time-domain dimension (i.e., the number of taus).

[0162] In some implementations, specific transformations of the channel matrix can yield channel characteristic parameters. These parameters include, but are not limited to, channel multipath delay, horizontal angle value, vertical angle value, Rice factor (KF), intra-cluster delay, and intra-cluster angle.

[0163] In some implementations, channel characteristic parameters are fitted to multiple measurement points in a scenario to obtain channel characteristic statistical parameters. These parameters include, but are not limited to: channel multipath delay spread, horizontal angle spread, vertical angle spread, Rice factor (KF), cluster delay spread, and cluster angle spread.

[0164] In some implementations, the UE receives a first address configured by the base station, where the first address is the address (e.g., an IP address) of the first network element. The UE sends a data packet to the base station, the data packet carrying a training dataset and the destination address of the data packet being the first address. The first address is used by the base station to route the data packet to the first network element. For the base station, after receiving the data packet sent by the UE, it routes the data packet to the first network element according to the first address.

[0165] In some implementations, the training dataset is used by the first network element to train an AI model; the first network element distributes M AI models to N base stations, as detailed above. Figure 3 , Figure 4 The relevant description is as follows: The base station is one of N base stations, and this base station configures the AI ​​model from the first network element to the UE. Accordingly, the UE receives the AI ​​model configured by the base station.

[0166] Figure 7 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 4 ,like Figure 7 As shown, the model management method includes:

[0167] Step 701: The first network element sends the user selection configuration to the base station.

[0168] Here, the user selection configuration is used to configure the user selection method for channel measurement. The user selection method includes, but is not limited to, selecting the UE based on parameters such as UE mobility speed, UE SINR, and UE configuration to complete the channel measurement. This allows for targeted channel measurement based on the dataset construction requirements of the first network element. Consequently, richer and more comprehensive channel propagation characteristics can be collected from base station deployment scenarios, improving the generalization performance of the AI ​​model.

[0169] Step 702: The base station selects a suitable UE for channel measurement based on the user's selected configuration.

[0170] Step 703: The UE performs channel matrix acquisition, channel feature parameter extraction, and channel feature statistical parameter fitting in sequence according to the measurement configuration.

[0171] Step 704: Establish a session connection between the UE and the base station so that the UE can feed back the channel characteristic statistical parameters / channel matrix to the base station.

[0172] Step 705: Establish a session connection between the base station and the first network element so that the base station can feed back the channel characteristic statistical parameters / channel matrix to the first network element.

[0173] Step 706: The UE feeds back the channel characteristic statistics parameters / channel matrix to the base station through the established session connection.

[0174] Step 707: The base station feeds back the channel characteristic statistical parameters / channel matrix to the first network element through the established session connection.

[0175] Step 708: The first network element clusters the channel characteristic statistical parameters, generates multiple datasets based on the clustering results, and trains multiple AI models using the multiple datasets.

[0176] Here, the channel feature statistical parameters obtained by the first network element can come from multiple base stations and / or multiple UEs and / or multiple cells. By clustering the channel feature statistical parameters, the channel feature statistical parameters of multiple base stations and / or multiple UEs and / or multiple cells can be clustered. The resulting dataset is more generalizable, and the AI ​​model trained on this dataset is naturally more generalizable.

[0177] Step 709: The first network element sends the AI ​​model to the base station.

[0178] Step 710: The base station configures the AI ​​model to the UE.

[0179] Figure 8 This is a flowchart illustrating the model management method provided in the embodiments of this application. Figure 5 ,like Figure 8 As shown, the model management method includes:

[0180] Step 801: The first network element sends the user selection configuration and the IP address of the first network element to the base station.

[0181] Here, the user selection configuration is used to configure the user selection method for channel measurement. The user selection method includes, but is not limited to, selecting the UE based on parameters such as UE mobility speed, UE SINR, and UE configuration to complete the channel measurement. This allows for targeted channel measurement based on the dataset construction requirements of the first network element. Consequently, richer and more comprehensive channel propagation characteristics can be collected from base station deployment scenarios, improving the generalization performance of the AI ​​model.

[0182] Step 802: The base station selects a suitable UE for channel measurement based on the user's configuration and configures the IP address of the first network element to the UE.

[0183] Step 803: The UE performs channel matrix acquisition, channel feature parameter extraction, and channel feature statistical parameter fitting in sequence according to the measurement configuration.

[0184] Step 804: Establish a connection between the UE and the base station so that the UE can feed back the channel characteristic statistical parameters / channel matrix to the base station.

[0185] Step 805: The UE sends a data packet with the destination address being the IP address of the first network element to the base station through the connection. This data packet carries channel characteristic statistical parameters / channel matrix.

