An intelligent radio access network

Abstract

Embodiments of the present application provide a communication method and device for introducing artificial intelligence (AI) in a radio access network (RAN). The method comprises: a radio intelligent controller (RIC) sending configuration information of one or more AI tasks to a base station. The configuration information of each AI task is used to indicate one or more of the following of the AI task: a task identifier (ID), a task type, a task content, a task execution subject, and a task status. The execution subject of the AI task can be the base station or a terminal.

Description

TECHNICAL FIELD

[0001] The present application relates to the communication technology field, and particularly relates to a smart radio access network. BACKGROUND

[0002] In a wireless communication network, for example, in a mobile communication network, the services supported by the network are more and more diverse, and therefore the needs to be met are more and more diverse. For example, the network needs to be able to support super-high rate, super-low latency, and / or super-large connection. This feature makes the network planning, network configuration, and / or resource scheduling more and more complex. In addition, as the network becomes more and more powerful, for example, supports higher and higher spectrum, supports high-order multiple input multiple output (MIMO) technology, supports beamforming, and / or supports new technologies such as beam management, network energy saving has become a hot research topic. These new needs, new scenarios, and new features bring unprecedented challenges to network planning, operation and maintenance, and efficient operation. SUMMARY

[0003] Embodiments of the present application provide a communication method, which aims to introduce artificial intelligence (AI) in a radio access network (RAN), so that the network can be intelligent, and an efficient RAN can be provided.

[0004] In a first aspect, a communication method is provided. The execution subject of the method can be a radio intelligent controller (RIC) or a device capable of supporting the RIC to implement the function, such as a chip, etc., which is not limited. Wherein the method comprises: sending first task configuration information to a base station, the first task configuration information being used to indicate the configuration information of one or more tasks; wherein the execution subject of each task in the one or more tasks is a terminal device or the base station.

[0005] Optionally, the execution subjects of different tasks can be the same or different, which is not limited. For example, the first task configuration information indicates the configuration information of X1 tasks, wherein the execution subjects of X2 tasks are the base station, and the execution subjects of X3 tasks are the terminal device. X1 is a positive integer greater than or equal to 1, X2 is an integer greater than or equal to 0 and less than or equal to X1, and X3 is an integer greater than or equal to 0 and less than or equal to X1. The sum of X2 and X3 is equal to X1.

[0006] Optionally, the execution subject of each task is a base station, including: the execution subject of each task is a centralized unit CU of the base station or a distributed unit DU of the base station. At this time, in one possible implementation, the execution subject of each of the one or more tasks is the terminal device, the CU, or the DU; the execution subject of each of the one or more tasks is the terminal device or the CU; or the execution subject of each of the one or more tasks is the terminal device or the DU.

[0007] Optionally, the execution subject of each task is the CU of the base station, including: the execution subject of each task is a centralized unit-control plane CU-CP of the base station or a centralized unit-user plane CU-UP of the base station. At this time, in one possible implementation, the execution subject of each of the one or more tasks is the terminal device, the CU-CP, the CU-UP, or the DU; the execution subject of each of the one or more tasks is the terminal device, the CU-CP, or the CU-UP; the execution subject of each of the one or more tasks is the terminal device, the CU-CP, or the DU; the execution subject of each of the one or more tasks is the terminal or the CU-CP; the execution subject of each of the one or more tasks is the terminal device or the CU-UP; or the execution subject of each of the one or more tasks is the terminal device, the CU-UP, or the DU.

[0008] By the above method, the RIC can send an AI task to the base station, and / or send an AI task to the terminal through the base station, for implementing an AI function in the RAN, so that AI can be effectively introduced into the RAN, and an efficient RAN can be implemented. For example, the efficiency of network planning, network configuration, and / or resource scheduling, etc. can be improved. When AI is introduced into the RAN by the method, the existing network can be better compatible, and new AI functions can be easily introduced.

[0009] In one possible implementation, for each of the one or more tasks, the configuration information of the task is used to indicate one or more of the following of the task: a task identifier ID, a task type, a task content, a task execution subject, and a task state.

[0010] By the method, the task information can be configured for the execution subject of the AI task, so that the execution subject of the task can know how to execute the corresponding AI task. Thus, the AI function can be implemented in the network.

[0011] In a possible implementation, the task type is data collection, inference result publishing, model publishing, or model training. For example, when the configuration information of a task does not indicate the task type, it can be agreed in the protocol or indicated in advance through other signaling that the type of the task is data collection, inference result publishing, model publishing, or model training. For another example, when the configuration information of a task indicates the task type, the type of the task can be indicated from a plurality of task types, which can include at least one of data collection, inference result publishing, model publishing, and model training. Alternatively, the plurality of task types can further include other task types, which are not limited.

[0012] Alternatively, when the task type is model training, the task type is further used to indicate that the task type is distributed model training or centralized model training. Alternatively, the task type is data collection, inference result publishing, model publishing, or model training, including that the task type is data collection, inference result publishing, model publishing, distributed model training, or centralized model training.

[0013] In the embodiments of the present application, the inference result inferred by the RIC using the AI model can be a parameter configuration on the RAN side and / or a parameter configuration on the terminal side. For example, the parameter value on the RAN side includes the parameter configuration of the cell and / or the parameter configuration of the base station. At this time, the inference result publishing can also be described as the configuration parameter, and the inference result can be described as the parameter value.

[0014] Through the method, data can be collected to assist in implementing model training and / or inference function in the AI function, the result inferred by the RIC can be published to the base station or the terminal device, the AI model can be published to the base station or the terminal device, and / or the base station or the terminal device can be instructed to perform model training. Thus, various possible AI functions can be introduced into the RAN, and an efficient RAN can be obtained.

[0015] In a possible implementation, when the type of the task is data collection, the task content indicates one or more of the following: a measurement type of data, a measurement condition, and a measurement result reporting (or described as data reporting) manner. Alternatively, the task content indicates one or more of the following: a data type and a data reporting manner.

[0016] When the type of the task is inference result publishing, the task content indicates the inference result,

[0017] When the type of the task is model publishing, the task content indicates model information, or

[0018] The type of the task is model training, and the task content indicates one or more of the following: a condition of reporting model parameter information or reporting model parameter gradient information, information of a reference neural network, and a neural network training data set.

[0019] Through the method, the specific content of the AI task can be indicated to the execution subject of the AI task, so that the execution subject can perform the corresponding task according to the indication of the task content, thereby realizing the AI function in the network.

[0020] In a possible implementation, the task state includes activation or deactivation, or the task state includes activation, deactivation, or release.

[0021] Through the method, the execution of the task can be flexibly controlled. For example, the task can be configured to be activated in part of the time period, or the configured task can be released, thereby saving the power consumption of the task execution subject.

[0022] In a possible implementation, the method further includes receiving a first interface establishment request message from a CU of the base station.

[0023] Through the method, the interface between the RIC and the CU can be established. Thus, communication can be performed between the RIC and the CU.

[0024] In a possible implementation, the first interface establishment request message is used to indicate one or more of the following: a message type, an ID of the CU, capability information of the CU, configuration information of the CU, and state information of the CU.

[0025] Through the method, the information of the CU can be obtained, thereby being used for AI model training and / or inference in the RIC.

[0026] In a possible implementation, the method further includes receiving a second interface establishment request message from a DU of the base station.

[0027] Through the method, the interface between the RIC and the DU can be established. Thus, communication can be performed between the RIC and the DU.

[0028] In a possible implementation, the second interface establishment request message is used to indicate one or more of the following: a message type, an ID of the DU, capability information of the DU, configuration information of the DU, and state information of the DU.

[0029] Through the method, the information of the DU can be obtained, thereby being used for AI model training and / or inference in the RIC.

[0030] In a possible implementation, the method further includes: receiving, from the base station, information of the terminal device, the information of the terminal device including one or more of the following: capability information of the terminal device, configuration information of the terminal device, and state information of the terminal device. Optionally, receiving, from the base station, the information of the terminal device includes: receiving, from a CU of the base station, the information of the terminal device.

[0031] By this method, the information of the terminal device can be obtained, which can be used for AI model training and / or inference in the RIC.

[0032] In a possible implementation, the one or more tasks include at least one data collection task, and the method further includes: receiving, from the base station, collected data. Receiving, from the base station, the collected data includes: receiving, from a CU of the base station, the collected data. Optionally, when the execution subject of the at least one data collection task is the base station, the data is collected by the base station. Optionally, when the execution subject of the at least one data collection task is the CU, the data is collected by the CU. Optionally, when the execution subject of the at least one data collection task is a DU, the data is collected by the DU. The data collected by the DU can be sent by the DU to the CU and then sent by the CU to the RIC. Optionally, when the execution subject of the at least one data collection task is a CU-CP, the data is collected by the CU-CP. Optionally, when the execution subject of the at least one data collection task is a CU-UP, the data is collected by the CU-UP. The data collected by the CU-UP can be sent by the CU-UP to the CU-CP and then sent by the CU-CP to the RIC. Optionally, when the execution subject of the at least one data collection task is the terminal device, the data is collected by the terminal device and sent by the terminal device to the base station.

[0033] By this method, the data that is intended to be collected can be obtained, which can be used to assist in implementing model training and / or inference functions.

[0034] In a possible implementation, the method further includes: publishing, to the base station, an inference result. Publishing, to the base station, the inference result includes: publishing, to a CU of the base station, the inference result.

[0035] By this method, the inference result can be published to the base station or the terminal device. For example, after the RIC collects data, the RIC can use the data to perform inference, so as to obtain a parameter configuration on the RAN side and / or a parameter configuration of the terminal device. The RIC can publish the parameter values to the base station and / or the terminal device, so that the base station and / or the terminal device can update the corresponding parameter values, thereby improving the performance of the network.

[0036] In a possible implementation, the one or more tasks include at least one model training task, and the method further includes: receiving model parameter information or model parameter gradient information from the base station. The model parameter information or model parameter gradient information is received from a CU of the base station. The parameter information or parameter gradient information can be sent by the terminal device to the base station.

[0037] Through the method, federated learning can be implemented on the terminal side. The efficiency of AI model training is improved.

[0038] In a second aspect, a communication method is provided. The execution subject of the method can be a base station, a CU, or a device capable of supporting the base station or the CU to implement the function, such as a chip, and the like, without limitation. The method includes: receiving first task configuration information from a radio intelligent controller (RIC), where the first task configuration information is used to indicate configuration information of one or more tasks; and the execution subject of each task in the one or more tasks is a terminal device or a base station.

[0039] The configuration information of the task can be referred to the first aspect, and details are not repeated here.

[0040] In a possible implementation, for at least one task in the one or more tasks, the execution subject of the at least one task is the terminal device, and the method further includes: indicating information of each task in the at least one task to the terminal device through radio resource control (RRC) signaling, a system information block (SIB), a master information block (MIB), or a paging message.

[0041] In a possible implementation, for at least one task in the one or more tasks, the execution subject of the at least one task is a DU, and the method further includes: sending content of each task in the at least one task to the DU.

[0042] In a possible implementation, for at least one task in the one or more tasks, the execution subject of the at least one task is a CU-UP, and the method further includes: sending content of each task in the at least one task to the CU-UP.

[0043] In a possible implementation, the method further includes: sending a first interface establishment request message to the RIC. The first interface establishment request message is described in the first aspect, and details are not repeated here.

[0044] In a possible implementation, the method further includes: sending information of the terminal device to the RIC. The information of the terminal is described in the first aspect, and details are not repeated here.

[0045] In a possible implementation, the method further includes that the one or more tasks include at least one data collection task, and the method further includes: sending the collected data to the RIC. The introduction about the data can be referred to the detailed description of the first aspect, which is not described here. For example, the method further includes: receiving the data from the terminal device, the DU, or the CU-UP.

[0046] In a possible implementation, the method further includes: receiving the inference result from the RIC. Optionally, the method further includes: sending the inference result to the terminal device, the DU, or the CU-UP. Optionally, when the inference result is sent to the terminal device, the sending includes: sending the inference result to the terminal device through RRC signaling, a SIB, a MIB, or a paging message.

[0047] In a possible implementation, the one or more tasks include at least one model training task, and the method further includes: receiving model parameter information or model parameter gradient information from the terminal device, and sending the model parameter information or the model parameter gradient information to the RIC.

[0048] In a third aspect, a communication method is provided. The execution subject of the method can be a terminal device or a device capable of supporting the terminal device to implement the function, for example, a chip, and the like, which is not limited. The method includes: receiving, from a base station, information of one or more tasks through radio resource control (RRC) signaling, a system information block (SIB), a master information block (MIB), or a paging message; and wherein the information of the one or more tasks includes information of each task, and the information of each task is used to indicate one or more of the following: a task identifier (ID), a task type, a task content, a task execution subject, and a task state.

[0049] The introduction about the task ID, the task type, the task content, the task execution subject, and the task state can be referred to the first aspect or the second aspect, which is not described here.

[0050] In a possible implementation, the method further includes: sending, to the base station, information of the terminal device, wherein the information of the terminal device includes one or more of the following: capability information of the terminal device, configuration information of the terminal device, and state information of the terminal device.

[0051] In a possible implementation, the one or more tasks include at least one data collection task, and the method further includes: sending collected data to the base station.

[0052] In a possible implementation, the method further includes: receiving an inference result from the base station.

[0053] In a possible implementation, the one or more tasks include at least one model training task, and the method further includes: sending model parameter information or model parameter gradient information to the base station.

[0054] In a fourth aspect, an apparatus is provided. The apparatus can be a RIC or another apparatus capable of implementing the method described in the first aspect. The apparatus can be installed in a RIC or used in conjunction with a RIC. In one design, the apparatus can include a module corresponding to each of the steps of the method described in the first aspect. The module can be implemented in hardware, software, or a combination thereof. In one design, the apparatus can include a processing module and a communication module.

[0055] In a possible implementation, the communication module is configured to send, to the base station, first task configuration information that indicates configuration information of one or more tasks, where an execution subject of each of the one or more tasks is a terminal device or the base station. The configuration information of the task can be generated by the processing module.

[0056] The communication module can receive and / or send other information as described in the first aspect, which will not be repeated here.

[0057] In a fifth aspect, an apparatus is provided. The apparatus can be a base station or another apparatus capable of implementing the method described in the second aspect. The apparatus can be installed in a base station or used in conjunction with a base station. In one design, the apparatus can include a module corresponding to each of the steps of the method described in the second aspect. The module can be implemented in hardware, software, or a combination thereof. In one design, the apparatus can include a processing module and a communication module.

[0058] In a possible implementation, the communication module is configured to receive, from a radio intelligent controller (RIC), first task configuration information that indicates configuration information of one or more tasks, where an execution subject of each of the one or more tasks is a terminal device or the base station. The processing module can be configured to receive the first task configuration information from the communication module and process the first task configuration information.

[0059] The communication module can receive and / or send other information as described in the second aspect, which will not be repeated here.

[0060] In a sixth aspect, an apparatus is provided. The apparatus can be a terminal device or another apparatus capable of implementing the method described in the third aspect. The apparatus can be installed in a terminal device or used in conjunction with a terminal device. In one design, the apparatus can include modules corresponding to each of the steps of the method described in the third aspect. The modules can be hardware circuits, software codes, or a combination thereof. In one design, the apparatus can include a processing module and a communication module.

[0061] In one possible implementation, the communication module is configured to receive information of one or more tasks from a base station via radio resource control (RRC) signaling, a system information block (SIB), a master information block (MIB), or a paging message. The information of each task of the one or more tasks can indicate one or more of a task identifier (ID), a task type, a task content, a task performer, and a task status.

[0062] The communication module can receive and / or transmit other information as described in the third aspect, which will not be repeated here.

[0063] In a seventh aspect, an apparatus is provided. The apparatus can include a processor configured to implement a method described in the first aspect. The apparatus can also include a memory configured to store instructions. The processor can implement the method described in the first aspect when executing the instructions stored in the memory. The apparatus can also include a communication interface configured to enable communication between the apparatus and another device. In an embodiment, the communication interface can be a transceiver, a circuit, a bus, a module, a pin, or another type of communication interface.

[0064] In one possible design, the apparatus includes:

[0065] a memory configured to store program instructions;

[0066] a processor configured to send first task configuration information to a base station via the communication interface. The first task configuration information can indicate configuration information of one or more tasks. The performer of each task of the one or more tasks can be a terminal device or the base station.

[0067] The processor can receive and / or transmit other information via the communication interface as described in the first aspect, which will not be repeated here.

[0068] In an eighth aspect, an embodiment of the present application provides a device, which comprises a processor configured to implement the method described in the second aspect. The device can further comprise a memory configured to store instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, the method described in the second aspect can be implemented. The device can further comprise a communication interface configured to enable the device to communicate with other devices.

[0069] In a possible design, the device comprises:

[0070] a memory configured to store program instructions;

[0071] a processor configured to receive, from a radio intelligent controller (RIC) via the communication interface, first task configuration information, where the first task configuration information is used to indicate configuration information of one or more tasks, and where an execution subject of each of the one or more tasks is a terminal device or a base station.

[0072] The processor can receive and / or send other information via the communication interface, which is described in the second aspect and will not be repeated here.

[0073] In a ninth aspect, an embodiment of the present application provides a device, which comprises a processor configured to implement the method described in the third aspect. The device can further comprise a memory configured to store instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, the method described in the third aspect can be implemented. The device can further comprise a communication interface configured to enable the device to communicate with other devices.

[0074] In a possible design, the device comprises:

[0075] a memory configured to store program instructions;

[0076] a processor configured to receive, from a base station via the communication interface, information of one or more tasks by using radio resource control (RRC) signaling, a system information block (SIB), a master information block (MIB), or a paging message, where each task information in the information of the one or more tasks is used to indicate one or more of the following for the each task: a task identifier (ID), a task type, a task content, a task execution subject, and a task state.

[0077] The processor can receive and / or send other information via the communication interface, which is described in the third aspect and will not be repeated here.

[0078] In a tenth aspect, a communication method is provided. The execution subject of the method can be a radio intelligent controller (RIC) or a device (e.g., a chip, etc.) capable of supporting the RIC to implement the function, which is not limited. The method includes: sending, to a terminal device, second task configuration information through a first protocol layer, the second task configuration information being used to indicate configuration information of one or more tasks, and the execution subject of each of the one or more tasks including one or more terminal devices.

[0079] Optionally, the execution subjects of different tasks can be the same or different, which is not limited. Optionally, the first protocol layer is a protocol layer other than an RRC protocol layer, or the first protocol layer is described as a non-RRC layer, or the first protocol layer is described as not being an RRC layer.

[0080] In a possible implementation, for each of the one or more tasks, the configuration information of the task is used to indicate one or more of the following of the task: a task identifier (ID), a task type, task content, a task execution subject, and a task state.

[0081] Through the above method, the RIC can send an AI task to a terminal device, which is used to implement an AI function in a RAN, so that AI can be effectively introduced into the RAN, and an efficient RAN is implemented. For example, the efficiency of network planning, network configuration, and / or resource scheduling, etc. can be improved. When AI is introduced into the RAN through the method, the existing network can be better compatible, and new AI functions can be easily introduced.

[0082] In a possible implementation, the method further includes: sending, to a base station, third task configuration information, the third task configuration information being used to indicate configuration information of one or more tasks, and the execution subject of each of the one or more tasks indicated by the third task configuration information including a centralized unit (CU) of the base station or a distributed unit (DU) of the base station.

[0083] In a possible implementation, the configuration information of each of the one or more tasks indicated by the third task configuration information is used to indicate one or more of the following of the task: a task identifier (ID), a task type, task content, a task execution subject, and a task state.

[0084] Through the above method, the RIC can send an AI task to a base station, which is used to implement an AI function in a RAN, so that AI can be effectively introduced into the RAN, and an efficient RAN is implemented. For example, the efficiency of network planning, network configuration, and / or resource scheduling, etc. can be improved. When AI is introduced into the RAN through the method, the existing network can be better compatible, and new AI functions can be easily introduced.

