Granted Patent B2

US 12,732,831 · App. 18/576,740 · Granted Sep 8, 2026

Inventors: Yingying Wang (Beijing, CN); Junshuai Sun (Beijing, CN); Xin Sun (Beijing, CN); Na Li (Beijing, CN); Yun Zhao (Beijing, CN); Guangyi Liu (Beijing, CN)

Assignees: CHINA MOBILE COMMUNICATION CO., LTD RESEARCH INSTITUTE; CHINA MOBILE COMMUNICATIONS GROUP CO., LTD.

H04W24/02H04W56/0015H04W76/20H04W80/02

Abstract

Claims (79)

1 . A base station, comprising:

a memory for storing instructions;

a first processor configured to execute the instructions to:

perform at least one of Artificial Intelligence (AI) task control for a terminal, AI task execution, network decision generation, or data management by interacting with the terminal,

wherein the first processor is further configured to execute the instructions to:

determine one or more configurations of configurations comprising a first AI model and algorithm required to be used, data acquisition requirement, data preprocessing mode, AI task configuration, and AI model evaluation indexes, and configure the one or more configurations to at least one the terminal or a local position, and receive at least a data acquisition result or AI model evaluation result sent by the terminal;

generate a corresponding execution result based on the data acquisition result according to the first AI model and algorithm;

generate network decision according to the execution result; or

maintain at least one of a task queue of AI tasks, an AI model and algorithm library, or an acquired data set.

2 . The base station of claim 1 , wherein the first processor is further configured to execute the instructions to form an AI model library determine the AI model evaluation indexes, evaluate AI models in the AI model library, and determine the first AI model required to be used, the evaluation indexes comprise at least model convergence time or prediction accuracy.

3 . The base station of claim 1 , wherein the base station further comprises a transceiver configured to send the data acquisition requirement to other processors in the base station, and receive the data acquisition result sent by the other processors in the base station,

the transceiver is further configured to send the network decision to the other processors in the base station.

4 . The base station of claim 1 , wherein the first processor is further configured to execute the instructions to determine other network elements required to participate in calculation of the task requirement, schedule and control the other network elements to participate in a training task of the first AI model, and perform synchronization of the first AI model and transmission of model parameters between the base station and the other network elements.

5 . The base station of claim 1 , further comprising:

a second processor configured to perform at least one of:

sending a first task request to the terminal;

receiving a second task requirement from the terminal and sending the second task requirement to the first processor;

receiving the one or more configurations, that are determined by the first processor, of configurations comprising the first AI model and algorithm, the data acquisition requirement, the data preprocessing mode, the AI task configuration, and the AI model evaluation indexes, and sending the one or more configurations to the terminal;

receiving at least one of data, labels, AI model evaluation results or AI model parameters sent by the terminal or other network element nodes, and sending the at least one of data, labels, AI model evaluation results or AI model parameters to the first processor; or

receiving network decision for a second task requirement fed back by the first processor and sending the network decision for the second task requirement to the terminal.

6 . The base station of claim 1 , further comprising:

a third processor configured to perform at least one of:

sending a third task requirement to the terminal;

receiving a fourth task requirement from the terminal and sending the fourth task requirement to the first processor;

receiving the one or more configurations, that are determined by the first processor, of configurations comprising the first AI model and algorithm, the data acquisition requirement, the data preprocessing mode, the AI task configuration, and the AI model evaluation indexes; and sending the one or more configurations to the terminal:

receiving at least one of data, labels, AI model evaluation results or AI model parameters sent by the terminal or other network element nodes and sending the at least one of data, labels, AI model evaluation results or AI model parameters to the first processor; or

receiving network decision for the second task requirement fed back by the first processor and sending the network decision for the second task requirement to the terminal.

7 . The base station of claim 1 , further comprising:

a fourth processor, configured to perform at least one of:

issuing a fifth task requirement to the first processor; or

receiving network decision for the fifth task requirement fed back by the first processor.