[0186] Step 806: The base station routes the data packet to the first network element based on the destination address of the data packet.

[0187] Step 807: The first network element clusters the channel characteristic statistical parameters, generates multiple datasets based on the clustering results, and trains multiple AI models using the multiple datasets.

[0188] Here, the channel feature statistical parameters obtained by the first network element can come from multiple base stations and / or multiple UEs and / or multiple cells. By clustering the channel feature statistical parameters, the channel feature statistical parameters of multiple base stations and / or multiple UEs and / or multiple cells can be clustered. The resulting dataset is more generalizable, and the AI ​​model trained on this dataset is naturally more generalizable.

[0189] Step 808: The first network element sends the AI ​​model to the base station.

[0190] Step 809: The base station configures the AI ​​model to the UE.

[0191] Figure 9 This is a schematic diagram of the structural composition of the model management device provided in the embodiments of this application. Figure 1 It is applied to the first network element, such as Figure 9 As shown, the model management device includes:

[0192] The first communication unit 901 is used to send user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select user equipment (UE) for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

[0193] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0194] In some implementations, the first communication unit 901 is further configured to receive training datasets sent by the N base stations; the training datasets are used for training M AI models, where M is a positive integer; and to distribute the M AI models to the N base stations.

[0195] In some implementations, the first network element further includes:

[0196] The first processing unit 902 is used to cluster the training dataset and generate M datasets based on the clustering results, where M is a positive integer; and to train the M AI models based on the M datasets.

[0197] In some implementations, the first communication unit 901 is further configured to establish a first connection with each of the base stations, the first connection being used by the base station to send a training dataset to the first network element.

[0198] In some implementations, each of the AI ​​models is associated with one or more identifiers, including base station identifiers and / or cell identifiers;

[0199] The first communication unit 901 is used to distribute the M AI models to the N base stations based on the identifiers associated with the M AI models.

[0200] In some implementations, the training dataset is a channel-dependent dataset.

[0201] Those skilled in the art should understand that Figure 9 The functions of each unit in the model management device shown can be understood by referring to the relevant descriptions of the aforementioned methods. Figure 9 The functions of each unit in the model management device shown can be implemented by a program running on a processor or by specific logic circuits.

[0202] Figure 10 This is a schematic diagram of the structural composition of the model management device provided in the embodiments of this application. Figure 2 Applications in base stations, such as Figure 10 As shown, the model management device includes:

[0203] The second communication unit 1001 is used to receive a user selection configuration and / or a first address sent by the first network element, wherein the user selection configuration is used to select the UE for channel measurement; the first address is the address of the first network element, and the first address is used to route the UE's data packets to the first network element.

[0204] In some implementations, the user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

[0205] In some embodiments, the second communication unit 1001 is further configured to configure the first address to the UE; after receiving a data packet sent by the UE, it routes the data packet to the first network element according to the first address, wherein the data packet carries a training dataset and the destination address of the data packet is the first address.

[0206] In some implementations, the second communication unit 1001 is further configured to select a UE for channel measurement based on the user selection configuration to obtain a training dataset; receive the training dataset sent by the UE and send the training dataset to the first network element; the training dataset is used by the first network element to train an AI model; and receive the AI ​​model distributed by the first network element.

[0207] In some implementations, the second communication unit 1001 is also used to configure the AI ​​model to the UE.

[0208] In some implementations, the second communication unit 1001 is further configured to establish a second connection with the UE and a first connection with the first network element. The second connection is used for the UE to send a training dataset to the base station, and the first connection is used for the base station to send a training dataset to the first network element.

[0209] In some implementations, the training dataset is a channel-dependent dataset.

[0210] Those skilled in the art should understand that Figure 10 The functions of each unit in the model management device shown can be understood by referring to the relevant descriptions of the aforementioned methods. Figure 10 The functions of each unit in the model management device shown can be implemented by a program running on a processor or by specific logic circuits.

[0211] Figure 11 This is a schematic diagram of the structural composition of the model management device provided in the embodiments of this application. Figure 3 Applied to UE, such as Figure 11 As shown, the model management device includes:

[0212] The third communication unit 1101 is used to receive a first address configured by the base station, the first address being the address of the first network element; and to send a data packet to the base station, the data packet carrying a training dataset and the destination address of the data packet being the first address, the first address being used to route the data packet to the first network element.

[0213] In some implementations, the training dataset is used by the first network element to train an AI model; the third communication unit 1101 is also used to receive the AI ​​model configured by the base station.

[0214] In some implementations, the training dataset is a channel-dependent dataset.