[0085] For a description of the task type, task content, and task status, please refer to the description in the first section; it will not be repeated here.

[0086] In one possible implementation, the method further includes receiving a first interface establishment request message from the CU of the base station. Specific information and beneficial effects can be found in the description of the first aspect, and will not be repeated here.

[0087] In one possible implementation, the method further includes receiving a second interface establishment request message from the DU of the base station. Specific information and beneficial effects can be found in the description of the first aspect, and will not be repeated here.

[0088] In one possible implementation, the method further includes: receiving information about the terminal device from the base station, wherein the information about the terminal device includes one or more of the following: capability information of the terminal device, configuration information of the terminal device, and status information of the terminal device. Optionally, receiving the information about the terminal device from the base station includes: receiving the information about the terminal device from a CU (CU) of the base station.

[0089] This method allows us to obtain terminal information, which can then be used for AI model training and / or inference in the RIC.

[0090] In one possible implementation, the one or more tasks include at least one data collection task, and the method further includes: receiving collected data from the base station. Receiving the collected data from the base station includes: receiving the collected data from the CU of the base station. Optionally, when the executing entity of the at least one data collection task is the base station, the data is collected by the base station. Optionally, when the executing entity of the at least one data collection task is the CU, the data is collected by the CU. Optionally, when the executing entity of the at least one data collection task is the DU, the data is collected by the DU. The data collected by the DU may be sent from the DU to the CU, and then sent by the CU to the RIC. Optionally, when the executing entity of the at least one data collection task is the CU-CP, the data is collected by the CU-CP. Optionally, when the executing entity of the at least one data collection task is the CU-UP, the data is collected by the CU-UP. The data collected by the CU-UP may be sent from the CU-UP to the CU-CP, and then sent by the CU-CP to the RIC.

[0091] This method allows you to obtain the data you want to collect, which can then be used to assist in model training and / or inference functions.

[0092] In one possible implementation, the method further includes: publishing the inference result to the base station. Publishing the inference result to the base station includes: publishing the inference result to the CU of the base station.

[0093] This method allows inference results to be published to the base station. For example, after collecting data, the RIC can use this data for inference to obtain the parameter configuration on the RAN side. The RIC can then publish these parameter values ​​to the base station, enabling the base station to update the corresponding parameter values ​​and thereby improve network performance.

[0094] In one possible implementation, the one or more tasks indicated by the second task configuration information include at least one data collection task, and the method further includes: receiving collected data from the terminal device through the first protocol layer. This method allows the acquisition of desired data, which can then be used to assist in model training and / or inference functions.

[0095] In one possible implementation, the one or more tasks indicated by the second task configuration information include at least one model training task. The method further includes receiving model parameter information or model parameter gradient information from the terminal through the first protocol layer. This method enables federated learning on the terminal side, improving the efficiency of AI model training.

[0096] In one possible implementation, the first protocol layer is an Artificial Intelligence Control (AIC) layer above the Packet Data Convergence Protocol (PDCP) layer, the first protocol layer is an AIC layer above the Radio Resource Control (RRC) layer, or the first protocol layer is an application layer.

[0097] The first protocol layer is an Artificial Intelligence Control (AIC) layer above the Packet Data Convergence Protocol (PDCP) layer, which includes: at the sending end, the data of the AIC layer is sequentially delivered to the PDCP layer, RLC layer, MAC layer and physical layer; after the physical layer at the receiving end receives the data, it is sequentially delivered to the MAC layer, RLC layer, PDCP layer and AIC layer.

[0098] The first protocol layer is the AIC layer above the Radio Resource Control (RRC) layer, including: at the transmitting end, the data of the AIC layer is sequentially delivered to the RRC layer, PDCP layer, RLC layer, MAC layer and physical layer; after the physical layer at the receiving end receives the data, it is sequentially delivered to the MAC layer, RLC layer, PDCP layer, RRC layer and AIC layer.

[0099] This method allows for the introduction of a new protocol layer for task publishing and / or the publication of models through the application layer, thereby achieving both effectiveness and scalability in task publishing.

[0100] Eleventhly, a communication method is provided. The executing entity of this method can be a base station, a CU, or a device capable of supporting the base station or CU to perform this function, such as a chip, etc., without limitation. The method includes: receiving third task configuration information from a wireless intelligent control RIC, the third task configuration information indicating configuration information for one or more tasks, wherein the executing entity of each of the one or more tasks indicated by the third task configuration information includes the centralized unit CU or the distributed unit DU of the base station.

[0101] For a detailed introduction to task configuration information, please refer to the description in Section 10, which will not be repeated here.

[0102] In one possible implementation, for at least one of the one or more tasks, the executing entity of the at least one task is the DU, and the method further includes sending the content of each of the at least one task to the DU.

[0103] In one possible implementation, for at least one of the one or more tasks, the execution entity of the at least one task is the CU-UP, and the method further includes: sending the content of each of the at least one task to the CU-UP.

[0104] In one possible implementation, the method further includes sending a first interface establishment request message to the RIC. Specific information and beneficial effects can be found in the description of aspect ten, and will not be repeated here.

[0105] In one possible implementation, the method further includes: at least one data collection task among the one or more tasks, and the method further includes: sending the collected data to the RIC. Optionally, the data is received from the DU of the base station or from the CU-UP of the base station.

[0106] In one possible implementation, the method further includes receiving inference results from the RIC.

[0107] In a twelfth aspect, a communication method is provided. The executing entity of this method can be a terminal device or a device capable of supporting the terminal device in implementing this function, such as a chip, and is not limited thereto. The method includes: receiving second task configuration information from a wireless intelligent control (RIC) through a first protocol layer, the second task configuration information indicating configuration information for one or more tasks; wherein the executing entity of each of the one or more tasks includes one or more terminal devices.

[0108] For a detailed introduction to task configuration information, please refer to the description in Section 10, which will not be repeated here.

[0109] In one possible implementation, the method further includes: sending terminal device information to a base station, wherein the terminal device information includes one or more of the following: capability information of the terminal device, configuration information of the terminal device, and status information of the terminal device.

[0110] In one possible implementation, the one or more tasks include at least one data collection task, and the method further includes: sending the collected data to the RIC via the first protocol layer.

[0111] In one possible implementation, the one or more tasks include at least one model training task, and the method further includes: sending model parameter information or model parameter information gradient information to the RIC through the first protocol layer.

[0112] In one possible implementation, the method further includes: receiving inference results from the RIC via the first protocol layer.

[0113] In one possible implementation, the first protocol layer is an Intelligent Control (AIC) layer above the Packet Data Convergence Protocol (PDCP) layer, or an AIC layer above the Radio Resource Control (RRC) layer, or an application layer.

[0114] In a thirteenth aspect, an apparatus is provided, which may be a reconfigurable integrated circuit (RIC) or other apparatus capable of implementing the methods described in the tenth aspect. This other apparatus may be installed in or used in conjunction with an RIC. In one design, the apparatus may include modules corresponding to each of the methods / operations / steps / actions described in the tenth aspect; these modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the apparatus may include a processing module and a communication module.

[0115] In one possible implementation, the processing module utilizes the communication module to send second task configuration information to the terminal device through a first protocol layer. The second task configuration information is used to indicate the configuration information of one or more tasks. The execution entity of each of the one or more tasks includes one or more terminal devices.

[0116] In one possible implementation, the communication module is used to send third task configuration information to the base station. This third task configuration information indicates configuration information for one or more tasks, wherein the execution entity of each of the one or more tasks indicated by the third task configuration information includes a centralized unit (CU) or a distributed unit (DU) of the base station. The third task configuration information is generated by the processing module.

[0117] For other information that the communication module can receive and / or send, please refer to the description in Section 10, which will not be repeated here.

[0118] In a fourteenth aspect, an apparatus is provided, which may be a base station or other apparatus capable of implementing the methods described in the eleventh aspect. The other apparatus may be installed in a base station or used in conjunction with a base station. In one design, the apparatus may include modules corresponding to each of the methods / operations / steps / actions described in the eleventh aspect; these modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the apparatus may include a processing module and a communication module.

[0119] In one possible implementation, the communication module is configured to receive third task configuration information from the wireless intelligent controller (RIC). This third task configuration information indicates configuration information for one or more tasks, wherein the execution entity of each of the one or more tasks indicated by the third task configuration information includes either a centralized unit (CU) or a distributed unit (DU) of the base station. The processing module is configured to receive the third task configuration information from the communication module and process it.

[0120] For other information that the communication module can receive and / or send, please refer to the description in Section 11, which will not be repeated here.

[0121] In a fifteenth aspect, an apparatus is provided, which may be a terminal device or other apparatus capable of implementing the methods described in the twelfth aspect. The other apparatus may be installed in or used in conjunction with a terminal device. In one design, the apparatus may include modules corresponding to each of the methods / operations / steps / actions described in the twelfth aspect; these modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the apparatus may include a processing module and a communication module.

[0122] In one possible implementation, the processing module is used to utilize the communication module to receive second task configuration information from the wireless intelligent control RIC via a first protocol layer. The second task configuration information is used to indicate the configuration information of one or more tasks. The execution entity of each of the one or more tasks includes one or more terminal devices.

[0123] For other information that the communication module can receive and / or send, please refer to the description in Section Twelve, which will not be repeated here.

[0124] In a sixteenth aspect, embodiments of this application provide an apparatus including a processor for implementing the method described in the tenth aspect above. The apparatus may further include a memory for storing instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, it can implement the method described in the tenth aspect above. The apparatus may also include a communication interface for communicating with other devices. In embodiments of this application, the communication interface may be a transceiver, circuit, bus, module, pin, or other type of communication interface.

[0125] In one possible design, the device includes:

[0126] Memory, used to store program instructions;

[0127] The processor is configured to send second task configuration information to a terminal device via a communication interface and a first protocol layer. The second task configuration information is used to indicate configuration information for one or more tasks. The execution entity of each of the one or more tasks includes one or more terminal devices.

[0128] In one possible implementation, the processor uses a communication interface to send third task configuration information to the base station. The third task configuration information is used to indicate the configuration information of one or more tasks. The execution entity of each of the one or more tasks indicated by the third task configuration information includes the centralized unit (CU) or the distributed unit (DU) of the base station.

[0129] For other information that the processor can receive and / or send using the communication interface, please refer to the description in Section 10, which will not be repeated here.

[0130] In a seventeenth aspect, embodiments of this application provide an apparatus including a processor for implementing the method described in the eleventh aspect above. The apparatus may further include a memory for storing instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, it can implement the method described in the eleventh aspect above. The apparatus may further include a communication interface for communicating with other devices.

[0131] In one possible design, the device includes:

[0132] Memory, used to store program instructions;

[0133] The processor is configured to receive third task configuration information from a wireless intelligent control (RIC) via a communication interface. The third task configuration information is used to indicate configuration information for one or more tasks, wherein the execution entity of each of the one or more tasks indicated by the third task configuration information includes the centralized unit (CU) or the distributed unit (DU) of the base station.

[0134] For other information that the processor can receive and / or send using the communication interface, please refer to the description in Section 11, which will not be repeated here.

[0135] Eighteenthly, embodiments of this application provide an apparatus including a processor for implementing the method described in the twelfth aspect above. The apparatus may further include a memory for storing instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, it can implement the method described in the twelfth aspect above. The apparatus may further include a communication interface for communicating with other devices.

[0136] In one possible design, the device includes:

[0137] Memory, used to store program instructions;

[0138] The processor is configured to receive second task configuration information from the wireless intelligent controller (RIC) via a first protocol layer using a communication interface. The second task configuration information is used to indicate configuration information for one or more tasks. The execution entity of each of the one or more tasks includes one or more terminal devices.

[0139] For other information that the processor can receive and / or send using the communication interface, please refer to the description in Section Twelve, which will not be repeated here.

[0140] Nineteenthly, a communication method is provided. The executing entity of this method can be a base station, a CU, or a device capable of supporting the base station or CU to perform this function, such as a chip, without limitation. The method includes: sending information about one or more tasks to a terminal device through the application layer, the Artificial Intelligence Control (AIC) layer, Radio Resource Control (RRC) signaling, System Information Block (SIB), Master Information Block (MIB), paging messages, Media Access Control (MAC) control elements (CE), or physical layer information. The AIC layer is located above the Packet Data Convergence Protocol (PDCP) layer, or the AIC layer is located above the RRC layer. Each of the one or more tasks is executed by one or more terminal devices.

[0141] Using the methods described above, AI functions can be implemented in the RAN, such as in base stations, thereby effectively introducing AI into the RAN and achieving a highly efficient RAN. For example, it can improve the efficiency of network planning, network configuration, and / or resource scheduling. When introducing AI into the RAN using this method, it can better maintain compatibility with existing networks and facilitate the introduction of new AI functions.

[0142] In one possible implementation, for each of the one or more tasks, the task information is used to indicate one or more of the following task contents: task identifier ID, task type, task content, task execution entity, and task status. A detailed description of this method can be found in the first aspect, and will not be repeated here.

[0143] In one possible implementation, the method further includes: receiving information from the terminal device, wherein the information of the terminal device includes one or more of the following: capability information of the terminal device, configuration information of the terminal device, and status information of the terminal device.

[0144] This method allows us to obtain information about the terminal device, which can then be used for AI model training and / or inference at the base station.

[0145] In one possible implementation, the one or more tasks include at least one data collection task, and the method further includes receiving the collected data from the terminal device.

[0146] In one possible implementation, the method further includes sending the inference result to the terminal.

[0147] In one possible implementation, the one or more tasks include at least one model training task, and the method further includes receiving gradient information of model parameters from the terminal device.

[0148] In a twentieth aspect, a communication method is provided. The executing entity of this method can be a terminal device or a device capable of supporting the terminal device in implementing this function, such as a chip, without limitation. The method includes: receiving information for one or more tasks from a base station through an application layer, an Artificial Intelligence Control (AIC) layer, Radio Resource Control (RRC) signaling, system messages, a Master Information Block (MIB), paging messages, MAC CE, or physical layer information, wherein the AIC layer is located above the Packet Data Convergence Protocol (PDCP) layer, or the AIC layer is located above the RRC layer; wherein the executing entity for each of the one or more tasks is one or more terminal devices.

[0149] In one possible implementation, the method further includes: sending terminal device information to the base station, wherein the terminal device information includes one or more of the following: terminal device capability information, terminal device configuration information, and terminal device status information.

[0150] In one possible implementation, the one or more tasks include at least one data collection task, and the method further includes: sending the collected data to the base station.

[0151] In one possible implementation, the method further includes receiving inference results from the base station.

[0152] In one possible implementation, the one or more tasks include at least one model training task, and the method further includes sending gradient information of model parameters to the base station.

[0153] In a twenty-first aspect, an apparatus is provided, which may be a base station or other apparatus capable of implementing the methods described in aspect nineteen. The other apparatus may be installed in a base station or used in conjunction with a base station. In one design, the apparatus may include modules corresponding to each of the methods / operations / steps / actions described in aspect nineteen; these modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the apparatus may include a processing module and a communication module.

[0154] In one possible implementation, the communication module utilizes the processing module to send information about one or more tasks to the terminal device through the application layer, the Artificial Intelligence Control (AIC) layer, the Radio Resource Control (RRC) signaling, the System Information Block (SIB), the Master Information Block (MIB), paging messages, the Media Access Control (MAC) control element (CE), or physical layer information. The AIC layer is located above the Packet Data Convergence Protocol (PDCP) layer, or the AIC layer is located above the RRC layer. Each of the one or more tasks is executed by one or more terminal devices.

[0155] For other information that the communication module can receive and / or send, please refer to the description in Section 19, which will not be repeated here.

[0156] In a twenty-second aspect, an apparatus is provided, which may be a terminal device or other apparatus capable of implementing the methods described in the twenty-fifth aspect. The other apparatus may be installed in or used in conjunction with a terminal device. In one design, the apparatus may include modules corresponding to each of the methods / operations / steps / actions described in the twenty-fifth aspect; these modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the apparatus may include a processing module and a communication module.

[0157] In one possible implementation, the communication module utilizes the processing module to receive information for one or more tasks from the base station via the application layer, the Artificial Intelligence Control (AIC) layer, Radio Resource Control (RRC) signaling, System Information Block (SIB), Master Information Block (MIB), paging messages, MAC CE, or physical layer information. The AIC layer is located above the Packet Data Convergence Protocol (PDCP) layer, or the AIC layer is located above the RRC layer. Each of the one or more tasks is executed by one or more terminal devices.

[0158] For other information that the communication module can receive and / or send, please refer to the description in Section 20, which will not be repeated here.

[0159] In a twentieth aspect, embodiments of this application provide an apparatus comprising a processor for implementing the method described in the nineteenth aspect above. The apparatus may further include a memory for storing instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, it can implement the method described in the nineteenth aspect above. The apparatus may further include a communication interface for communicating with other devices.

[0160] In one possible design, the device includes:

[0161] Memory, used to store program instructions;

[0162] The processor is configured to use a communication interface to send information about one or more tasks to a terminal device via an application layer, an artificial intelligence control (AIC) layer, radio resource control (RRC) signaling, a system information block (SIB), a master information block (MIB), paging messages, a media access control (MAC) control element (CE), or physical layer information. The AIC layer is located above the packet data aggregation layer protocol (PDCP) layer, or the AIC layer is located above the RRC layer. Each of the one or more tasks is executed by one or more terminal devices.

[0163] For other information that the processor can receive and / or send using the communication interface, please refer to the description in Section 19, which will not be repeated here.

[0164] In a twentieth aspect, embodiments of this application provide an apparatus including a processor for implementing the method described in the twentieth aspect. The apparatus may further include a memory for storing instructions. The memory is coupled to the processor, and when the processor executes the instructions stored in the memory, it can implement the method described in the twentieth aspect. The apparatus may also include a communication interface for communicating with other devices.

[0165] In one possible design, the device includes:

[0166] Memory, used to store program instructions;

[0167] The processor is configured to receive information for one or more tasks from a base station via a communication interface through an application layer, an artificial intelligence control (AIC) layer, radio resource control (RRC) signaling, a system information block (SIB), a master information block (MIB), paging messages, MAC CE, or physical layer information, wherein the AIC layer is located above the packet data aggregation layer protocol (PDCP) layer, or the AIC layer is located above the RRC layer; wherein the executing entity of each of the one or more tasks is one or more terminal devices.

[0168] For other information that the processor can receive and / or send using the communication interface, please refer to the description in Section 20, which will not be repeated here.

[0169] In a twenty-fifth aspect, an apparatus is provided, comprising: an Artificial Intelligence Application (AIA) module, a Model and Strategy Information Repository (MPIR), an Operational Status Information Repository (OSIB), and an Artificial Intelligence Process Management (AIPM) module; wherein,

[0170] The AIA module includes at least one application, wherein an application is used to train or update a model using one or more of the following: model information corresponding to the application, policy information of the application, operational status information of the terminal device, and operational status information of the radio access network (RAN).

[0171] The MPIR is used to store the policy information and corresponding model information of each application in the at least one application;

[0172] The OSIB is used to store the operational status information of the terminal device and the operational status information of the RAN;

[0173] The AIPM is used to manage the model training or model update process.

[0174] The above modules can be software modules, hardware circuits, or a combination of software models and hardware circuits, without any restrictions.

[0175] In a twenty-sixth aspect, an apparatus is provided, comprising: an Artificial Intelligence Application (AIA) module, a Model and Strategy Information Repository (MPIR), an Operational Status Information Repository (OSIB), and an Artificial Intelligence Process Management (AIPM) module; wherein,

[0176] The AIA module includes at least one application, wherein an application is used to perform inference using one or more of the following: model information corresponding to the application, policy information of the application, operational status information of the terminal device, and operational status information of the radio access network (RAN).