8 . A terminal, comprising:

a memory for storing instructions;

a first processor configured to execute the instructions to:

perform at least one of Artificial Intelligence (AI) task control for the terminal, data acquisition and report, or AI task execution by interacting with a base station,

wherein the terminal further comprises a transceiver:

the transceiver is configured to receive one or more configurations of configurations comprising a first AI model and algorithm, data acquisition requirement, data preprocessing mode, AI task configuration, and AI model evaluation indexes sent by the base station;

the first processor is further configured to execute the instructions to execute an AI task according to the first AI model and algorithm configured by the base station, and generate an inference result; perform evaluation according to the AI model evaluation indexes configured by the base station, and obtain at least AI model evaluation results or AI model parameters; and the transceiver is further configured to send at least the AI model evaluation results or AI model parameters to the base station;

the first processor is further configured to execute the instructions to: perform AI model synchronization between the terminal and the base station; or

acquire data according to the data acquisition requirement and the data preprocessing mode, generate at least acquired data or labels; and

the transceiver is further configured to report at least the acquired data or labels to the base station.

9 . The terminal of claim 8 , further comprising:

a second processor configured to perform at least one of:

receiving a first task requirement from the base station and issuing the first task requirement to the first processor;

sending a second task requirement generated by the first processor itself to the base station through the RRC layer;

receiving one or more configurations, that are sent by the base station, of configurations comprising the first AI model and algorithm, the data acquisition requirement, the data preprocessing mode, the AI task configuration, and the AI model evaluation indexes; or

sending at least one of the acquired data, labels, AI model evaluation results or AI model parameters to the base station.

10 . The terminal of claim 8 , further comprising:

a third processor configured to perform at least one of:

receiving a third task requirement from the base station and issuing the third task requirement to the first processor;

sending a fourth task requirement generated by the first processor itself to the base station through the MAC layer;

receiving one or more configurations, that are sent by the base station, of configurations comprising the first AI model and algorithm, the data acquisition requirement, the data preprocessing mode, the AI task configuration, and the AI model evaluation indexes; or

sending at least one of the acquired data, labels, AI model evaluation results or AI model parameters to the base station.

11 . The terminal of claim 8 , further comprising:

a fourth processor, configured to perform at least one of:

issuing a sixth task requirement to the first processor; or

acquiring data according to the data acquisition requirement generated by the first processor, and feeding back to the first processor.

12 . An Artificial Intelligence (AI) task control method, applied to a terminal, comprising:

performing, by a first processor of the terminal, at least one of AI task control for the terminal, data acquisition and report, or AI task execution by interacting with a base station,

wherein the AI task control comprises:

receiving, by the first processor, one or more configurations of configurations comprising: a first AI model and algorithm, data acquisition requirement, data preprocessing mode, AI task configuration, and AI model evaluation indexes that are sent by the base station,

the AI task execution comprises at least one of:

executing, by the first processor, an AI task according to the first AI model and algorithm configured by the base station, and generating an inference result;

performing evaluation according to the AI model evaluation indexes configured by the base station, obtaining at least AI model evaluation results or AI model parameters and sending at least the AI model evaluation results or AI model parameters to the base station; or

performing, by the first processor, AI model synchronization between the terminal and the base station,

wherein the data acquisition and report comprises:

acquiring, by the first processor, data according to the data acquisition requirement and the data preprocessing mode, generates at least acquired data or labels, and reports at least the acquired data or labels to the base station.

13 . The method of claim 12 , further comprising:

performing, by the first processor through interacting with a second processor of the terminal, at least one of:

receiving a first task requirement from the base station forwarded by the second processor;

sending, by the second processor, a second task requirement generated by the first processor itself to the base station;

receiving one or more configurations, that are sent from the base station and forwarded by the second processor, of configurations comprising the first AI model and algorithm, the data acquisition requirement, the data preprocessing mode, the AI task configuration, and the AI model evaluation indexes; or

sending, by the second processor, at least one of the acquired data, labels, AI model evaluation results or AI model parameters to the base station.

14 . The method of claim 12 , further comprising:

performing, by the first processor through interacting with a third processor of the terminal, at least one of:

receiving a third task requirement which is sent from the base station forwarded by the third processor;

sending, by the third processor, a fourth task requirement generated by the first processor itself to the base station;

receiving one or more configurations, that are sent from the base station and forwarded by the third processor, of configurations comprising the first AI model and algorithm, the data acquisition requirement, the data preprocessing mode, the AI task configuration, and the AI model evaluation indexes, and sending the one or more configurations to a transceiver of the terminal; or

sending, by the third processor, at least one of the acquired data, labels, AI model evaluation results or AI model parameters to the base station.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (20)