[0215] Those skilled in the art should understand that Figure 11 The functions of each unit in the model management device shown can be understood by referring to the relevant descriptions of the aforementioned methods. Figure 11 The functions of each unit in the model management device shown can be implemented by a program running on a processor or by specific logic circuits.

[0216] Figure 12 This is a schematic structural diagram of a communication device 1200 provided in an embodiment of this application. Figure 12 The communication device 1200 shown includes a processor 1210, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0217] Optionally, such as Figure 12 As shown, the communication device 1200 may further include a memory 1220. The processor 1210 can retrieve and run computer programs from the memory 1220 to implement the methods described in this embodiment.

[0218] The memory 1220 can be a separate device independent of the processor 1210, or it can be integrated into the processor 1210.

[0219] Optionally, such as Figure 12 As shown, the communication device 1200 may also include a transceiver 1230. The processor 1210 can control the transceiver 1230 to communicate with other devices. Specifically, it can send information or data to other devices or receive information or data sent by other devices.

[0220] The transceiver 1230 may include a transmitter and a receiver. The transceiver 1230 may further include an antenna, and the number of antennas may be one or more.

[0221] Optionally, the communication device 1200 may specifically be the first network element in the embodiments of this application, and the communication device 1200 may implement the corresponding processes implemented by the first network element in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0222] Optionally, the communication device 1200 may specifically be a base station in the embodiments of this application, and the communication device 1200 may implement the corresponding processes implemented by the base station in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0223] Optionally, the communication device 1200 may specifically be a UE in the embodiments of this application, and the communication device 1200 may implement the corresponding processes implemented by the UE in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0224] Figure 13 This is a schematic structural diagram of the chip according to an embodiment of this application. Figure 13 The chip 1300 shown includes a processor 1310, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0225] Optionally, such as Figure 13 As shown, chip 1300 may further include memory 1320. Processor 1310 can retrieve and run computer programs from memory 1320 to implement the methods described in this embodiment.

[0226] The memory 1320 can be a separate device independent of the processor 1310, or it can be integrated into the processor 1310.

[0227] Optionally, the chip 1300 may also include an input interface 1330. The processor 1310 can control the input interface 1330 to communicate with other devices or chips; specifically, it can acquire information or data sent by other devices or chips.

[0228] Optionally, the chip 1300 may also include an output interface 1340. The processor 1310 can control the output interface 1340 to communicate with other devices or chips, specifically, to output information or data to other devices or chips.

[0229] Optionally, the chip can be applied to the first network element in the embodiments of this application, and the chip can implement the corresponding processes implemented by the first network element in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0230] Optionally, the chip can be applied to the base station in the embodiments of this application, and the chip can implement the corresponding processes implemented by the base station in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0231] Optionally, the chip can be applied to the UE in the embodiments of this application, and the chip can implement the corresponding processes implemented by the UE in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0232] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0233] It should be understood that the processor in the embodiments of this application may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method embodiments can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor described above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0234] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0235] It should be understood that the above-described memory is exemplary and not a limiting description. For example, the memory in the embodiments of this application may also be static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM), etc. That is to say, the memory in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0236] This application also provides a computer-readable storage medium for storing computer programs.

[0237] Optionally, the computer-readable storage medium can be applied to the first network element in the embodiments of this application, and the computer program causes the computer to execute the corresponding processes implemented by the first network element in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0238] Optionally, the computer-readable storage medium can be applied to the base station in the embodiments of this application, and the computer program causes the computer to execute the corresponding processes implemented by the base station in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0239] Optionally, the computer-readable storage medium can be applied to the UE in the embodiments of this application, and the computer program causes the computer to execute the corresponding processes implemented by the UE in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0240] This application also provides a computer program product, including computer program instructions.

[0241] Optionally, the computer program product can be applied to the first network element in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the first network element in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0242] Optionally, the computer program product can be applied to the base station in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the base station in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0243] Optionally, the computer program product can be applied to the UE in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the UE in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0244] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0245] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0246] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0247] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0248] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0249] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0250] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A model management method, characterized in that, The method includes: The first network element sends a user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select a user equipment (UE) for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

2. The method according to claim 1, characterized in that, The user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

3. The method according to claim 1, characterized in that, The method further includes: The first network element receives training datasets sent by the N base stations; the training datasets are used for training M AI models, where M is a positive integer; The first network element distributes the M AI models to the N base stations.

4. The method according to claim 3, characterized in that, After the first network element receives the training dataset sent by the N base stations, the method further includes: The first network element clusters the training dataset and generates M datasets based on the clustering results, where M is a positive integer; The first network element trains the M AI models based on the M datasets.