[0177] The MPIR is used to store the policy information and corresponding model information of each application in the at least one application;

[0178] The OSIB is used to store the operational status information of the terminal device and the operational status information of the RAN;

[0179] The AIPM is used to manage the inference process.

[0180] In the method embodiments of the first to twenty-fourth aspects described above, the entity (e.g., RIC or base station) that publishes task configuration information or task information may include the means of the twenty-sixth and / or twenty-seventh aspects. For example, the acquired information may be stored accordingly in the OSIB. The task publishing process may be triggered or managed by the AIPM.

[0181] The twenty-seventh aspect provides a method comprising: for an application, using one or more of the following for model training or model updating:

[0182] The model information corresponding to an application;

[0183] The policy information of an application;

[0184] Operational status information of terminal equipment; and,

[0185] Operational status information of the Radio Access Network (RAN).

[0186] The above modules can be software modules, hardware circuits, or a combination of software models and hardware circuits, without any restrictions.

[0187] The twenty-eighth aspect provides a method comprising: for an application, reasoning using one or more of the following:

[0188] The model information corresponding to an application;

[0189] The policy information of an application;

[0190] Operational status information of terminal equipment; and,

[0191] Operational status information of the Radio Access Network (RAN).

[0192] In a twenty-ninth aspect, a communication system is provided, comprising:

[0193] The apparatus of the fourth aspect, the apparatus of the fifth aspect, and the apparatus of the sixth aspect;

[0194] The apparatus of the fourth aspect, the apparatus of the fifth aspect, the apparatus of the sixth aspect, and the apparatus of the twenty-first aspect;

[0195] The seventh aspect device, the eighth aspect device, and the ninth aspect device;

[0196] The apparatus of the seventh aspect, the apparatus of the eighth aspect, the apparatus of the ninth aspect, and the apparatus of the twenty-third aspect;

[0197] The apparatus of the seventh aspect, the apparatus of the ninth aspect, and the apparatus of the twenty-third aspect;

[0198] The apparatus of aspect thirteen, aspect fourteen, and aspect fifteen;

[0199] The apparatus of aspect thirteen, aspect fourteen, aspect fifteen, and aspect twenty-one;

[0200] The apparatus of the sixteenth aspect, the apparatus of the seventeenth aspect, and the apparatus of the eighteenth aspect;

[0201] The apparatus of the sixteenth aspect, the apparatus of the seventeenth aspect, the apparatus of the eighteenth aspect, and the apparatus of the twenty-third aspect;

[0202] The apparatus of aspect twenty-one and the apparatus of aspect twenty-two; or,

[0203] The apparatus of aspect twenty-three and aspect twenty-four.

[0204] In a thirtieth aspect, a computer-readable storage medium is provided, including instructions that, when executed on a computer, cause the computer to perform the method described in any of the above-described method embodiments.

[0205] In a thirty-first aspect, a computer program product is provided, including instructions that, when run on a computer, cause the computer to perform the method described in any of the method embodiments.

[0206] In a thirty-second aspect, a chip system is provided, comprising a processor and potentially a memory, for implementing the methods described in any of the above-described method embodiments. The chip system may be composed of chips or may include chips and other discrete devices. Attached Figure Description

[0207] Figures 1(a) and 1(b) show example diagrams of the protocol stack provided in the embodiments of this application;

[0208] Figure 2 The diagram shown is an example of the RAN structure provided in an embodiment of this application;

[0209] Figure 3 The diagram shown is an example of the structure of the gNB provided in this application embodiment;

[0210] Figure 4 The diagram shown is an example of an air interface protocol stack provided in an embodiment of this application.

[0211] Figure 5 The diagram shown is a structural example of the RIC module provided in an embodiment of this application;

[0212] Figures 6(a)-6(c) The diagram shown is an example of a network architecture provided in an embodiment of this application.

[0213] Figures 7(a)-7(d) The diagram shown is an example of a protocol stack provided in an embodiment of this application;

[0214] Figure 8(a) shows an example of the architecture provided in this application embodiment, and Figure 8(b) shows a flowchart of information interaction using the architecture in Figure 8(a).

[0215] Figure 9(a) shows an example of the architecture provided in the embodiment of this application, and Figure 9(b) shows a flowchart of information interaction using the architecture of Figure 9(a).

[0216] Figure 10 The image shown is an example three of the architectures provided in this application.

[0217] Figure 11 The following is an example of the architecture provided in the embodiments of this application;

[0218] Figure 12 and Figure 13 The diagram shown is an example of the device structure provided in an embodiment of this application. Detailed Implementation

[0219] The technical solutions provided in this application can be applied to various communication systems, such as: Long Term Evolution (LTE) systems, 5th generation (5G) mobile communication systems, Wireless-Fidelity (WiFi) systems, future 6th generation mobile communication systems, or systems integrating multiple communication systems, etc. This application does not limit the scope of these systems. 5G can also be referred to as New Radio (NR).

[0220] The technical solutions provided in this application can be applied to various communication scenarios, such as one or more of the following communication scenarios: enhanced mobile broadband (eMBB) communication, ultra-reliable low-latency communication (URLLC), machine type communication (MTC), massive machine type communication (mMTC), device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, vehicle-to-vehicle (V2V) communication, and Internet of Things (IoT), etc.

[0221] The technical solutions provided in this application can be applied to communication between communication devices, especially for communication between communication devices in mobile communication networks. Communication between communication devices can include: communication between network devices and terminal devices, communication between network devices, and / or communication between terminal devices. In this application, the term "communication" can also be described as "transmission," "information transmission," "data transmission," or "signal transmission," etc. Transmission can include sending and / or receiving. In this application, the technical solution is described using communication between network devices and terminal devices as an example. Those skilled in the art can also use this technical solution for communication between other scheduling entities and subordinate entities, such as communication between macro base stations and micro base stations, and / or, for example, communication between a first terminal device and a second terminal device. For example, a scheduling entity can perform radio resource management (RRM) on subordinate entities. In this application, "multiple" can refer to two, three, four, or more types; this application does not impose limitations. "At least one" can refer to one or more types; this application does not impose limitations.

[0222] In this embodiment of the application, communication between the network device and the terminal device includes: the network device sending downlink data, signals or information to the terminal device, and / or the terminal device sending uplink data, signals or information to the network device.

[0223] In this application embodiment, " / " can indicate that the related objects are in an "or" relationship, for example, A / B can mean A or B; "and / or" can be used to describe three relationships between related objects, for example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. A and B can be singular or plural. In this application embodiment, terms such as "first," "second," "A," and "B" can be used to distinguish technical features with the same or similar functions. These terms do not limit the quantity or execution order, and "first" and "second" are not necessarily different. In this application embodiment, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Embodiments or designs described as "exemplary" or "for example" should not be construed as being more preferred or advantageous than other embodiments or designs. The use of "exemplary" or "for example" is intended to present related concepts in a concrete manner for ease of understanding.

[0224] The terminal device involved in the embodiments of this application can also be called a terminal, which can be a device with wireless transceiver capabilities. The terminal can be deployed on land, including indoors, outdoors, handheld, and / or vehicle-mounted; it can also be deployed on water (such as ships); and it can also be deployed in the air (e.g., on airplanes, balloons, and satellites). The terminal device can be user equipment (UE). UE includes handheld devices, vehicle-mounted devices, wearable devices, or computing devices with wireless communication capabilities. For example, the UE can be a mobile phone, tablet computer, or computer with wireless transceiver capabilities. The terminal device can also be a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a wireless terminal in autonomous driving, a wireless terminal in telemedicine, a wireless terminal in a smart grid, a wireless terminal in a smart city, and / or a wireless terminal in a smart home, etc.

[0225] In this application embodiment, the device for implementing the functions of the terminal device can be the terminal device itself; it can also be a device capable of supporting the terminal device in implementing the functions, such as a chip system. This device can be installed in the terminal device or used in conjunction with the terminal device. In this application embodiment, the chip system can be composed of chips, or it can include chips and other discrete components. In the technical solutions provided in this application embodiment, the device for implementing the functions of the terminal device is the terminal device, and the terminal device is a UE (User Equipment) as an example, to describe the technical solutions provided in this application embodiment.

[0226] The network devices involved in this application embodiment include base stations (BS), which can be devices deployed in a radio access network (RAN) capable of communicating with terminal devices. Optionally, the radio access network can also be simply referred to as the access network. Base stations may take various forms, such as macro base stations, micro base stations, relay stations, or access points. The base stations involved in this application embodiment can be base stations in 5G systems, base stations in LTE systems, or base stations in other systems, without limitation. Among them, the base station in the 5G system can also be called a transmission reception point (TRP) or a generation Node B (gNB or gNodeB).

[0227] In this application embodiment, the device for implementing the function of the network device can be a network device itself; it can also be a device capable of supporting the network device in implementing that function, such as a chip system. This device can be installed in the network device or used in conjunction with the network device. In the technical solutions provided in this application embodiment, the device for implementing the function of the network device is a network device, and a base station is used as an example to describe the technical solutions provided in this application embodiment.

[0228] For example, taking a gNB as the base station, the gNB and the UE can communicate using the air interface. The network architecture and / or protocol stack for communication between other types of base stations and UEs are similar to or the same as those for communication between gNB and UE, and will not be described in detail here.

[0229] Figure 1(a) shows an example of the protocol stack used by the gNB and UE for user plane data interaction. This interaction involves the service data adaptation protocol (SDAP) layer, packet data convergence protocol (PDCP) layer, radio link control (RLC) layer, media access control (MAC) layer, and physical layer (PHY) layer on both the gNB and UE sides.

[0230] Figure 1(b) shows an example of the protocol stack used when the gNB and UE interact with each other in the control plane. This interaction involves the radio resource control (RRC) layer, PDCP layer, RLC layer, MAC layer, and PHY layer on both the gNB and UE sides.

[0231] In this embodiment, the RRC layer can be used to control air interface radio resources and air interface connections. The SDAP layer can be used to map quality of service (QoS) flows to data radio bearers (DRBs). Here, the QoS flow refers to a service data flow with specific QoS requirements.

[0232] In this embodiment of the application, when the control plane and user plane include protocol layers with the same name, such as PDCP layer, RLC layer, MAC layer or PHY layer, for the network side (e.g. gNB side) or UE side, it means that the corresponding protocol layer supports both user plane functions and control plane functions.

[0233] The base station is part of the RAN (Radio Access RAN) and is used for wireless communication with the UE (User Equipment). For example, Figure 2 The diagram illustrates a possible RAN architecture (e.g., in a 5G system). Optionally, the RAN in a 5G system can be referred to as a next-generation radio access network (NG-RAN). Figure 2As shown, the RAN can communicate or exchange data with the core network (CN) through the NG interface. Taking the base station name as gNB as an example, the RAN can include one or more gNBs. Different gNBs can communicate or exchange data through the Xn-C interface. For any gNB, it can be an integrated gNB, that is, a complete module, entity, network element, or device; or it can include multiple modules, entities, network elements, or devices. For example, a gNB can include two parts: a central unit (CU) and a distributed unit (DU). This design can be called CU and DU separation, or CU / DU separation. The CU of a gNB can also be referred to as gNB-CU, and the DU of a gNB can also be referred to as gNB-DU. The CU and DU of a gNB can communicate, exchange data, or exchange information through the F1 interface. A gNB can include one or more CUs. A gNB can include one or more DUs. A DU can connect to a CU. A CU can connect to one or more DUs. For other gNBs, the core network, and / or UEs, the components of a gNB can be considered as a single gNB. For example, if a gNB consists of a CU and a DU, then for other gNBs, the core network, and / or UEs, the CU and DU of that gNB can be considered as a whole.

[0234] In the embodiments of this application, the interface between any two network elements or any two entities in the RAN can be wired or wireless. That is, the interface can be a wired interface, such as an optical fiber or cable, or it can be a wireless interface. This application embodiment does not impose any restrictions. Different interfaces can have the same or different forms, which is not limited.

[0235] In this embodiment, the interface between any two network elements or entities in the RAN is used for the exchange of data or information between the two network elements or entities. This embodiment does not limit the name of the interface; for example, the interface can be called the z-th interface, where z is a positive integer. The value of z is different for different interfaces.

[0236] Optionally, for a CU in a gNB, the CU can be a complete module, entity, network element, or device, or the CU can include multiple modules, entities, network elements, or devices. For example, the CU can include a CU-CP (control plane) and a CU-UP (user plane). This design can be referred to as CP and UP separation, or CP / UP separation. The CU-CP of a gNB can also be referred to as gNB-CU-CP, and the CU-UP of a gNB can also be referred to as gNB-CU-UP. For example, Figure 3The diagram shown is an example of the structure of gNB. Figure 3 As shown, a gNB can include one CU-CP. A gNB can include one or more CU-UPs. A CU-UP can connect to one CU-CP, and a CU-CP can connect to one or more CU-UPs. The interface between a CU-UP and a CU-CP can be called an E1 port. A DU can connect to one CU-CP, and a CU-CP can connect to one or more DUs. The interface between a DU and a CU-CP can be called an F1-C port. A DU can connect to one or more CU-UPs. A CU-UP can connect to one or more DUs. The interface between a DU and a CU-UP can be called an F1-U port.

[0237] Figure 4 The diagram shows an example of the air interface protocol stack on the gNB side when the CP / UP is separated. Figure 4 As shown, the two RLC layers in the DU implement the control plane function and the user plane function, respectively; the MAC layer and PHY layer in the DU can implement the control plane function and the user plane function simultaneously. The RRC layer and the control plane PDCP layer are located in the CU-CP, and the SDAP and the user plane PDCP layer are located in the CU-UP.

[0238] The new demands, scenarios, and characteristics of mobile communication networks have brought unprecedented challenges to network planning, operation, and maintenance. At this juncture, relying solely on human experience or simple algorithms may not result in efficient mobile communication network operation. For example, network planning, network configuration optimization, and / or resource scheduling may suffer from one or more of the following drawbacks: long processing times, high costs, poor adaptability of optimization algorithms, and poor adaptability of scheduling algorithms. Consequently, it may be unable to meet the new challenges of mobile communication networks.

[0239] To address the aforementioned issues, artificial intelligence (AI) technology can be introduced into mobile communication networks. The goal of AI technology is to enable machines to learn, accumulate experience, and solve problems that humans can solve through experience, such as natural language understanding, image recognition, and / or chess playing. The AI ​​technology involved in this application primarily pertains to machine learning. Machine learning can be considered a method that endows machines with the ability to perform functions that cannot be accomplished through direct programming alone. In machine learning, machines can train or learn from training data to obtain an AI model. This AI model can be used to predict test samples and obtain prediction results. In this application, the AI ​​model may be simply referred to as a model, machine learning (ML) model, AI / ML model, AI network, or other names, and this application does not impose any limitations. In this application, using a model for prediction can also be referred to as using a model for reasoning, etc., and this application does not impose any limitations.

[0240] To enable RAN to meet new challenges, AI technology can be introduced to improve the efficiency of network planning, configuration, and / or resource scheduling, thereby achieving network intelligence and high-efficiency RAN. For example, taking the Multiple Input Multiple Output (MIMO) algorithm in RAN as an example, commonly used MIMO algorithms mainly involve linear matrix operations and are based on assumptions such as Gaussian distribution. However, real-world channel environments are complex and variable, and these commonly used algorithms have limited adaptability to these environments; for example, they cannot simulate complex nonlinear environments and struggle to reach theoretical performance limits. AI technology, however, can simulate nonlinear models, thus effectively adapting to real-world channel environments and approaching performance limits. For instance, using AI technology, machines can obtain training data and use machine learning algorithms and this training data to train models. The trained models can then be used to infer inference data to obtain inference results. For example, AI technology can predict or infer the amount of service data over a future period.

[0241] Based on the above considerations, how to introduce AI technology into the RAN is a key issue that needs to be addressed. In RAN-related protocols, including but not limited to those related to the 3rd Generation Partnership Project (3GPP), new functions can be introduced into the RAN through patching to meet new network optimization needs. However, this approach suffers from poor scalability.

[0242] To introduce easily extensible AI functionality into the RAN, this application provides the following four parts: Part 1: A functional module for implementing AI functionality; Part 2: The network architecture when this functional module is applied to the RAN; Part 3: A protocol stack for implementing AI functionality; Part 4: A communication method between the RAN and the UE for implementing AI functionality. These will be described in detail below.

[0243] The methods provided in this application are not limited to specific machine learning algorithms, and may include supervised learning, unsupervised learning, neural networks, and / or reinforcement learning. Supervised learning may include one or more of the following specific algorithms: Support Vector Machine (SVM), Decision Tree, Naive Bayesian classification, and k-nearest neighbor (KNN). Unsupervised learning may include one or more of the following specific algorithms: Principal Component Analysis (PCA), Singular Value Decomposition (SVD), and k-means clustering.

[0244] Optionally, in this embodiment, the learning of an AI model can be implemented in the RAN. This method can save processing resources on the UE side and reduce power consumption on the UE side.

[0245] Alternatively, in this embodiment, federated learning can be used to learn an AI model. Federated learning can be viewed as a distributed AI training method. In federated learning, the training process of the AI ​​model can be performed on multiple UEs instead of being aggregated to a base station or server on the RAN side, which can save learning time and signaling overhead. In some specific scenarios, federated learning algorithms are beneficial for protecting user privacy. For example, the specific process of federated learning is as follows: a central node (e.g., a base station or other network element in the RAN) sends an AI model to multiple participating nodes (e.g., UEs). The participating nodes train the AI ​​model based on the AI ​​model and the data they have collected or measured, and report their trained AI model to the central node in the form of gradients. The central node processes the gradient information fed back by the participating nodes (e.g., performs averaging or other operations) to obtain a new AI model. Optionally, the central node can send the new AI model to multiple participating nodes, so that the multiple participating nodes can train the AI ​​model based on the AI ​​model and the data they have collected or measured. The central node can obtain an updated AI model based on the gradient information fed back by the participating nodes. In federated learning, the participating nodes may be the same or different in each training process, and this application does not impose any restrictions on this.

[0246] In the embodiments of this application, the learning methods of different AI models can be the same or different, and there is no limitation.

[0247] First part: Introduction of modules for implementing AI functions.

[0248] In this application embodiment, the module used to implement AI functions may be called a radio intelligence control (RIC) module, AI module, intelligent module, machine learning module, or other names, and this application embodiment is not limited to any specific name. For ease of description, the module used to implement AI functions is described as an RIC module. In this application embodiment, AI functions may also be called RIC functions, and AI functions include, but are not limited to, one or more of the following functions: data collection, model download, model training, model updating, model publishing, and inference. The collected data may be used for one or more of the following functions: model training, model updating, and inference. The data used for model training or model updating may also be called training data, training samples, or other names, and this application embodiment is not limited to any specific name. The data used for inference may also be called test samples, prediction samples, or other names, and this application embodiment is not limited to any specific name.

[0249] In this embodiment of the application, in order to implement AI functions in the RAN, information can be transmitted between different network elements. This information can be called AI data, AI information, RIC data, RIC information, or other names, without limitation. For example, the information stored in the RIC module described below can be collectively referred to as AI information.

[0250] A RIC module can be an integrated module or it can consist of multiple modules. In one possible implementation, Figure 5 The diagram shown is an example of the structure of a RIC module, which includes multiple modules.

[0251] The RIC module includes a first module and a second module.

[0252] The first module in the RIC module is used for model training, model updating, and / or inference. For ease of description, model training and model updating will be collectively referred to as model training below. The output of the first module is at least one of the following: information of the trained model, information of the updated model, and inference results. This first module may also be called by other names, and this application embodiment does not impose any limitations. For example... Figure 5 As shown, the first module can be referred to as the Artificial Intelligence Application (AIA) module. For ease of description, this embodiment of the application can be described using the name AIA module as an example.

[0253] In this application embodiment, the AIA module may include one or more application instances, each corresponding to (for implementing or assisting in implementing) one or more network functions. For example, an application instance may correspond to one or more of the following network functions: radio access technology (RAT) selection, load balancing, mobility management, network power saving, coverage optimization, flow control, scheduling, channel coding, or modulation, etc., without limitation in this application embodiment. For example, an application instance may also correspond to multiple network functions, based on a certain strategy, with the goal of achieving comprehensive optimization of multiple network functions. In this application embodiment, application instances may also be collectively referred to as applications, AI application instances, AI applications, RAN applications, or other names, without limitation. In this application embodiment, for an application with a specific function, the specific name of the application is not limited; for example, an application used to implement RAT selection may be called a RAT selection application, or the r-th application, etc., where r is a positive integer, and the value of r may be different for different applications.

[0254] In the embodiments of this application, the differences between different application instances may be: implementing different network functions; or implementing the same type of function for different nodes, such as for different UEs, different base stations, different DUs, different CUs, or one for UEs and another for base stations, etc., without limitation.

[0255] When implementing network functions, applications in the AIA module need to obtain their model information, policy information, and / or AI operation status information, and use this information as input parameters for model training or inference. The acquisition and storage of this information can be implemented by the second module in the RIC module. Here, an application's model can be referred to as the application's corresponding model. An application can be configured to: be able to perform model training but not inference, be able to perform inference but not model training, or be able to perform both model training and inference; this application's embodiments do not impose such limitations.

[0256] For different applications, one or more of the following information will differ: model information, strategy information, AI operational status information, and the target node. One application corresponds to one model. One model can correspond to one application. Alternatively, a module can correspond to multiple different applications simultaneously, but one or more of the following information will differ between applications: strategy information, AI operational status information, and the target node.

[0257] The second module in the RIC module is used to store AI information and / or manage AI-related processes. This second module can also be called by other names, and this application embodiment does not impose any limitations. For example... Figure 5 As shown, the second module can be called the Artificial Intelligence Platform (AIP) module. For ease of description, this embodiment of the application can be described using the name AIP module as an example.

[0258] The AIP module can be an integrated module or it can include multiple functional modules; this application embodiment does not impose any limitations. In one possible implementation, the AIP module includes a first submodule, a second submodule, and a third submodule.

[0259] The first submodule of the AIP module is used to store model information for one or more models. Model information for a model can be obtained by the AIA module and used by the corresponding application within the AIA module for model training and / or inference. The first submodule can also be called by other names, which are not limited in this embodiment, such as a model repository (MR) or a model information repository (MIR). For ease of description, this embodiment can be described using the name MIR for the first submodule. For example, the RIC module obtains model information for one or more models from a CN, network management (also known as Operation, Administration and Maintenance (OAM)), or a third-party application, and this model information can be stored in the MIR. As another example, after the application in the AIA module performs model training or model updates, it stores the model information in the MIR.

[0260] Optionally, the MIR can also be used to store policy information. This policy information can also be referred to as decision information, parameters, conditions, auxiliary information, or other names, without limitation in this embodiment. In this case, the MIR can also be called a model and policy information repository (MPIR) or other names. As described above, for an application, the AIA module can obtain relevant policy information stored in the MPIR for model training or inference. This relevant policy information can also be called the application's policy information. The application's policy information can be used as input parameters for model update training and / or inference. Optionally, the MPIR can store application-specific policy information for each of one or more applications. For example, the MPIR can store information on one or more of the following policies: switching decision policies for mobility load balancing applications, priority information among multiple RATs in a RAT application, and shutdown priority information among multiple RATs in a network energy saving application. Optionally, the MPIR can store common policy information for multiple applications. For example, the MPIR stores conflict resolution policies for multiple application optimization goals, used to resolve conflicts when optimization goals of multiple applications conflict.

[0261] Optionally, the policy information can be stored in a different module than the MIR module within the RIC module; this application embodiment does not impose any restrictions.

[0262] The second submodule of the AIP module is used to store the status information of the UE and / or RAN (e.g., base station). This status information can also be called operational status information, parameters, or other names; this application embodiment does not impose any limitations. The second submodule can also be called other names; this application embodiment does not impose any limitations, such as a status information base (SIB) or an operation status information base (OSIB), etc. For ease of description, such as... Figure 5 As shown, this embodiment of the application can be described using the name OSIB for the second submodule. Exemplarily, the RIC obtains the operational status information of the UE and / or RAN from the CN, OAM, third-party applications, UE, and / or RAN, and this operational status information can be stored in the OSIB. As described above, for an application, the AIA module can obtain the operational status information stored in the OSIB for the application to perform model training and / or inference.

[0263] The operational status information of the UE or RAN may include one or more of the following: AI capability information, AI configuration information, and AI status information.

[0264] AI capability information (capability information):

[0265] The AI ​​capability information of a UE may include one or more of the following: AI application support capability information, AI data collection capability information, and AI result application capability.

[0266] Specifically, the AI ​​application support capability information of the UE includes one or more of the following: the computing power of the UE, the storage capacity of the UE, the AI ​​applications supported by the UE (for example, in the embodiments of this application, they can be represented as application identifier, application name, application index or application type), the AI / ML models supported by the UE (for example, in the embodiments of this application, they can be represented as model identifier, model name, model index or model type), and the AI ​​working modes supported by the UE, etc.

[0267] In this application embodiment, the AI ​​working mode includes one or more of the following: performing AI training, performing AI inference, performing centralized AI training or distributed AI training, and performing centralized AI training or distributed AI training.

[0268] In the embodiments of this application, centralized AI training refers to AI training performed by a single node (such as a UE or base station), while distributed AI training refers to a single node performing a part of an AI training process. For example, a distributed AI training process may include multiple training segments, with each of multiple different nodes performing a portion of the training, and these multiple nodes working together to complete the complete training process. The training processes of these nodes can be sequential, parallel, or a combination of sequential and parallel, without limitation. For example, federated training is a type of distributed training.

[0269] In this embodiment, centralized AI inference refers to a single node performing AI inference, while distributed AI inference refers to a single node performing a part of an AI inference process. For example, a distributed AI inference process may include multiple inference segments, with each of multiple different nodes performing a portion of the inference, and these multiple nodes working together to complete the complete inference process. The inference processes of these nodes can be sequential, parallel, or a combination of sequential and parallel processes, without limitation.

[0270] The UE's AI data collection capability information includes one or more of the following: UE type, UE-supported slice types, UE-supported data collection related measurement types or methods (refer to the relevant descriptions in 3GPP protocol TS37.320, but not limited to the measurement types defined in that protocol), UE-supported positioning methods, UE-supported RAT types (i.e., which RAT data the UE supports collecting), UE-supported clock types, UE-supported clock precision, UE-supported quality of experience (QoE) measurement types or methods, UE-supported power control methods, UE-supported Layer 1 (physical layer) measurement methods, UE-supported frequency band range, UE-supported latency measurement methods, and UE-supported latency measurement granularity or step size, etc.

[0271] The UE's AI result application capability information includes one or more of the following: power saving schemes supported by the UE, mobility schemes supported by the UE, RATs supported by the UE (i.e., which RATs the UE supports applying AI results to), maximum transmit power supported by the UE, and the number of multiple input multiple output (MIMO) layers supported by the UE.

[0272] RAN AI capability information includes one or more of the following: supported network slice types, supported measurement types (refer to the relevant descriptions in 3GPP protocol TS37.320, but not limited to the measurement types defined by that protocol), supported location methods, supported RAT types, clock type, clock accuracy, supported measurement types when performing QoE measurements, supported Layer 1 measurement methods, supported power control methods, frequency band range, frequency, delay measurement type, and delay measurement granularity. Layer 1 measurement methods include: whether beam-level measurements, such as RSRP and / or RSRQ measurements, are supported; and / or whether zero-power interference measurements are supported. Delay measurement types include: RLC / MAC segmented delay measurement, and / or total delay measurement. Delay measurement granularity includes: packet, QoS flow, and / or bearer (RB).

[0273] In the embodiments of this application, the RAT can be Global System for Mobile Communications (GSM), Universal Mobile Telecommunications System (UMTS), LTE, 5G, WiFi, or Bluetooth, etc., and this application embodiment is not limited thereto. The types of RATs supported by the RAN or UE include at least one of these multiple RATs.

[0274] AI configuration information (configuration information):

[0275] The AI ​​configuration information of a UE may include one or more of the following information configured for the UE: AI application (e.g., in this embodiment, it may be represented as a configured application identifier, application name, application index, or application type), AI / ML model corresponding to the AI ​​application (e.g., in this embodiment, it may be represented as a model identifier, model name, model index, or model type), AI working mode, power saving scheme, service parameters (e.g., in this embodiment, it may include data radio bearer (DRB) configuration, protocol data unit (PDU) session configuration, etc.), serving base station, and / or serving cell, etc.

[0276] The RAN's AI configuration information may include one or more of the following: the network slice type used by the base station, the operating frequency of subordinate cells, the bandwidth of subordinate cells, the downlink transmit power of the synchronization signal and PBCH block (SSB) of subordinate cells, the SSB period of subordinate cells, the private network configuration in the RAN, the RAN sharing related configuration, and information about nodes in the RAN (such as WiFi access points (APs), non-standalone base stations, small cells, IAB nodes, and / or relay nodes). SSB can also be replaced with synchronization signal, broadcast channel, or downlink reference signal, etc.

[0277] AI status information (status information):

[0278] When the AIA module is used for model training, the corresponding AI state information can be considered as training data. When the AIA module is used for inference, the corresponding AI state information can be considered as test data, prediction data, or inference data.

[0279] The AI ​​status information of a UE may include one or more of the following: service status information, resource usage status information, radio channel status information, location information, moving speed, movement trajectory, UE's RRC connection status, and user / UE preference information, etc.

[0280] In this application embodiment, the service status information may include one or more of the following: running service, average data rate, average air interface transmission latency, packet loss rate, QoE satisfaction level, and service model (traffic pattern), etc.

[0281] In this application embodiment, the resource usage status information may include one or more of the following: the computing resources used, the storage resources used, and the percentage of air interface resources used, etc.

[0282] In this embodiment, the wireless channel state information may include one or more of the following: the reference signal received power (RSRP) of the serving cell measured by the UE, the reference signal received quality (RSRQ) of the serving cell measured by the UE, the uplink reception interference measured by the base station or the integration of access and backhaul (IAB) node, and the downlink shared spectrum resource conflict probability measured by the base station or the IAB node, etc. A UE may have one or more serving cells, and this embodiment does not impose any limitation. The UE-related uplink reception interference may be measured at the network side.

[0283] In this embodiment, user / UE preferences include one or more of the following: power saving mode preference, RAT selection preference, operator selection preference, and AI operating mode preference, etc. User / UE preference information can be set by the user using the UE. For example, a user can set one or more of the following: power saving mode, operator network selection order, and RAT selection order, etc. The UE or RAN node may set different preferences based on changes in its own device power consumption. Alternatively, user / UE preference information can be stored as policy information in the MPIR; or, a portion of the user / UE preference information (e.g., preference information set in subscription information, as this preference information rarely changes) can be stored in the MPIR, and another portion in the OSIB. For example, user / UE preference information obtained by the RIC from the CN, OAM, or third-party applications is stored in the MPIR, while user / UE preference information reported by the UE or RAN node to the RIC is stored in the OSIB.

[0284] The AI ​​status information of the RAN can include one or more of the following: service status information, resource usage status information, radio channel status information, location information of RAN nodes (such as base stations, IABs, or relay nodes), movement speed, movement trajectory, etc., load of subordinate cells, cell shutdown status of subordinate cells, service QoS guarantee status of subordinate cells, and network slice type used by the base station.

[0285] The third submodule of the AIP module is used to manage or implement one or more of the following AI functions: initialization, model deployment, data collection, model training, inference, and inference result deployment. The third submodule may also be called by other names, which are not limited in this embodiment, such as a procedure management (PM) module, an AI procedure management (AIPM) module, or an AI processing module. For ease of description, this embodiment may use the name AIPM as an example.

[0286] For example, the AIPM can trigger or manage the initialization process of the RIC module. This process can occur during RIC module SETUP (e.g., power-on). At this time, the RIC module can obtain the UE's AI capability information and AI configuration information from the CN, UE, and / or base station. The RIC module can obtain the RAN's AI capability information and AI configuration information from the CN and / or base station. Optionally, when the UE's AI capability information or AI configuration information changes, the UE can inform the RIC module of the updated AI capability information or AI configuration information through the base station or CN. When the RAN's AI capability information or AI configuration information changes, the RAN can inform the RIC module of the updated AI capability information or AI configuration information through the base station or CN. The RIC module can store the obtained AI capability information and AI configuration information in the OSIB.

[0287] For example, the AIPM can trigger or manage the data collection process. This process can occur while the RIC module is running. The AIPM can trigger the RIC module to periodically, event-triggered, and / or upon request, obtain AI status information from the UE and / or RAN. Optionally, the UE and / or RAN can be made aware of the type of AI status information that the RIC module wants to collect through a protocol agreement or by the RIC module issuing a data collection task. The RIC module can store the obtained AI status information in the OSIB.

[0288] For example, AIPM can trigger or manage the model publishing function. This process can occur while the RIC module is running. It is used by the RIC module to publish model information to the base station and / or UE. This model information is stored in MPIR.

[0289] For example, the AIPM can trigger or manage the model training (including model update) process. This process can occur while the RIC module is running. For instance, for an application, the AIPM can trigger the AIA module to obtain one or more of the following information used by the application during model training from the RIC module: model information, policy information, and UE and / or RAN operational status information. The application trains the model based on this information. The AIA module can store the trained model information or the updated model information in the aforementioned MPIR.

[0290] For example, the AIPM can trigger or manage the inference process. This process can occur during the runtime of the RIC module. For instance, for an application, the AIPM can trigger the AIA module to obtain one or more of the following information used by the application for inference from the RIC module: model information, policy information, and operational status information of the UE and / or RAN. The application performs inference based on this information. Optionally, the AIPM can trigger an inference result publication process to publish the application's inference results to the RAN and / or UE. The inference results may also be published to the CN, OAM, and / or third-party applications, which is not limited in this embodiment.

[0291] Optionally, any two of the first, second, and third submodules described above can be combined into one submodule. For example, the first and second submodules can be combined into one submodule. The information stored in the first and second submodules can be collectively referred to as AI information. This AI information includes, but is not limited to, the aforementioned related information.

[0292] Second part: Network architecture when RIC modules are applied to RAN.

[0293] In this application embodiment, the RIC module can be a physical network element or a functional module, and this application embodiment does not impose any limitations. The functional module can be a software module, a hardware circuit, or a combination of software and hardware circuits. When the RIC module is a physical network element, a hardware circuit, or a combination of software and hardware circuits, the RIC module can also be called a radio intelligence controller (RIC). For ease of description, in this application embodiment, the RIC module can also be simply referred to as RIC. The RIC module can exist in the RAN but not in the UE, or the RIC can exist in both the RAN and the UE. The RIC module will be described below from both the network side and the UE side.

[0294] When the RIC module exists on the network side, it can have Figures 6(a)-6(c) The three network architectures shown are shown.

[0295] Figure 6(a): Independent RIC architecture.

[0296] In this network architecture, the RIC module and the base station are separate network elements. This network architecture can be applied to integrated base stations, CU / DU separate base stations, and CP / UP separate base stations.

[0297] In this architecture, the RIC module is logically a network element or logical entity independent of the base station. For example, the RIC module can be an independent RIC node, RIC network element, AI node, or AI network element, or it can be a software module and / or hardware circuit included in at least one node, which is a node separate from the base station. This application embodiment does not impose any limitations.

[0298] Figure 6(b): Embedded RIC network architecture.

[0299] In this network architecture, the RIC module is an integral part of the base station. This network architecture is applicable to integrated base stations, CU / DU separate base stations, and CP / UP separate base stations.

[0300] For a CU / DU separated base station, the RIC module may be included in the CU but not in the DU; or included in the DU but not in the CU; or may be partially included in the CU and partially included in the DU.

[0301] The RIC module in the CU can be called a non-real-time wireless intelligent control (nrt-RIC) module, module A, CU intelligent module, CU AI module, or other names, and this application embodiment does not impose any limitations. For CP / UP separation scenarios, the nrt-RIC module may be included in the CU-CP but not in the CU-UP; or included in the CU-UP but not in the CU-CP; or it may be partially included in the CU-UP and partially included in the CU-CP, and this application embodiment does not impose any limitations.

[0302] In this embodiment, the nrt-RIC module can be an integrated module or include multiple separate sub-modules, without limitation. For example, the nrt-RIC module includes an AIA module, where applications are used to implement or assist in implementing radio resource management (RRM) functions. RRM functions can be considered network functions or network optimization functions with relatively low real-time requirements, i.e., non-real-time functions. Layer 3 functions can be regarded as RRM functions. The model information and policy information corresponding to the applications in the nrt-RIC module are stored in the MPIR in the nrt-RIC module, and the operational status information to be used by the applications in the nrt-RIC module is stored in the OSIB in the nrt-RIC module. The nrt-RIC module also includes an AIPM module, which is used to manage one or more of the following AI processes of the applications in the AIA module of the nrt-RIC module: data collection, model training, model download, model deployment, inference, and inference result deployment.

[0303] The nrt-RIC module can be a software module, a hardware circuit, or a combination of software and hardware modules; this application does not impose any restrictions on the embodiments thereof.

[0304] The RIC module in DU can be called a real-time wireless intelligent control (real-time RIC, rt-RIC) module, module B, DU intelligent module, DU AI module, or other names. This application embodiment does not impose any restrictions.

[0305] In this embodiment, the rt-RIC module can be an integrated module or include multiple separate sub-modules, without limitation. For example, the rt-RIC module includes an AIA module, where applications are used to implement or assist in implementing Layer 1 and / or Layer 2 functions, which are functions with relatively high real-time requirements, i.e., real-time functions. For example, applications in the AIA module are used to implement or assist in implementing one or more of the following functions: channel status information (CSI) compression, power control, precoding, modulation, and channel coding. The model information and policy information corresponding to the applications in the rt-RIC module are stored in the MPIR of the rt-RIC module, and the operational status information to be used by the applications in the rt-RIC module is stored in the OSIB of the rt-RIC module. Optionally, the rt-RIC module also includes an AIPM module, which manages the following processes of the applications in the AIA module of the rt-RIC module: data collection, model download, model training, model deployment, and inference result deployment. Alternatively, at least one of the following processes in the AIA module of the rt-RIC module can be managed by the AIPM module in the nrt-RIC module described above: model training, model download, and model deployment.

[0306] In this embodiment of the application, layer 1 is the physical layer, layer 2 is the SDAP layer, PDCP layer, RLC layer and / or MAC layer, and layer 3 is the RRC layer.

[0307] The rt-RIC module can be a software module, a hardware circuit, or a combination of software and hardware modules; the embodiments in this application do not impose any restrictions.

[0308] Figure 6(c): Hybrid RIC architecture.

[0309] In this architecture, some RIC modules and the base station are separate network elements, while another RIC module is an integral part of the base station. This network architecture can be applied to integrated base stations, CU / DU separated base stations, and CP / UP separated base stations. In one possible implementation, as shown in Figure 6(c), the nrt-RIC module is independent of the base station, while the rt-RIC is included in the base station. In another possible implementation, the rt-RIC module is independent of the base station, while the nrt-RIC is included in the base station.

[0310] The RIC module may or may not exist in the UE.

[0311] When the RIC module exists in the UE, it can be a software module, a hardware circuit, or a combination of software and hardware modules; this application embodiment does not impose any limitations. In this case, the UE can implement one or more of the following AI functions: data collection, model download, model training, model updating, model publishing, inference, and inference result publishing.