5. The method according to claim 3, characterized in that, The method further includes: The first network element establishes a first connection with each of the base stations, and the first connection is used by the base station to send a training dataset to the first network element.

6. The method according to any one of claims 3 to 5, characterized in that, Each AI model is associated with one or more identifiers, including base station identifiers and / or cell identifiers; The step of distributing the M AI models to the N base stations includes: Based on the identifiers associated with the M AI models, the M AI models are distributed to the N base stations.

7. The method according to any one of claims 3 to 5, characterized in that, The training dataset is a channel-related dataset.

8. A model management method, characterized in that, The method includes: The base station receives a user selection configuration and / or a first address sent by a first network element, wherein the user selection configuration is used to select a UE for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

9. The method according to claim 8, characterized in that, The user selection configuration includes at least one of the following: user movement speed selection condition, user signal quality selection condition, and user configuration selection condition.

10. The method according to claim 8, characterized in that, The method further includes: The base station configures the first address to the UE; After receiving the data packet sent by the UE, the base station routes the data packet to the first network element according to the first address. The data packet carries a training dataset and the destination address of the data packet is the first address.

11. The method according to claim 8, characterized in that, The method further includes: The base station selects a UE based on the user selection configuration to perform channel measurements and obtain a training dataset. The base station receives the training dataset sent by the UE and sends the training dataset to the first network element; the training dataset is used by the first network element to train an AI model. The base station receives the AI ​​model distributed by the first network element.

12. The method according to claim 11, characterized in that, The method further includes: The base station configures the AI ​​model to the UE.

13. The method according to claim 11, characterized in that, The method further includes: The base station establishes a second connection with the UE and a first connection with the first network element. The second connection is used for the UE to send a training dataset to the base station, and the first connection is used for the base station to send a training dataset to the first network element.

14. The method according to any one of claims 11 to 13, characterized in that, The training dataset is a channel-related dataset.

15. A model management method, characterized in that, The method includes: The UE receives the first address configured by the base station, where the first address is the address of the first network element; The UE sends a data packet to the base station. The data packet carries a training dataset and the destination address of the data packet is the first address. The first address is used to route the data packet to the first network element.

16. The method according to claim 15, characterized in that, The training dataset is used to train the AI ​​model for the first network element; the method further includes: The UE receives the AI ​​model configured by the base station.

17. The method according to claim 15 or 16, characterized in that, The training dataset is a channel-related dataset.

18. A model management device, characterized in that, Applied to the first network element, the device includes: The first communication unit is used to send user selection configuration and / or a first address to N base stations, where N is a positive integer; wherein, the user selection configuration is used to select user equipment (UE) for channel measurement; the first address is the address of the first network element, and the first address is used to route data packets of the UE to the first network element.

19. The apparatus according to claim 18, characterized in that, The first communication unit is further configured to receive training datasets sent by the N base stations; the training datasets are used for training M AI models, where M is a positive integer; and to distribute the M AI models to the N base stations.

20. The apparatus according to claim 18, characterized in that, The first network element also includes: The first processing unit is used to cluster the training dataset and generate M datasets based on the clustering results, where M is a positive integer; and to train the M AI models based on the M datasets.

21. A model management device, characterized in that, Applied to a base station, the device includes: The second communication unit is used to receive a user selection configuration and / or a first address sent by the first network element, wherein the user selection configuration is used to select the UE for channel measurement; the first address is the address of the first network element, and the first address is used to route the UE's data packets to the first network element.

22. The apparatus according to claim 21, characterized in that, The second communication unit is further configured to receive a training dataset sent by the UE, and send the training dataset to the first network element; the training dataset is used by the first network element to train an AI model; and receive an AI model distributed by the first network element.

23. A model management device, characterized in that, Applied to a UE, the device includes: The third communication unit is used to receive a first address configured by the base station, the first address being the address of the first network element; and to send a data packet to the base station, the data packet carrying a training dataset and the destination address of the data packet being the first address, the first address being used to route the data packet to the first network element.

24. The apparatus according to claim 23, characterized in that, The third communication unit is also used to receive the AI ​​model configured by the base station.

25. A communication device, characterized in that, include: A processor and a memory for storing a computer program, the processor for calling and running the computer program stored in the memory to perform the method as described in any one of claims 1 to 17.

26. A computer-readable storage medium, characterized in that, Used to store a computer program that causes a computer to perform the method as described in any one of claims 1 to 17.

27. A computer program product, characterized in that, It includes computer program instructions that cause a computer to perform the method as described in any one of claims 1 to 17.

Images