[0312] Even when the RIC module is not present in the UE, the UE can be triggered to collect data and apply inference results. For example, the UE can report the collected data to the network side for the network to implement AI functions. Another example is that the UE can receive and apply inference results published by the network, and / or can receive and apply network reconfiguration triggered by the inference results.

[0313] In communication systems, the RAN and UE communicate based on a protocol architecture. The following section will introduce the protocol architecture after incorporating AI functionality.

[0314] Part Three: The protocol stack used to implement AI functions.

[0315] When implementing AI functions in RAN or UE, there are... Figures 7(a)-7(d) The diagram shows four protocol stacks. The protocol stack in the RAN corresponds to the protocol stack on the UE side. That is, the RAN and UE can use... Figures 7(a)-7(d) They communicate using the same protocol stack. Each protocol stack is suitable for... Figures 6(a)-6(c) Any network architecture.

[0316] Figure 7(a): RRC layer enhancement.

[0317] In this protocol architecture, the AI ​​functions of the non-real-time network (optimization) function are implemented by the RRC layer. In this embodiment, the non-real-time network (optimization) function can be simply referred to as the non-real-time function, and its specific description is the same as in Part II above, so it will not be repeated here. For example, defining new RRC messages or adding new information elements (IEs) to RRC messages can trigger or manage one or more of the following AI processes of the non-real-time function: data collection, model training, model download, model deployment, inference, and inference result deployment. Furthermore, this method of defining new RRC messages or adding new information elements to RRC messages can also be used to implement one or more of the following AI functions of the real-time network (optimization) function: model training, model download, and model deployment. In this embodiment, the real-time network (optimization) function can be simply referred to as the real-time function, and its specific description is the same as in Part II above, so it will not be repeated here. These newly added RRC messages or newly added information elements can be referred to as nrt-AI data.

[0318] At the sending end, the nrt-AI data in the RRC layer is sequentially submitted to the PDCP layer, RLC layer, MAC layer and physical layer for processing, and then sent from the sending end to the receiving end at the physical layer. After the data is received at the physical layer of the receiving end, it is sequentially submitted to the MAC layer, RLC layer, PDCP layer and RRC layer for processing, so that the receiving end can interpret the nrt-AI data at the RRC layer.

[0319] In this protocol architecture, real-time AI functions are implemented by Layer 1 and / or Layer 2. For example, one or more of the following AI functions are implemented in real time by carrying information through physical layer data channels, physical layer control channels, or MAC control elements (CE): data collection, model training, inference, and inference result publication. This information can be referred to as rt-AI data.

[0320] At the sending end, Layer 1 rt-AI data is sent to the receiving end, where it is deciphered at Layer 1. At the sending end, Layer 2 rt-AI data is submitted to the physical layer for processing and then sent from the physical layer to the receiving end. After the receiving end's physical layer receives the data, it is submitted to Layer 2 for processing, allowing the receiving end to decipher the rt-AI data at Layer 2.

[0321] Figure 7(b): A new artificial intelligence control (AIC) layer that runs parallel to the RRC layer.

[0322] The parallel operation of the AIC layer and the RRC layer can also be described as the AIC layer's data not passing through the RRC layer, or as the AIC layer being above the PDCP layer. Without limitation, there may be no other protocol layers between the AIC layer and the PDCP layer, or other protocol layers may exist, such as those to be introduced in the future.

[0323] In this protocol architecture, the AIC layer is used to implement one or more of the following AI functions for non-real-time functionality: data collection, model training, model download, model deployment, inference, and inference result deployment. Additionally, the AIC layer can also be used to implement one or more of the following AI functions for real-time functionality: model training, model download, and model deployment. The information or data used to implement these two types of functions can be referred to as nrt-AI data or AIC (layer) messages.

[0324] At the sending end, nrt-AI data is sequentially submitted to the PDCP layer, RLC layer, MAC layer and physical layer for processing, and then sent from the sending end to the receiving end at the physical layer. After the receiving end receives the data at the physical layer, the data is sequentially submitted to the MAC layer, RLC layer, PDCP layer and AIC layer for processing, so that the receiving end can interpret the nrt-AI data at the AIC layer.

[0325] In one possible implementation, the protocol architecture includes one or more of the following AI functions implemented in real-time by Layer 1 and / or Layer 2: data collection, model training, inference, and inference result publishing. The information or data used to implement this function can be referred to as rt-AI data, as described in Figure 7(a).

[0326] In one possible implementation, the protocol architecture utilizes one or more of the following AI functions for real-time functionality by the AIC layer: data collection, model training, inference, and inference result publishing. The information or data used to implement this function can be referred to as rt-AI data or AIC (layer) messages. At the sending end, the rt-AI data is sequentially submitted to the PDCP layer, RLC layer, MAC layer, and physical layer for processing, and then sent from the sending end to the receiving end at the physical layer. Upon receiving the data at the physical layer of the receiving end, the data is sequentially submitted to the MAC layer, RLC layer, PDCP layer, and AIC layer for processing, allowing the receiving end to interpret the rt-AI data at the AIC layer.

[0327] Optionally, in this implementation, to ensure the real-time performance of the rt-AI data, the transmission mode of the PDCP layer and RLC layer of the radio bearer (RB) used to carry the rt-AI data can be configured to transparent mode (TM). For example, when the transmitter sends the rt-AI data, no processing of the interaction information is performed at the PDCP layer and RLC layer; it is directly submitted to the subsequent protocol layer. A detailed introduction to RB is provided below.

[0328] Figure 7(c): The AIC layer above the newly added RRC layer.

[0329] Without limitation, there may be no other protocol layer between the AIC layer and the RRC layer, or there may be other protocol layers, such as those to be introduced in the future.

[0330] In this protocol architecture, the AIC layer, as a newly added control plane protocol layer, sits above the RRC layer. The AIC layer implements one or more of the following non-real-time AI functions: data collection, model training, model download, model deployment, inference, and inference result deployment. Optionally, the AIC layer can also implement one or more of the following real-time AI functions: model training, model download, and model deployment. The information or data used to implement these two types of functions can be called nrt-AI data or AIC (layer) messages. At the sending end, the nrt-AI data is sequentially submitted to the RRC layer, PDCP layer, RLC layer, MAC layer, and physical layer for processing, and then sent from the sending end to the receiving end at the physical layer. After receiving the data at the physical layer of the receiving end, the data is sequentially submitted to the MAC layer, RLC layer, PDCP layer, RRC layer, and AIC layer for processing, so that the receiving end can interpret the nrt-AI data at the AIC layer.

[0331] In this protocol architecture, one or more of the following AI functions are implemented by Layer 1 and / or Layer 2 for real-time functionality: data collection, model training, inference, and inference result publishing. See the corresponding description in Figure 7(a) for details.

[0332] Figure 7(d): Some AI functions are executed by the application layer.

[0333] In this protocol architecture, the application layer is the user plane protocol layer, which can execute the model publishing function. At the sending end, data related to the model publishing function is sequentially submitted to the SDAP layer (optional), PDCP layer, RLC layer, MAC layer, and physical layer for processing, and then sent by the sending end to the receiving end at the physical layer. After the receiving end receives the data at the physical layer, the data is sequentially submitted to the MAC layer, RLC layer, PDCP layer, SDAP layer (optional), and application layer for processing, so that the receiving end can interpret the data related to the model publishing function at the application layer.

[0334] In addition to the model publishing function, other AI functions can adopt the above-mentioned approach. Figures 7(a)-7(c) Any of the following methods. For example:

[0335] The first possible implementation (Figure 7(d) + Figure 7(a)): The model publishing function is executed by the application layer for both real-time and non-real-time functions, and the implementation of other AI functions is the same as described in Figure 7(a) above.

[0336] The second possible implementation (Figure 7(d) + Figure 7(b)): The application layer performs real-time and non-real-time functions, and the model publishes the functions. The implementation of other AI functions is the same as described in Figure 7(b) above.

[0337] The third possible implementation (Figure 7(d) + Figure 7(c)): The application layer performs real-time and non-real-time functions, and the model publishes the functions. The implementation of other AI functions is the same as described in Figure 7(c) above.

[0338] Based on the aforementioned protocol layers, the RAN and UE can exchange AI information to achieve RAN intelligence. When transmitting AI information between the RAN and UE, the AI ​​information can be carried through an RB. For a single RB, parameters or information for each protocol layer corresponding to that RB can be configured.

[0339] In this embodiment, the RB is used to carry data exchanged between the RAN and UE via the air interface. Different types of data can be mapped to different RBs and sent from the sender to the receiver.

[0340] In one possible implementation, the RBs between the RAN and UE include two types: signal radio bearers (SRBs) and data radio bearers (DRBs). SRBs primarily carry control plane data, which passes through the RRC, PDCP, RLC, MAC, and physical layers. DRBs primarily carry user plane data, which passes through the SDAP, PDCP, RLC, MAC, and physical layers. For example, RRC layer messages or non-access stratum (NAS) messages can be carried on SRBs, while data from the application layer can be carried on DRBs. One RB can correspond to one QoS requirement. One or more SRBs and / or one or more DRBs can exist between a UE and the RAN.

[0341] In this embodiment, according to the QoS requirements of AI information, the AI ​​information can be carried in an SRB (e.g., for carrying the nrt-AI data in Figure 7(a), the nrt-AI data in Figure 7(b) (optional), and the nrt-AI data in Figure 7(c)) and / or a DRB (e.g., for carrying the data used for model publishing function in Figure 7(d) and the nrt-AI data in Figure 7(b) (optional)). Alternatively, a new RB can be defined for the AI ​​information, that is, an RB dedicated to carrying AI information can be established between the RAN and the UE.

[0342] AI information (or data) includes public AI information and AI information specific to a particular UE.

[0343] In one possible implementation, a common RB can be established between the RAN and the UE to carry common AI information broadcast or multicast by the RAN to multiple UEs. This common RB can also be called an Artificial Intelligence-Common Radio Bearer (AI-CRB) or other names, which are not limited in this embodiment. The configuration information of the AI-CRB can be notified to the corresponding UE via system information or L1 / L2 / L3 dedicated signaling, where the dedicated signaling is specifically for that UE. For example, the base station sends system information to the UE, which indicates the configuration information of the AI-CRB. The configuration information of the AI-CRB may indicate the radio network temporary identifier (RNTI) used to scramble the PDCCH, and / or the time-frequency location information of the PDCCH (e.g., the search space and / or control resource set (CORESET) of the PDCCH). The PDCCH is used to schedule the AI ​​information carried on the AI-CRB. For example, the base station sends a PDCCH to the UE. The control information on this PDCCH is scrambled using Artificial Intelligence (AI) - Radio Network Temporary Identifier (RNTI). This control information schedules a PDSCH, which carries the configuration information of the AI-CRB.

[0344] For example, when it is necessary to transmit public AI information, such as information used to update AI models for federated learning, an AI-CRB can be configured to carry the public AI information. If there are multiple types of public AI information and thus multiple QoS requirements, multiple AI-CRBs can be configured to carry RIC public data with different QoS requirements respectively.

[0345] One or more AI-CRBs can be established between the RAN and the UE. Each AI-CRB can have corresponding QoS requirements. In this embodiment, the QoS requirements include scheduling priority information.

[0346] For example, a specific RB can be established between the RAN and a UE to carry AI information sent by the RAN to that specific UE. This specific RB can also be called an Artificial Intelligence Data Radio Bearer (AI-DRB) or other names; this application embodiment is not limited to any particular name.

[0347] The AI-DRB can carry the nrt-AI data shown in Figure 7(d) above.

[0348] One or more AI-DRBs can be established between the RAN and a UE. Each AI-DRB can have corresponding QoS requirements.

[0349] The establishment of the RB between the RAN and the UE can be triggered by the RAN or by the RIC instructing the RAN to trigger it; this application embodiment does not impose any restrictions.

[0350] Fourth part: Communication method between RAN and UE.

[0351] Based on the previous introduction, the AI ​​information interaction process between RAN and UE will be described in detail below.

[0352] Example 1: Independent RIC architecture (Figure 6(a)) + RRC layer enhancement (Figure 7(a))

[0353] Figure 8(a) shows an example of the architecture (network architecture + protocol stack) between the base station and the UE. Figure 8(b) shows an example of the information interaction process between the RAN and the UE using the architecture shown in Figure 8(a). In Example 1, there is no independent AIC protocol layer; the non-real-time functions, the AI ​​initialization function of the rt-RIC, and / or the AI ​​model distribution function are performed by the RRC layer.

[0354] In this embodiment of the application, the execution order of each operation in the interaction process is not limited.

[0355] In the architecture shown in Figure 8(a), the RAN contains a RIC module independent of the base station. In this architecture, the RAN's RIC module performs at least one of the following operations: model download, model training, data collection, inference, and publishing inference results.

[0356] When the UE does not support RIC functionality (or can be described as having no RIC module), the RAN's RIC module can request the base station to collect data and / or publish inference results to the base station. The RAN's RIC module can also instruct the UE to collect data or send inference results to the UE through the base station's RRC layer, for example, by sending a new RRC message or a new IE to the UE. The data requested from the UE can be RRC layer data, Layer 2 data, or physical layer data, without limitation. In this embodiment, the model used for inference can be a downloaded original model or an updated model trained based on training data, without limitation. The training data used by the RIC module for model training can be collected from the base station and / or the UE, or obtained from the CN, without limitation. The inference data used by the RIC module for inference can be collected from the base station and / or the UE.

[0357] In this application embodiment, the UE's lack of RIC functionality can be described as the absence of an RIC module in the UE. For ease of description, unless otherwise specified, the RIC module referred to below refers to the RIC module on the RAN side.

[0358] When the UE supports RIC functionality, the RAN's RIC module can request the base station to collect data and / or publish inference results to the base station. The RAN's RIC module can also instruct the UE to collect data, perform federated learning, publish a model (for inference or federated learning), perform inference, or send inference results to the UE via the base station's RRC layer, for example, by sending a new RRC message or a new IE to the UE. The data requested from the UE can be RRC layer data, Layer 2 data, or physical layer data, without limitation. In this embodiment, the model used for inference can be a downloaded original model or an updated model trained based on training data, without limitation. The training data used by the RIC module for model training can be collected from the base station and / or the UE, or obtained from the CN, without limitation. The inference data used by the RIC module for inference can be collected from the base station and / or the UE.

[0359] In this application embodiment, the base station can be an integrated base station or a base station with separate CU / DU; this application embodiment is not limited to either. The following architecture diagram illustrates a base station in the form of separate CU / DU. Optionally, when the CU / DU of the base station is separated, the base station can be a base station with separate CP / UP.

[0360] In this embodiment, when the RIC module and the base station are separated, i.e., when the RIC module is not included in the base station, the base station and the RIC module can communicate through an interface. The interface between the base station and the RIC module includes one or more of the following interfaces: the interface between the integrated base station and the RIC module, the interface between the CU and the RIC module, the interface between the CU-CP and the RIC module, the interface between the CU-UP and the RIC module, and the interface between the DU and the RIC module. In this embodiment, the interface between the base station and the RIC module can be referred to as the G1 interface, or an interface with other names, such as the first interface, etc., and this embodiment is not limited. For ease of description, this embodiment uses the G1 interface as an example. The G1 interface can be a wired connection interface, a wireless connection interface, or an interface with other forms of connection, and this embodiment is not limited. In this embodiment, the wired connection can be through a cable, optical fiber, or other medium, and is not limited.

[0361] For example, based on the architecture shown in Figure 8(a), on the RAN side: the RIC module implements non-real-time AI functions through the RRC layer in the base station, and implements some real-time AI functions through Layer 1 and / or Layer 2 in the base station. Some configuration information of the real-time AI functions can be sent from the base station to the UE through RRC layer signaling. For example, the base station can implement at least one of the following through newly added RRC messages or newly added network elements: instructing the UE to collect RRC layer data, instructing the UE to collect Layer 2 data (such as the data transmission delay between the base station's PDCP layer and the UE's PDCP layer, two peer protocol layers), instructing the UE to collect physical layer data, instructing the configuration information of the RB carrying AI data or AI information, publishing the RRC layer inference results (parameter values) to the UE, publishing the physical layer inference results (parameter values) to the UE, instructing the parameter configuration related to Layer 1 and / or Layer 2 AI functions, publishing model information to the UE, and instructing the UE to perform model training (such as federated training). As another example, the base station can implement at least one of the following through physical layer channels and / or MAC CE: instructing the UE to collect physical layer data and publishing the physical layer inference results (parameter values) to the UE.

[0362] In this embodiment, the RRC layer signaling can be a message carried on the broadcast channel (such as a master information block (MIB)), a system message (such as a system information block (SIB)), or an RRC message. This embodiment does not impose any restrictions.

[0363] For example, on the UE side: accordingly, based on information received from the base station, AI functions for both non-real-time and real-time functions are implemented. For example:

[0364] If the base station instructs the UE to collect RRC layer data, the UE will report the collected data to the base station via RRC messages.

[0365] If the base station instructs the UE to collect physical layer data via an RRC message, the UE will report the collected data to the base station via an RRC message, MAC CE, or physical layer channel.

[0366] If the base station instructs the UE to collect physical layer data via MAC CE, the UE will report the collected data to the base station via MAC CE or the physical layer channel.

[0367] If the base station instructs the UE to collect physical layer data through the physical layer channel, the UE will report the collected data to the base station through MAC CE or the physical layer channel.

[0368] If the base station instructs the UE on the configuration information of the RB carrying AI data or AI information, the UE establishes the RB with the base station based on the configuration information of the RB.

[0369] If the base station publishes the inference results (parameter values) of the RRC layer and / or physical layer to the UE, the UE can apply the inference results to the UE side.

[0370] If the base station instructs the UE on the parameter configuration related to Layer 1 and / or Layer 2 AI functions, the UE sets the parameters related to these functions according to the parameter configuration.

[0371] If the base station publishes model information to the UE, the UE can use the model to perform inference. The UE can apply the inference results and / or report the inference results to the RAN side.

[0372] If the base station instructs the UE to perform federated training, the UE can report the gradient information of the trained model parameters to the RIC module.

[0373] Figure 8(b) shows a flowchart illustrating the information interaction process between the RAN and UE using the architecture shown in Figure 8(a). This mainly includes: the RIC module issuing tasks to the base station, and / or the RIC module issuing tasks to the UE through the base station; and the UE and / or the base station executing the corresponding tasks. Optionally, when the task issued by the RIC module includes data collection, the base station reports the collected data to the RIC module, and / or the UE reports the collected data to the RIC module through the base station. Optionally, after the RIC module trains or updates the model based on the collected data, it can use the model for inference. Optionally, after the RIC module performs inference based on the collected data, it can publish the inference results to the UE and / or the base station. The method shown in Figure 8(b) will be described in detail below.

[0374] S801, the RIC module sends the first task configuration information to the base station; the base station receives the first task configuration information.

[0375] In this method, the first task configuration information is used by the RIC module to publish a new task to the base station, or by the RIC module to publish a new task to the UE through the base station. In this embodiment, the name of the message carrying the first task configuration information is not limited. For example, in this embodiment, the message carrying the first task configuration information can be called the y-th message, the RIC TASKADDITION message, or the RIC TASK ADDITION REQUEST message. Here, y is a positive integer. In this embodiment, the value of y is different for different messages. This embodiment uses the message name RIC TASK ADDITION REQUEST as an example.

[0376] In this application embodiment, a task may also be referred to as an operation, transaction, project, or other name, and this application does not impose any limitations. In this application embodiment, the type of a task may be one of at least two task types. The at least two task types may be at least two of the following: data collection (or data gathering), model deployment, model training, inference, and inference result deployment. In this application embodiment, the type of task may also be referred to as the name of the task, etc., and this application embodiment does not impose any limitations.

[0377] The first task configuration information is used when the RIC module issues a new task to the base station. The base station can then execute the corresponding task according to the instructions in this configuration information. As mentioned earlier, when the RIC module issues a new task to the UE through the base station, the base station sends one or more tasks to the UE in the form of RRC layer signaling, and / or sends one or more tasks to the UE in the form of MAC CE signaling. Accordingly, as described above, the UE can report the collected data to the RIC module through the base station according to the instructions of the base station.

[0378] Optionally, when the first task configuration information can configure multiple tasks (in this case, the actual message sent may include one or more tasks), the RIC TASK ADDITION REQUEST message includes one or more of the information elements (IEs) shown in the first column of Table 1. The second column of Table 1 provides a description of each IE in the first column.

[0379] In this embodiment, an IE in a message may be explicitly included in the message or implicitly indicated by the message; this embodiment does not impose any limitations. This embodiment does not limit the names of each IE in the message; for example, an IE may be replaced with the xth IE, where x is a positive integer. The value of x may be different for different IEs. In this embodiment, when an IE in a table is not included in the corresponding message, the IE configuration information may be agreed upon by the protocol.

[0380] Table 1

[0381]

[0382] The above-mentioned TASK CONFIGURATION INFORMATION message includes one or more of the IEs described in the first column of Table 2.

[0383] Table 2

[0384]

[0385]

[0386] Optionally, S801 may further include: the base station sending first task confirmation information to the RIC module; and the RIC receiving the first task confirmation information.

[0387] The first task confirmation information is used by the base station to confirm the RIC TASKADDITION REQUEST message to the RIC module. In this embodiment, the name of the message carrying the first task confirmation information is not limited. For example, in this embodiment, the message carrying the first task confirmation information can be called the RIC TASK ADDITION RESPONSE message, the y-th message, or other names. Here, y is a positive integer.

[0388] In this embodiment of the application, S801 may be referred to as the task addition process, the task configuration process, or other names.

[0389] Optionally, in order to exchange information between the RIC module and the base station, an interface can be established between the RIC module and the base station. In the embodiments of this application, the interface between the RIC module and the base station has the following possible scenarios:

[0390] Scenario 1: The base station is an integrated base station, and there is an interface between the base station and the RIC module. In this scenario, establishing the interface between the RIC module and the base station includes: establishing the interface between the RIC module and the integrated base station.

[0391] Scenario 2: The base station is a CU / DU separated base station, with interfaces between the CU and the RIC module, and between the DU and the RIC module. In this scenario, establishing the interface between the RIC module and the base station includes establishing an interface between the RIC module and the CU, and establishing an interface between the RIC module and the DU.

[0392] Sub-scenario 1 of Scenario 2: The base station is a CP / UP separated base station, with interfaces between the CU-CP and the RIC module, and between the CU-UP and the RIC module. In this scenario, establishing the interface between the RIC module and the CU includes: establishing the interface between the RIC module and the CU-CP, and establishing the interface between the RIC module and the CU-UP.

[0393] Sub-scenario 2 of Scenario 2: The base station is a CP / UP separated base station. There is an interface between the CU-CP and the RIC module, but no interface between the CU-UP and the RIC module. In this scenario, establishing the interface between the RIC module and the CU includes establishing an interface between the RIC module and the CU-CP. The RIC module and the CU-UP can exchange data through the forwarding function of the CU-CP.

[0394] In this embodiment of the application, forwarding data includes: transparently forwarding the data (without processing the data), or processing the data and then forwarding the processed data.

[0395] Scenario 3: The base station is a CU / DU separated base station. There is an interface between the CU and the RIC module, but no interface between the DU and the RIC module. In this scenario, establishing the interface between the RIC module and the base station includes: establishing an interface between the RIC module and the CU. The RIC module and the DU can exchange data through the forwarding function of the CU.

[0396] Scene 3's sub-scene 1: Same as Scene 2's sub-scene 1.

[0397] Scene 3's sub-scene 2: Same as Scene 2's sub-scene 1.

[0398] For simplicity, Figure 8(b) uses a CU / DU separate base station as an example, where both the CU and DU have interfaces with the RIC module. The methods for other scenarios are similar and will not be elaborated further. As mentioned above, to establish the interface between the CU and the RIC module, the method shown in Figure 8(b) may further include: S802: The CU sends a G1 interface establishment request message to the RIC module. This G1 interface establishment request message may be referred to as a first G1 interface establishment request message or a first interface establishment request message. Optionally, the method in S802 may further include: The RIC module returns a G1 interface establishment confirmation message to the CU. This G1 interface establishment confirmation message may be referred to as a first G1 interface establishment confirmation message or a first interface establishment confirmation message.

[0399] In this embodiment, the G1 interface setup request message is used to establish a connection with the RIC module, and its name is not limited. For example, this message can be called the y-th message, the G1 SETUP REQUEST message, or other names. Here, y is a positive integer.

[0400] In this embodiment, the G1 interface establishment confirmation message is used by the RIC module to confirm the establishment of the connection between the RIC and the RIC, and its name is not limited. For example, this message can be called the y-th message, the G1 SETUP RESPONSE message, the G1 interface establishment response message, or other names. Here, y is a positive integer.

[0401] S802 can be referred to as the G1 interface establishment process between the CU and the RIC module. Through this process, the G1 interface between the RIC module and the CU is established.

[0402] In this embodiment, the CU can report the AI ​​operation status information of the base station to the RIC module via the G1 interface. It can report the AI ​​operation status information of the base station itself, or the AI ​​operation status information related to the CU of that base station. The AI ​​operation status information of the base station is part of the AI ​​operation status information of the RAN. In addition to reporting the AI ​​operation status information of the base station where the CU is located, this operation can optionally also report the AI ​​operation status information of other base stations in the RAN. For example, the CU can control multiple DUs and has subordinate IABs and non-standalone base stations. The CU, the DUs controlled by the CU, the IABs subordinate to the CU, and the non-standalone base stations subordinate to the CU can be considered as a RAN, and the CU can report some or all of the AI ​​operation status information of this RAN.

[0403] For example, the G1 SETUP REQUEST message sent by CU to RIC includes one or more of the IEs shown in the first column of Table 3.

[0404] Table 3

[0405]

[0406]

[0407] Optionally, the aforementioned CU capability information, configuration information, and / or status information may be reported via other messages after the interface between the CU and RIC is established; this embodiment of the application does not impose any restrictions. These other messages may include one or more of the IEs shown in the first column of Table 3, without limitation. In this case, the message type is the type of these other messages.

[0408] Optionally, the method shown in Figure 8(b) further includes: S803: The DU sends a G1 interface establishment request message to the RIC module. This G1 interface establishment request message may be referred to as a second G1 interface establishment request message or other names, or a second interface establishment request message. Optionally, the method in S803 further includes: The RIC module returns a G1 interface establishment confirmation message to the DU. This G1 interface establishment confirmation message may be referred to as a second G1 interface establishment confirmation message or a second interface establishment confirmation message.

[0409] S803 can be referred to as the G1 interface establishment process between the DU and the RIC module. Through this process, the G1 interface between the RIC module and the DU is established.

[0410] The G1 SETUP REQUEST message sent by DU to RIC includes one or more of the IEs shown in the first column of Table 4.

[0411] Table 4

[0412]

[0413] Optionally, the aforementioned DU capability information, configuration information, and / or status information may be reported via other messages after the interface between the DU and RIC is established; this embodiment of the application does not impose any restrictions. These other messages may include one or more of the IEs shown in the first column of Table 4, without limitation. In this case, the message type is the type of these other messages.

[0414] When the base station is an integrated base station, the base station sends a G1 SETUP REQUEST message to the RIC. This message specifically includes the functions of the G1 SETUP REQUEST message shown in Tables 3 and 4, as shown in Table 5.

[0415] Table 5

[0416]

[0417] Optionally, the aforementioned base station capability information, configuration information, and / or status information may be reported via other messages after the interface between the base station and the RIC is established; this embodiment of the application does not impose any restrictions. These other messages may include one or more of the IEs shown in the first column of Table 5, without limitation. In this case, the message type is the type of these other messages.

[0418] The RIC module obtains the AI ​​capability information of the aforementioned base station, or, in other words, after obtaining the AI ​​capability information of the CU and / or DU, it can trigger the AI-CRB establishment process over the air interface. Using this AI-CRB, the base station can send broadcast or multicast data collection messages to the UE via the air interface. One possible implementation of the RIC module triggering the AI-CRB establishment process over the air interface is as follows: the RIC module instructs the CU and / or DU to establish one or more AI-CRBs over the air interface; or the CU and / or DU receive a task from the RIC module instructing all UEs or a group of UEs in the cell to establish one or more AI-CRBs over the air interface when collecting data.

[0419] In this embodiment of the application, for a base station with separate CU and DU, an interface can exist between the CU and DU, allowing for data interaction. Therefore, the method shown in Figure 8(b) may include establishing a connection between the CU and DU. Exemplarily, the interface between the CU and DU is called the F1 interface. The F1 interface establishment process between the CU and DU is used to establish a connection between the CU and DU, after which the CU and DU can exchange information. The F1 interface establishment process between the CU and DU includes: the DU sending an F1 establishment request message to the CU, and the CU returning an F1 establishment response message to the DU. The F1 establishment request message indicates one or more of the following information: message type, DU identifier, and DU subordinate cell list information. The F1 establishment response message indicates one or more of the following information: message type, active cell list, and system information.

[0420] The above describes how the RIC module can send tasks to the base station that will be executed by the UE. In some scenarios, the RIC module can obtain the UE's operational status information through a protocol-defined method, so that the RIC module can determine the UE's tasks. In other scenarios, the RIC module needs to obtain the UE's operational status information through the base station or CN, so that the RIC module can determine the UE's tasks.

[0421] Therefore, optionally, the method shown in Figure 8(b) includes: S804: The CU sends the UE's operational status information to the RIC module. This operational status information may be included in the aforementioned G1 SETUP REQUEST message, or it may be included in another message; this embodiment of the application does not impose limitations. For example, the operational status information may be included in a UE context establishment request message. In this method, if the base station is an integrated base station, the CU can be replaced by the base station. If it is a CP / UP separation scenario, the CU in this method can also be replaced by a CU-CP.

[0422] In this embodiment, the UE context establishment request message is used to trigger the establishment of the corresponding UE's AI context in the RIC, so that the RIC can perform AI-related operations on the UE. Its name is not limited. For example, this message can be called the y-th message, the UEAI CONTEXT SETUP REQUEST message, or other names. Here, y is a positive integer.

[0423] The UE context establishment request message includes one or more of the IEs shown in the first column of Table 6.

[0424] Table 6

[0425]

[0426] Optionally, the UE context establishment request sent by the CU to the RIC module may include information about one or more UEs.

[0427] Optionally, the method of S804 also includes: the RIC module returning a UE context establishment confirmation message to the CU.

[0428] In this embodiment, the UE context establishment confirmation message is used by the RIC to confirm the establishment of the corresponding UE's AI context in the RIC or to determine the aforementioned UE AI CONTEXT SETUP REQUEST message, and its name is not limited. For example, this message can be called the y-th message, the UE AI CONTEXT SETUP RESPONSE message, or other names. Wherein, y is a positive integer.

[0429] S804 can be referred to as the UE context establishment process or the UE AI context establishment process, etc., and the embodiments in this application are not limited thereto.

[0430] Optionally, when the UE's operational status information changes, the method shown in Figure 8(a) further includes S805: the CU sends the updated operational status information of the UE to the RIC module. In this method, if the base station is an integrated base station, the CU can be replaced by the base station. If it is a CP / UP separation scenario, the CU in this method can also be replaced by the CP and / or UP.

[0431] In this embodiment, the message including the aforementioned updated operational status information can be named either a UE context establishment request message or a UE context update request message; this embodiment does not impose any restrictions. The UE context update request message is used to inform the RIC module of the updated operational status information of the UE, and its name is not limited. For example, this message can be called the y-th message, the UE AI CONTEXT MODIFICATION REQUEST message, or other names. Here, y is a positive integer. The message structure of the UE context update request message can be the same as that of the UE context establishment request message, or it can include updated information but exclude information that has not been updated (such as including one or more of the IEs shown in Table 7); this embodiment does not impose any restrictions.

[0432] Optionally, S805 may further include: the RIC module sending a UE context update confirmation message to the CU. In this embodiment, the UE context update confirmation message is used by the RIC module in response to the UE context update request message, and its name is not limited. For example, the message may be called the y-th message, the UE AI CONTEXT MODIFICATION RESPONSE message, or other names. Wherein, y is a positive integer.

[0433] Optionally, S805 can be referred to as the UE context update process or the UE AI context modification process, and this application embodiment does not impose any limitations.

[0434] Table 7

[0435]

[0436] Optionally, when the UE enters the RRC idle state, the RRC inactive state, the UE drops the connection, or the UE switches to another base station's cell, the method shown in Figure 8(a) further includes: the CU sending a UE operation status information release request message to the RIC module. Optionally, the method may also include: the RIC module sending a UE operation status information release confirmation message to the CU.

[0437] In this method, if the base station is an integrated base station, the CU can be replaced by the base station. In a CP / UP separation scenario, the CU in this method can also be replaced by the CP and / or UP.

[0438] In this embodiment, the UE operational status information release request message is used to inform the user of the operational status information of the released or deleted UE, and its name is not limited. For example, this message can be called the y-th message, UE context release message, UE AICONTEXT RELEASE REQUEST message, or other names. Here, y is a positive integer. The UE operational status information release request message may include one or more of the IEs shown in Table 8, and this embodiment does not impose any restrictions.

[0439] In this embodiment, the UE operational status information release confirmation message is used by the RIC module in response to the UE operational status information release request message, and its name is not limited. For example, this message can be called the y-th message, the UE context release confirmation message, the UE AI CONTEXT RELEASE RESPONSE message, or other names. Wherein, y is a positive integer.

[0440] Optionally, the above method may be referred to as the UE context release process, the UE operation status information release process, or the UEAI context release process, and the embodiments of this application are not limited thereto.

[0441] Table 8

[0442]

[0443] In order to report the UE's operational status information to the RIC module, the CU needs to obtain the UE's operational status information. As mentioned above, the UE's operational status information may include one or more of the following: AI capability information, AI configuration information, and AI status information.

[0444] In one possible implementation, the UE's AI capability information can be reported by the UE to the base station or CU. Taking the UE reporting to the CU as an example, the reporting method includes: the UE sending its AI capability information to the CU, which indicates the UE's AI capabilities. Optionally, this reporting method can be triggered by the CU sending a UE Capability Enquiry message to the UE. After receiving the UE Capability Enquiry message, the UE reports its capability information to the CU. This process can be referred to as the UE capability acquisition process S806. Optionally, this reporting method can be embedded in the RRC connection establishment process between the UE and the CU, for example, the UE reports its AI capability information to the CU during the RRC connection establishment process.

[0445] In one possible implementation, when the UE switches from the source base station to the target base station, if the target base station has already obtained the UE's AI capability information from the core network or from the source base station, then the UE does not need to report the UE's AI capability information to the target base station, that is, the target base station and the UE do not need to perform a capability acquisition process; or, the UE reports the UE's AI capability information to the target base station during the RRC connection establishment process with the target base station.

[0446] Optionally, the UE's AI configuration information can be configured for the UE by the RAN (e.g., CU), so the CU does not need to obtain the UE's configuration information from the UE itself. The CU itself knows the UE's AI configuration information, or it can obtain the UE's AI configuration information from other network elements or nodes in the RAN (such as DU or other CUs).

[0447] Optionally, the UE's AI status information is information generated during operation. The UE's AI status information may be reported by the UE to the CU, monitored or measured by the CU, or monitored or tested by other network elements in the RAN (such as DU or other CUs) and then communicated to the CU.

[0448] The above describes the detailed process by which the RIC module issues tasks to the base station via S801. Optionally, in the method shown in Figure 8(b), the RIC module can execute S801 multiple times to issue multiple first tasks to the base station. Furthermore, as described below, the RIC module can also delete or update tasks.

[0449] In the method shown in Figure 8(b), the RIC module can also delete or terminate tasks already published to the base station through a task deletion process. For example, the method includes: the RIC module publishing a task release message to the base station. The task release message is used to release or terminate one or more tasks, and its name is not limited; for example, it can be called a task release request message or the y-th message, where y is a positive integer. Optionally, the method may further include: the base station sending a task release confirmation message to the RIC module. The task release confirmation message is used to confirm the release or termination of one or more tasks, and its name is not limited; for example, it can be called the y-th message, where y is a positive integer.

[0450] In one possible implementation, the task release message includes one or more of the following information:

[0451] Message type;

[0452] Message ID; and

[0453] Task Identifier; ---- Used to indicate one or more tasks to be released. This identifier is similar to the Task ID shown in Table 2.

[0454] In another possible implementation, the task release message includes one or more of the following information:

[0455] Message type;

[0456] Message ID; and

[0457] Transaction identifier; ---- Used to indicate the task to be released. This identifier is similar to the transaction identifiers shown in Table 1 and is used to release one or more tasks configured by the procedure carrying this transaction identifier. This message may include one or more transaction identifiers.

[0458] In the method shown in Figure 8(b), the RIC module can also publish the added, modified, and / or deleted tasks to the base station through a task update process. For example, the RIC module publishes a task update message to the base station to add one or more tasks, modify one or more tasks, and / or release one or more tasks. This application embodiment does not limit the name of the task update message, such as a Task Modification Request message, or the y-th message, where y is a positive integer. Optionally, the method may further include: the base station sending a task update confirmation message to the RIC module. The task update confirmation message is used to confirm the update of one or more tasks, and the name of this message is not limited; for example, it can be called the y-th message, where y is a positive integer.

[0459] For example, a task update message may include one or more of the following information:

[0460] Message type;

[0461] Message ID;

[0462] Information on one or more tasks to be added;

[0463] Information for one or more tasks to be modified; and,

[0464] The identifier of one or more tasks to be released.

[0465] The information for the one or more tasks to be added is the same as that in the TASK CONFIGURATION INFORMATION message mentioned earlier, and will not be repeated here. Alternatively, the information for each task to be added may include the parameters updated in the TASK CONFIGURATION INFORMATION message mentioned earlier, or the information for each task may include the parameters updated in the TASK CONFIGURATION INFORMATION message mentioned earlier, and will not be repeated here. Optionally, the RIC TASK ADDITION REQUEST message described above can be considered a special type of task update message.

[0466] The identifier of one or more tasks to be released can be a task ID similar to that shown in Table 2, or a transaction identifier similar to that shown in Table 1, used to release one or more tasks configured with information carrying that transaction identifier.

[0467] The above details the process by which the RIC module publishes new tasks, publishes updated tasks, and instructs the release of tasks to the base station. If the base station is an integrated base station, for each task in the first task:

[0468] (1) When the task is performed by a base station, the base station performs the task. In this embodiment of the application, for example, the RIC module instructs the base station to collect the uplink data packet loss rate of the UE.

[0469] (2) When the executor of the task is the UE, the base station publishes the task to the UE through RRC layer messages. In this embodiment of the application, for example, the base station or CU can establish a PDCP and RLC protocol example corresponding to SRB, AI-CRB or AI-DRB with the UE based on the QoS information of the task, and carry the task information through the SRB, AI-CRB or AI-DRB instance.

[0470] If the base station's CU / DU are separate, the CU receives the configuration information. For each task in the first task:

[0471] (1) When the execution subject of the task is DU, CU forwards the task to DU, and DU executes the task. Optionally, if there is a G1 interface between DU and RIC, the task for DU can be sent to DU by the RIC module through the G1 interface without CU forwarding.

[0472] In this embodiment of the application, for example, the RIC module instructs the CU to collect the uplink data packet loss rate of the UE. The CU can instruct the DU to collect the uplink data packet loss rate of the UE through a UE context modification process (S807) between the CU and the DU. The DU can send the collected data to the CU. In this embodiment of the application, the UE context modification process between the CU and the DU can be used to update the information of the UE.

[0473] In this embodiment of the application, for example, the inference results published by the RIC module to the DU or base station indicate at least one of the following information: handover threshold configuration information, cell RACH configuration information, cell downlink reference signal transmit power, cell uplink maximum transmit power information, UE serving cell configuration, UE DRX configuration, and UE DRB configuration, etc.

[0474] For example, the RIC module indicates the downlink reference signal transmit power of the cell, and the CU can carry the downlink reference signal transmit power value of the cell by sending a CU configuration update message to the DU. That is, the CU and DU can apply the inference result to the DU through the CU configuration update process S808. In this embodiment of the application, the CU configuration update process between the CU and DU can be used to update the cell information.

[0475] (2) When the executor of the task is the CU, the CU executes the task. In the embodiments of this application, by way of example, the inference results published by the RIC module to the CU or the base station indicate at least one of the following: the downlink synchronization signal transmit power of the cell, the threshold for handover decision, and the RRC connection status of the UE.

[0476] (3) When the executor of the task is the UE, the CU publishes the task to the UE through the RRC layer. Optionally, the CU can establish a PDCP and RLC protocol instance corresponding to SRB, AI-CRB or AI-DRB with the UE based on the QoS information of the task, and carry the task information through the SRB, AI-CRB or AI-DRB instance.

[0477] (4) When the CP / UP of the base station is separated, and the CU-CP that receives the configuration information receives the configuration information, then: if the CU-CP is the execution subject of the task, the CU-CP executes the task; if the CU-UP is the execution subject of the task, the CU-CP forwards the task to the CU-UP, and the CU-UP executes the task. Optionally, if there is an interface between the CU-UP and the RIC module, the task for the CU-UP can be sent by the RIC module to the CU-UP through the interface without the need for forwarding by the CU-CP.

[0478] In this embodiment, if the task issued by the RIC module to the base station, CU, DU, CU-CP, and / or CU-UP is an inference result, the corresponding module can apply the result. Optionally, if the inference result needs to be communicated to the UE by the base station, the base station will also instruct the UE to use the inference result as a parameter via signaling. For example, the task content may be adjusting the time-frequency resource location of the UE's sounding reference signal (SRS) or adjusting the UE's maximum uplink transmit power.

[0479] For any AI task published by the RIC module, if the executor of the task is the UE, the method shown in Figure 8(b) further includes S809, which is used to publish the task to the UE. For example, the base station or CU (Figure 8(b) uses CU as an example) can send the task to the UE via broadcast, multicast, or unicast. The base station can send the task content directly to the UE or send it to the UE after processing, without restriction.

[0480] In this embodiment, if the task issued by the RIC module to the UE is data collection, the base station can send the task content directly to the UE, or send it to the UE after processing. If the task issued by the RIC module to the UE is inference result, the base station can indicate the inference result to the UE in the form of parameters.

[0481] In one possible implementation, for an AI task, if the executor of the AI ​​task is all UEs in the cell, a group of UEs, or UEs meeting certain conditions, the CU can broadcast the task to the UEs through system information, paging messages, or AI-CRB. In another possible implementation, if the executor of the task is a specific UE, the CU can publish the task to that specific UE through specific RRC messages, paging messages, or AI-DRB.

[0482] Optionally, when the subject of an AI task is a UE that meets certain conditions, such as the conditions that the UEs performing the task meet as indicated in Table 2 above, the CU can determine which UE(s) will perform the task based on the conditions and the corresponding information of the UE, such as one or more of the following: AI capabilities, service status, location, and RRC connection status.

[0483] Optionally, when the subject of an AI task is a UE that meets certain conditions, such as the conditions that the UEs performing the task meet as indicated in Table 2 above, the CU can indicate the conditions to the UEs in the cell through broadcast. Each UE can determine whether to perform the AI ​​task based on the conditions and one or more of the following information: AI capabilities, service status, location, and RRC connection status.

[0484] In S809, the CU can publish one or more tasks to the UE through any of the following methods A1-A3.

[0485] Method A1: The CU publishes or instructs the UE on the content of one or more tasks via system messages or MIB.

[0486] This method can be applied when the executing entity of one or more tasks is all or multiple UEs in the cell.

[0487] For example, the CU can broadcast an updated system message to the UE through a system message update process. This system message is used to indicate the content of one or more tasks to the UE. In this embodiment of the application, from the perspective of the air interface, the CU indicating the content of one or more tasks to the UE can be regarded as the CU configuring parameters for the UE. For example, the system message includes one or more of the IEs shown in the first column of Table 9.

[0488] Table 9

[0489]

[0490] Method A2: The CU publishes or instructs the UE on the content of one or more tasks via paging messages.

[0491] This method is applicable when the subjects performing one or more tasks are multiple UEs or a specific UE in the cell.

[0492] Method A3: The CU publishes or instructs the UE on the content of one or more tasks via RRC messages.

[0493] This method is applicable when the executing entity of one or more tasks is multiple UEs or a specific UE in the cell.

[0494] In one possible implementation, the CU uses an RRC reconfiguration message to indicate the task content to the UE through the RRC reconfiguration process. Optionally, the UE can reply to the CU with an RRC reconfiguration completion message.

[0495] In another possible implementation, the CU indicates the task content to the UE through a newly added RRC message, such as a newly added RIC reconfiguration message. Optionally, the UE can reply to the CU with a response message to the newly added RRC message, such as a newly added RIC reconfiguration completion message.

[0496] For example, in mode A2 or A3, the paging message or RRC message sent by the CU to the UE includes one or more of the IEs shown in the first column of Table 10.

[0497] Table 10

[0498]

[0499] The CU can issue different tasks to the UE in the same or different ways, and this application embodiment does not impose any restrictions.

[0500] For the higher-layer messages exchanged in method A3 above, the CU can determine the RB configuration information of the higher-layer message and inform the DU of the RB configuration information, for example, through a downlink RRC message forwarding (DL RRC MESSAGE TRANSFER) message or a CU reconfiguration (CU CONFIGURATION UPDATE) message. Based on the RB configuration information, the DU sends the higher-layer message to the corresponding UE via the physical channel corresponding to the RB. Optionally, after receiving the CU CONFIGURATION UPDATE message, the DU can send a CU CONFIGURATION UPDATE ACK message back to the CU.

[0501] In the method shown in Figure 8(b), for any task, if the task type is data collection, the base station will also report the collected information to the RIC. This includes the following scenarios:

[0502] Scenario B1: For a data collection task, if the task is performed by a base station, and the base station is an integrated base station, the base station reports the collected data to the RIC.

[0503] Scenario B2: For a data collection task, if the executor of the task is the CU, and there is a G1 interface between the RIC module and the CU, then the CU reports the collected data to the RIC module through the G1 interface.

[0504] Scenario B3: For a data collection task, if the executor of the task is CU-CP and there is a G1 interface between the RIC module and CU-CP, then CU-CP reports the collected data to the RIC module through the G1 interface.

[0505] Scenario B4: For a data collection task, if the executor of the task is CU-UP and there is a G1 interface between the RIC module and CU-UP, then CU-UP reports the collected data to the RIC module through the G1 interface.

[0506] Scenario B5: For a data collection task, if the execution entity of the task is CU-UP, there is no G1 interface between the RIC module and CU-UP, but there is a G1 interface between the RIC and CU-CP, then CU-UP will forward the collected data to CU-CP, and CU-CP will report the data collected by CU-UP to the RIC module through the G1 interface.

[0507] Scenario B6: For a data collection task, if the execution entity of the task is DU, and there is a G1 interface between the RIC module and DU, then DU reports the collected data to the RIC module through the G1 interface.

[0508] Scenario B7: For a data collection task, if the execution entity of the task is DU, there is no G1 interface between the RIC module and DU, but there is a G1 interface between the RIC module and CU, then DU forwards the collected data to CU through the F1 interface, and CU reports the data collected by DU to the RIC module through the G1 interface.

[0509] When the aforementioned base station reports the collected data to the RIC module, it can do so via a task reporting message (S810). This task reporting message is used to report the collected data to the RIC module. This data can be operational status information. For example, this task reporting message can be called an RIC TASK REPORT message, a data reporting message, the y-th message, or other names, where y is a positive integer. For example, the task reporting message may include one or more of the IEs shown in the first column of Table 11 or Table 12.

[0510] Optionally, when the task is executed by a DU, and there is no interface between the DU and the RIC module, after receiving the task, the DU executes the task according to its content and sends a RIC TASK REPORT message to the CU (S811) when the reporting conditions are met. The CU can then report the data collected by the DU to the RIC module via the RIC TASK REPORT message. This condition can be periodic, event-based, or other, and is not limited. The message contains the RIC task execution result.

[0511] Table 11

[0512]

[0513] Table 12

[0514]

[0515] In the method shown in Figure 8(b), for any task, if the type of the task is data collection and the executing entity is the UE, the base station will also report the information collected by the UE to the RIC.

[0516] In one possible implementation, taking a CU / DU separated base station as an example, the UE can report its operational status information to the CU using the method described above. The CU then forwards the UE's operational status information to the RIC module using S804 and / or S805. This operational status information includes data that the RIC module requests the UE to collect.

[0517] In another possible implementation, after the UE receives the task content, or receives the task content and determines that it needs to perform the RIC task, the UE performs the task according to the task content and sends a RIC TASK REPORT message to the DU when the task reporting conditions are met (S812). Optionally, before sending this message, the UE determines the RB carrying the message according to the protocol or task priority information, and sends the RIC TASK REPORT to the DU through the RB. The DU receives the message and forwards it to the CU through an uplink RRC message forwarding (UL RRC MESSAGE TRANSFER) message.

[0518] After receiving data collected by the base station and / or UE from the CU, the RIC module can use the data to perform inference and publish the inference results to the base station and / or UE.

[0519] In one possible implementation, the RIC module publishes the inference results using the RIC TASK ADDITION REQUEST message or task update message in S801 above. The task type for publishing the inference results in this message is "Inference Result Publication". Optionally, this message may also include other types of tasks, without limitation.

[0520] In another possible implementation, the RIC module uses the inference result indication procedure S813 to send an inference result indication message to the base station, CU, or DU, thereby publishing the inference result to the base station. The inference result indication message is used to publish the inference result and its name is not limited; for example, it may be called the RIC RESULT INDICATION message or the y-th message, where y is a positive integer. For example, the inference result indication message may include one or more of the IEs shown in Table 13.

[0521] For example, after the RIC receives data collected by the UE and / or gNB, the relevant AI application performs AI calculations over a period of time and outputs inference results (calculation results). If the AI ​​calculation results require parameter configuration for the RAN or UE, such as adjusting the handover decision threshold, adjusting the downlink synchronization signal transmit power of a cell, or changing the UE's RRC connection status, the RIC sends the AI ​​inference results to the CU via the RICRESULT INDICATION message. When the RIC has a G1 interface with the CU-UP, the RIC's AI inference results may be directly published to the CU-UP. When the RIC has a G1 interface with the DU, the RIC's AI inference results may be directly published to the DU.

[0522] Table 13

[0523]

[0524] Optionally, if the AI ​​inference result is a new AI model or a modification of the original AI data collection task, the RIC module can send a RIC TASK MODIFICATION REQUEST message to the CU through the task update process described above. This message contains the new AI model or the updated AI data collection task. When the RIC has a G1 interface with the CU-UP, the RIC can directly send the RIC TASK MODIFICATION REQUEST to the CU-UP. When the RIC has a G1 interface with the DU, the RIC can directly send the RIC TASK MODIFICATION REQUEST to the DU.

[0525] After receiving the AI ​​inference results published to them, the relevant nodes apply the results, such as adjusting the threshold for handover decisions or adjusting the downlink synchronization signal transmission power of a certain cell.

[0526] For example, in the above method, if the application object of an inference result is a UE, then similar to the method described in S807 above, the base station can publish the inference result to the UE through system messages, MIBs, or paging messages, etc., and the UE can use the inference result. For example, if the application object of the inference result is the base station, then the corresponding network element in the base station uses the inference result.

[0527] For example, in the above method, when the RIC module publishes the inference results applied by the base station to the base station, the following scenarios exist. The inference results in the following scenarios can be replaced with indication information of the inference results.

[0528] Scenario C1: If the base station is an integrated base station, the inference result is applied by the base station.

[0529] Scenario C2: If the base station is a CU / DU separated base station, the application object of the inference result is the CU. There is a G1 interface between the RIC module and the CU. The CU receives the inference result through the G1 interface and applies the inference result.

[0530] Scenario C3: If the base station is a CU / DU separated base station, the application object of the inference result is the CU-CP. There is a G1 interface between the RIC module and the CU-CP. The CU-CP receives the inference result through the G1 interface and applies the inference result.

[0531] Scenario C4: If the base station is a CU / DU separated base station, the application object of the inference result is the CU-UP. There is a G1 interface between the RIC module and the CU-UP. The CU-UP receives the inference result through the G1 interface and applies the inference result.

[0532] Scenario C5: If the base station is a CU / DU separated base station, the application object of the inference result is CU-UP. There is no G1 interface between the RIC module and CU-UP. Then, CU-CP receives the inference result through the G1 interface, forwards the inference result to CU-UP, and CU-UP applies the inference result.

[0533] Scenario C6: If the base station is a CU / DU separated base station, the application object of the inference result is the DU. There is a G1 interface between the RIC module and the DU. The DU receives the inference result through the G1 interface and applies the inference result.

[0534] Scenario C7: If the base station is a CU / DU separated base station, the application object of the inference result is the DU. There is no G1 interface between the RIC module and the DU. Then the CU receives the inference result through the G1 interface, forwards the inference result to the DU, and the DU applies the inference result.

[0535] Example 2: Independent RIC architecture (Figure 6(a)) + AIC layer in parallel with RRC layer (Figure 7(b))

[0536] Figure 9(a) shows Example 2 of the architecture (network architecture + protocol stack) between the base station and the UE. Figure 9(b) shows an example of the information interaction process between the RAN and the UE using the architecture shown in Figure 9(a). Example 2 has an independent AIC protocol layer. As described in the above description of Figure 7(b), the AIC layer performs non-real-time AI functions. In addition, the AIC layer can also perform AI initialization of rt-RIC and / or some AI functions of real-time functions, such as at least one of model training, model download, and model deployment.

[0537] In the architecture shown in Figure 8(a), there is a RIC module in the RAN. In this architecture, the RIC module of the RAN performs at least one of the following operations: model download, model training, data collection, inference, and publishing inference results.

[0538] When the UE does not support RIC functionality (or can be described as having no RIC module), the RAN's RIC module can request the base station to collect data and / or publish inference results to the base station. The RAN's RIC module can also instruct the UE to collect data or send inference results to the UE via AIC layer signaling (or AIC layer messages). The AIC layer is a protocol layer parallel to the RRC layer. As described in the above description of Figure 7(b), the AIC layer signaling can be sent to the UE sequentially through lower protocol layers in the base station. The data collected by the RIC module can be either RRC layer data or physical layer data, without restriction. In this embodiment, the model used for inference can be a downloaded original model or an updated model trained based on training data, without restriction. The training data used by the RIC module for model training can be collected from the base station and / or the UE. The inference data used by the RIC module for inference can be collected from the base station and / or the UE.

[0539] When the UE supports RIC functionality, the RAN's RIC module can request the base station to collect data and / or publish inference results to the base station. The RAN's RIC module can also, through the AIC layer, instruct the UE to collect data, instruct the UE to perform federated learning, publish a model to the UE (for inference), or send inference results to the UE. The data requested from the UE can be RRC layer data or physical layer data, without limitation. In this embodiment, the model used for inference can be a downloaded original model or an updated model trained based on training data, without limitation. The training data used by the RIC module for model training can be collected from the base station and / or the UE. The inference data used by the RIC module for inference can be collected from the base station and / or the UE.

[0540] For ease of description, unless otherwise specified, the RIC module referred to below refers to the RIC module on the RAN side.

[0541] For example, based on the architecture shown in Figure 9(a), on the RAN side: the RIC module implements non-real-time AI functions through the AIC layer, and implements partial real-time AI functions through Layer 1 and / or Layer 2 in the base station. Some configuration information of the partial real-time AI functions can be sent to the UE through AIC layer signaling. For example, the RIC module implements at least one of the following through AIC layer messages: instructing the UE to collect data, instructing the configuration information of AI-related RBs, publishing inference results (parameter values) to the UE, instructing the parameter configuration related to Layer 1 and / or Layer 2 AI functions, publishing model information to the UE, and instructing the UE to perform model training (such as federated training).

[0542] For example, on the UE side: accordingly, based on the information received from the RAN side, assist in implementing AI functions for non-real-time functions and AI functions for real-time functions.

[0543] Figure 9(b) shows a flowchart illustrating the information interaction process between the RAN and UE using the architecture shown in Figure 9(a). It mainly includes: the RIC module issuing tasks to the UE; and the UE executing the corresponding tasks. Optionally, when the task issued by the RIC module includes data collection, the UE reports the collected data to the RIC module. Optionally, after the RIC module trains or updates the model based on the collected data, it can use the model for inference. Optionally, after the RIC module performs inference based on the collected data, it can publish the inference results to the UE. Optionally, when the task issued by the RIC module is model publishing, the UE receives the model information. The UE can use the model for inference. Optionally, when the task issued by the RIC module is model training, the UE reports the gradient information of the trained model parameters to the RIC module. The method shown in Figure 9(b) will be described in detail below.

[0544] S901, the RIC sends the second task configuration information to the UE; the UE receives the second task configuration information.

[0545] In this method, the second task configuration information is used by the RIC module to issue new tasks to the UE. This information is carried through AIC layer messages or signaling. In this embodiment, the name of the message carrying the second task configuration information is not limited. For example, in this embodiment, the message carrying the second task configuration information can be called the y-th message, the RIC TASKADDITION message, or the RIC TASK ADDITION REQUEST message, where y is a positive integer. This embodiment uses the name RIC TASK ADDITION REQUEST as an example. The UE can execute the corresponding task according to the instructions of this configuration information.

[0546] Optionally, S901 includes: the UE returning a confirmation message of second task configuration information to the RIC module. The second task confirmation information is used by the UE to confirm the RIC TASK ADDITION REQUEST message in S901 above with the RIC module. In this embodiment, the name of the message carrying the second task confirmation information is not limited. For example, in this embodiment, the message carrying the second task confirmation information can be called the RIC TASK ADDITION RESPONSE message, the y-th message, or other names. Where y is a positive integer.

[0547] Optionally, when the second task configuration information can configure multiple tasks (in this case, the actual message sent may include one or more tasks), the RIC TASK ADDITION REQUEST message includes one or more of the IEs shown in the first column of Table 14.

[0548] Table 14

[0549]

[0550]

[0551] Table 15

[0552]

[0553] In this embodiment of the application, S901 may be referred to as the task addition process, the task configuration process, or other names.

[0554] Similar to the method shown in Figure 8(b), in the method of Figure 9(b), the RIC module can execute S901 multiple times to issue multiple second tasks to the UE. Furthermore, as described below, the RIC module can also delete or update tasks.

[0555] For example, the RIC module can delete or terminate a task already issued to the UE through a task deletion process. For instance, this method includes: the RIC module issuing the aforementioned task release message to the UE. This task release message can be referred to as the RICTASK RELEASE REQUEST message. For details regarding the content of the RIC TASK RELEASE REQUEST message, please refer to the method shown in Figure 8(b), which will not be elaborated upon here.

[0556] For example, the RIC module can also publish the added, modified, and / or deleted tasks to the UE through the task update procedure S902. For instance, the RIC module publishes a task update message to the UE to add one or more tasks, modify one or more tasks, and / or release one or more tasks. This task update message can be called a RIC TASK MODIFICATION REQUEST message. The specific content of the RIC TASK MODIFICATION REQUEST message can be similar to the method shown in Figure 8(b), and will not be repeated here.

[0557] In the method shown in Figure 9(b), AIC layer messages sent by the RIC module to the UE, such as RIC TASKADDITION REQUEST, RIC TASK RELEASE REQUEST, and RIC TASK MODIFICATION REQUEST messages, can be sent to the UE via the base station. As described above for Figure 7(b), these AIC layer messages can sequentially pass through the PDCP layer, RLC layer, MAC layer, and physical layer of the base station, and are sent from the physical layer of the base station to the physical layer of the UE. At the UE layer, they are sequentially passed through the MAC layer, RLC layer, PDCP layer, and AIC layer, and then interpreted by the UE's AIC layer. For example, if the CU receives this message from the RIC module via the G1 interface, it can determine the RB carrying the message based on the priority information corresponding to the message, configure the protocol layers of the RB accordingly, and then the DU sends the message to the UE.

[0558] Similar to the method in Figure 8(b), the method shown in Figure 9(b) may include S802 (referred to as S903 in this method), S803 (referred to as S904 in this method), S804 / S805 (referred to as S905 / S906 in this method), and the F1 interface establishment process.

[0559] Similar to S806 in the method of Figure 8(b), the method in Figure 9(b) may include S907: The CU can obtain the UE's AI capability information through the UE's capability acquisition process; or, the CU can obtain the UE's AI capability information from the core network or the source base station. The CU can send this information to the RIC module through the above-mentioned S905 and / or S906 processes. After obtaining the UE's AI capability information, the RIC module can initiate the establishment process of one or more air interface RBs according to QoS requirements, etc., in order to carry uplink and downlink AIC layer data or signaling for AI-related processes such as AI data collection, AI model download, and AI model update.

[0560] Optionally, the method shown in Figure 9(b) may further include S902: the RIC module issues a task to the base station, the content of which is: data collection or inference results. The specific configuration method is similar to the corresponding method in Figure 8(b). The difference from the method in Figure 8(b) is that the task issued by the RIC module to the base station in S902 does not include the task issued by the RIC module to the UE. The reason for this is that, as mentioned above, in the method shown in Figure 9(b), the task is issued from the AIC layer of the RIC module to the AIC layer of the UE.

[0561] The RIC module can issue tasks to at least one of the CU, DU, CU-CP, and CU-UP. For example, the data collection task issued by the RIC module to the CU or DU is: requesting to measure the uplink transmission latency of the UE.

[0562] Optionally, the method shown in Figure 9(b) may also include a configuration update process for the RIC and the base station.

[0563] The above describes the RIC module issuing tasks to the UE or base station. The RIC module may require the CU and / or DU to perform corresponding operations to cooperate with or assist in completing the issued RIC task. The process for performing this operation can be called a configuration update process. For example, if the RIC module instructs multiple UEs to collect data, the RIC module can configure uplink resources and other uplink transmission parameters for the base station. The base station can then instruct these uplink transmission parameters to the UE via the air interface, allowing the UE to anonymously report the collected data using these parameters. As another example, if the RIC module instructs the UE to perform federated learning, the RIC module can configure uplink resources and other uplink transmission parameters for the base station. The base station can then instruct these uplink parameters to the UE via the air interface, allowing the UE to report the gradient information of the model parameters using these parameters. For instance, the base station configures the time-frequency resources, dedicated preambles, etc., used by the UE to send anonymous data collection or federated learning uplink reports via system information on the air interface, so that the UE can use these radio resources for uplink transmission when it has the data to report. The RIC module can send a configuration update message to the CU, requesting the CU to perform corresponding operations. Upon receiving the request, the CU returns a configuration update acknowledgment message to the RIC module. Similarly, optionally, a configuration update process can occur between the RIC module and the DU. Alternatively, a configuration update process can occur between the RIC module and the CU-UP.

[0564] In the method shown in Figure 9(b), after receiving the task issued by the RIC module, the UE executes the task according to its content and sends a RIC TASK REPORT message (S908) to the RIC module when the task reporting conditions are met. This message is an AIC layer message. Optionally, before sending this message, the UE determines the RB carrying the message according to the protocol or task priority information, and sends information to the base station through the RB. This information includes the RIC TASK REPORT message. The base station or its CU can submit this information to the RIC module, and the RIC module interprets the RIC TASK REPORT message sent by the UE at the AIC layer. For a description of the RIC TASK REPORT message, please refer to Table 11; it will not be repeated here. Optionally, after successfully receiving the RIC TASK REPORT message, the RIC module can return an acknowledgment message to the UE, which can be called a RIC TASK REPORTACKNOWLEDGE message. This message is also an AIC layer message.

[0565] Similar to the method in Figure 8(b), after receiving the task issued by the RIC module, the CU and / or DU execute the task according to the task content and send a RIC TASK REPORT message to the RIC module (S909) when the task reporting conditions are met. Optionally, after successfully receiving the RIC TASK REPORT message, the RIC module can return an acknowledgment message to the CU and / or DU, which can be called a RIC TASK REPORT ACKNOWLEDGE message.

[0566] After receiving data collected by the base station and / or UE from the CU, the RIC module can use the data to perform inference and publish the inference results to the base station and / or UE.

[0567] In one possible implementation, similar to the method shown in Figure 8(b), the method shown in Figure 9(b) may include S910: the RIC module publishes the inference results to the base station. The RIC module can publish the AI ​​inference results to the base station (CU, DU, CU-CP, and / or CU-UP).

[0568] In one possible implementation, the RIC module can utilize the S902 described above to publish updated models, inference results, and / or updated data collection tasks to the UE via the RIC TASKMODIFICATION REQUEST message.

[0569] Example 3: Independent RIC architecture (Figure 6(a)) + AIC layer above the RRC layer (Figure 7(c))

[0570] Figure 10The diagram shows Example 3 of the architecture (network architecture + protocol stack) between the base station and the UE. Example 3 features an independent AIC protocol layer, which performs non-real-time AI functions. Furthermore, the AIC layer can also perform AI initialization for rt-RIC and / or partial AI functions for real-time functions, such as at least one of model training, model download, and model deployment.

[0571] use Figure 10 In the architecture shown, except for the differences described below, the example diagram of the information interaction process between the RAN and UE is the same as that in Figure 9(b). The similarities between the two will not be repeated.

[0572] The difference between Example 2 and Example 3 lies in the information exchange process between the RAN and UE:

[0573] As described in Figures 7(b) and 7(c), in Example 2, the AIC layer messages sent by the RIC module are delivered from the AIC layer to the PDCP layer of the CU. In this architecture, the mapping of AIC layer messages to RB can be done at the G1 interface according to the application layer protocol. In Example 3, the AIC layer messages sent by the RIC module are delivered from the AIC layer to the RRC layer. In this architecture, the mapping of AIC layer messages to RB is done at the RRC layer.

[0574] Example 4: Standalone RIC architecture (Figure 6(a)) + application layer (Figure 7(d))

[0575] Figure 11 The diagram shows Example 4 of the architecture (network architecture + protocol stack) between the base station and the UE. The functional descriptions of each protocol layer in Example 4 are shown in Figure 7(d), and will not be repeated here.

[0576] use Figure 10 In the architecture shown, except for the differences described below, the information interaction flow diagram between the RAN and UE is the same as in Example 1, Example 2, or Example 3. For details on which example it is identical to, please refer to the corresponding description in Figure 7(d) above; it will not be repeated here. The similarities between the two will not be elaborated further.

[0577] In Example 4, the RIC TASK ADDITION REQUEST message is not used for model publishing; that is, the task type indicated by the RIC TASK ADDITION REQUEST does not include model publishing. When the RIC module publishes a task to the UE publishing module, this function is implemented through the application layer of the RIC module and the application layer of the UE. When the RIC module publishes a model, the model information is carried on a data RB, such as an AI-CRB, and sent to the UE.

[0578] Furthermore, unlike the F1 interface establishment process in the examples above, in Example 4, a General Packet Radio Service (GPRS) Tunnel Protocol-User Plane (GTP-u) public tunnel is created during the F1 interface establishment process for the transmission of AI public signaling / data between the RIC module and the UE. For example, this tunnel is used to carry AI models published by the RIC module to the UE, updated AI models, or AI model gradients reported by the UE to the RIC module. Between the RIC module and the UE, only one GTP-u public tunnel can be established, using priority information carried in the GTP-u header to distinguish data with different QoS requirements, thus mapping the data to the corresponding AI-CRB on the air interface; alternatively, a GTP-u public tunnel can be established for each priority level, with one GTP-u public tunnel corresponding to one AI-CRB, and data from different GTP-u public tunnels directly carried on the corresponding AI-CRB.

[0579] Unlike the examples above, in Example 4, if the task received by the UE is model training, the UE sends the gradient information of the model parameters to the application layer of the RIC module through the application layer. For example, the UE sends the gradient of the AI ​​model parameters to the CU through AI-CRB, and then the CU sends the gradient of the AI ​​model parameters to the RIC module through the GTP-u common tunnel of the G1 interface.

[0580] Example 5: Embedded RIC architecture (Figure 6(b)) + various possible protocol layers ( Figures 7(a)-7(d) )

[0581] In Example 5, an embedded RIC architecture is used, meaning that the RIC module is part of the base station and there is no RIC module independent of the base station.

[0582] The transmission process using this architecture is similar to Examples 1 to 4 described above. The difference is that in Example 5, the processes and operations involving the G1 interface are changed to the corresponding processes and operations inside the base station, which gives us the implementation method of Example 5.

[0583] Example 6: Hybrid RIC architecture (Figure 6(c)) + various possible protocol layers ( Figures 7(a)-7(d) )

[0584] In Example 6, a hybrid RIC architecture is adopted, that is, the nrt-RIC module is part of the base station, while the rt-RIC module is independent of the base station.

[0585] For non-real-time AI functions, the transmission process using this architecture is similar to Examples 1 to 4 described above. For real-time AI functions, the transmission process using this architecture differs from Examples 1 to 4 in that processes such as publishing the model to the UE and reporting the gradient of AI model parameters to the UE are carried by the AI-CRB over the air interface. For other AI processes, the corresponding AI information can be carried by MAC CE, DCI, or uplink control information (UCI).

[0586] In the embodiments provided above, the methods provided by this application are described from the perspectives of individual network elements and the interaction between different network elements. To implement the functions of the methods provided in the embodiments of this application, each network element may include hardware structures and / or software modules, implementing the above functions in the form of hardware structures, software modules, or a combination of hardware structures and software modules. Whether a particular function is executed in the form of hardware structures, software modules, or a combination of hardware structures and software modules depends on the specific application and design constraints of the technical solution.

[0587] Figure 12 The diagram shown is a structural example of the device 300 provided in an embodiment of this application.

[0588] In one possible implementation, device 300 is used to implement the functions of the RIC in the above method. This device can be the RIC itself, or it can be other devices capable of implementing the functions of the RIC. These other devices can be installed in the RIC or used in conjunction with the RIC.

[0589] In one possible implementation, device 300 is used to implement the functions of a base station in the above method. This device can be a base station itself, or it can be other devices capable of implementing the functions of a base station. These other devices can be installed in the base station or used in conjunction with the base station.

[0590] In one possible implementation, device 300 is used to implement the functions of the terminal device in the above method. This device can be the terminal device itself, or it can be other devices capable of implementing the functions of the terminal device. These other devices can be installed in the terminal device or can be used in conjunction with the terminal device.

[0591] The device 300 includes a receiving module 301 for receiving signals or information. The device 300 also includes a transmitting module 302 for transmitting signals or information. The device 300 further includes a processing module 303 for processing the received signals or information, such as decoding the signals or information received by the receiving module 301. The processing module 303 can also generate signals or information to be transmitted, for example, for generating signals or information to be transmitted via the transmitting module 302.

[0592] The module division in this embodiment is illustrative and represents a logical functional division; in actual implementation, other division methods may be used. For example, the receiving module 301 and the transmitting module 302 can also be integrated into a transceiver module or a communication module. Furthermore, the functional modules in the various embodiments of this application can be integrated into one module, exist as separate physical entities, or have two or more modules integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0593] Figure 13 The diagram shown is a structural example of the device 400 provided in an embodiment of this application.

[0594] In one possible implementation, device 400 is used to implement the functionality of the RIC in the above-described method. This device can be the RIC itself, or it can be other devices capable of implementing the functionality of the RIC. These other devices can be installed within the RIC or used in conjunction with the RIC. For example, device 400 can be a chip system. In this embodiment, the chip system can be composed of chips or can include chips and other discrete devices. For example, device 400 includes at least one processor 420 for implementing the functionality of the RIC in the method provided in this embodiment.

[0595] In one possible implementation, device 400 is used to implement the functions of a base station in the above-described method. This device can be a base station itself, or it can be other devices capable of implementing the functions of a base station. These other devices can be installed in the base station or used in conjunction with the base station. For example, device 400 can be a chip system. For example, device 400 includes at least one processor 420 for implementing the functions of a base station in the method provided in the embodiments of this application.

[0596] In one possible implementation, the device 400 is used to implement the functions of the terminal device in the above-described method. This device can be the terminal device itself, or it can be other devices capable of implementing the functions of the terminal device. These other devices can be installed in the terminal device or used in conjunction with the terminal device. For example, the device 400 can be a chip system. For example, the device 400 includes at least one processor 420 for implementing the functions of the terminal device in the method provided in the embodiments of this application.

[0597] The device 400 may further include at least one memory 430 for storing program instructions and / or data. The memory 430 is coupled to the processor 420. The coupling in this embodiment is an indirect coupling or communication connection between devices, units, or modules, which may be electrical, mechanical, or other forms, for information exchange between devices, units, or modules. The processor 420 may cooperate with the memory 430 to implement the functions described in the above method embodiments. The processor 420 may execute program instructions stored in the memory 430. At least one of the at least one memory may be included in the processor 420.

[0598] The device 400 may further include a communication interface 410 for communicating with other devices via a transmission medium, thereby enabling the devices in the device 400 to communicate with other devices. The processor 420 uses the communication interface 410 to send and receive signals to implement the functions described in the above method embodiments. In this application embodiment, the communication interface may be a transceiver, circuit, bus, module, pin, or other type of communication interface.

[0599] This application embodiment does not limit the specific connection medium between the communication interface 410, processor 420, and memory 430. This application embodiment... Figure 13 The memory 430, processor 420, and transceiver 410 are connected via a bus 440, and the bus is in Figure 13 The connections between other components are shown in bold and are for illustrative purposes only, not as limiting information. The bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, Figure 13 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0600] In the embodiments of this application, the processor may be a general-purpose processor, a digital signal processor, 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, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0601] In the embodiments of this application, the memory can be non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), or it can be volatile memory, such as random-access memory (RAM). Memory is any other medium capable of carrying or storing desired program code in the form of instructions or data structures, and accessible by a computer, but is not limited thereto. The memory in the embodiments of this application can also be a circuit or any other device capable of implementing storage functions, used to store program instructions and / or data.

[0602] The technical solutions provided in this application can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, they can be implemented in whole or in part as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a terminal device, a RIC, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs (DVDs)), or semiconductor media, etc.

[0603] In the embodiments of this application, provided there is no logical contradiction, the embodiments may reference each other. For example, the methods and / or terms between method embodiments may reference each other, the functions and / or terms between device embodiments may reference each other, and the functions and / or terms between device embodiments and method embodiments may reference each other.

[0604] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A communication method, characterized in that, include: Send first task configuration information to the base station, wherein the first task configuration information is used to indicate the configuration information of one or more artificial intelligence (AI) tasks; Wherein, for each of the one or more AI tasks, the configuration information of the AI ​​task is used to indicate the task type and task content in the AI ​​task; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

2. The method according to claim 1, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status; wherein the execution subject of the AI ​​task is a terminal device or the base station.

3. The method according to claim 2, characterized in that, The task status includes activated or deactivated, or the task status includes activated, deactivated, or released.

4. The method according to claim 1 or 2, characterized in that, The method further includes: The base station receives a first interface establishment request message from its central unit (CU), wherein the first interface establishment request message indicates one or more of the following: Message type; The ID of the CU; The capability information of the CU; The configuration information of the CU; and The status information of the CU.

5. The method according to claim 1 or 2, characterized in that, The method further includes: The distribution unit (DU) of the base station receives a second interface establishment request message, wherein the second interface establishment request message is used to indicate one or more of the following: Message type; The ID of the DU; The capability information of the DU; The configuration information of the DU; and The status information of the DU.

6. The method according to claim 2, characterized in that, The terminal device receives information from the base station, and the terminal device information includes one or more of the following: the terminal device's capability information, the terminal device's configuration information, and the terminal device's status information.

7. The method according to claim 1 or 2, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: The collected data is received from the base station.

8. The method according to claim 1 or 2, characterized in that, The method further includes: The inference results are published to the base station.

9. The method according to claim 1 or 2, characterized in that, The one or more AI tasks include at least one model training task, and the method further includes: Receive model parameter information or model parameter gradient information from the base station.

10. A communication method, characterized in that, include: Receive first task configuration information from the wireless intelligent controller RIC, the first task configuration information being used to indicate the configuration information for one or more artificial intelligence (AI) tasks; Wherein, for each of the one or more AI tasks, the configuration information of the AI ​​task is used to indicate the task type and task content in the AI ​​task; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

11. The method according to claim 10, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status; wherein, the execution subject of the AI ​​task is a terminal device or a base station.

12. The method according to claim 11, characterized in that, The task status includes activated or deactivated, or the task status includes activated, deactivated, or released.

13. The method according to claim 11, characterized in that, For at least one of the one or more AI tasks, where the executing entity of the at least one AI task is the terminal device, the method further includes: Information about each AI task in the at least one AI task is indicated to the terminal device via Radio Resource Control (RRC) signaling, System Information Block (SIB), Master Information Block (MIB), or paging messages.

14. The method according to claim 11, characterized in that, The method further includes: Send a first interface establishment request message to the RIC, wherein the first interface establishment request message is used to indicate one or more of the following: Message type; The ID of the centralized unit (CU) of the base station; The capability information of the CU; The configuration information of the CU; and The status information of the CU.

15. The method according to claim 11, characterized in that, The method further includes: Send the terminal device information to the RIC, wherein the terminal device information includes one or more of the following: the terminal device's capability information, the terminal device's configuration information, and the terminal device's status information.

16. The method according to claim 11, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: The collected data is sent to the RIC.

17. The method according to claim 16, characterized in that, The data is received from the distribution unit (DU) of the base station or the terminal device.

18. The method according to claim 17, characterized in that, The method further includes: Receive inference results from the RIC.

19. The method according to claim 18, characterized in that, The method further includes: The inference result is sent to the terminal device.

20. The method according to claim 19, characterized in that, The one or more AI tasks include at least one model training task, and the method further includes: The model parameter information or model parameter gradient information is sent to the RIC, and the model parameter information or model parameter gradient information comes from the terminal device.

21. A communication method, characterized in that, include: Receive information about one or more artificial intelligence (AI) tasks from the base station via Radio Resource Control (RRC) signaling, System Information Block (SIB), Master Information Block (MIB), or paging messages; Wherein, for each of the one or more AI tasks, the AI ​​task information is used to indicate the task type and task content in the AI ​​task; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

22. The method according to claim 21, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status; wherein, the execution subject of the AI ​​task includes one or more terminal devices.

23. The method according to claim 22, characterized in that, The task status includes activated or deactivated, or the task status includes activated, deactivated, or released.

24. The method according to any one of claims 22-23, characterized in that, The method further includes: Send terminal device information to the base station, wherein the terminal device information includes one or more of the following: the terminal device's capability information, the terminal device's configuration information, and the terminal device's status information.

25. The method according to any one of claims 21-23, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: The collected data is sent to the base station.

26. The method according to any one of claims 21-23, characterized in that, The method further includes: Receive inference results from the base station.

27. The method according to any one of claims 21-23, characterized in that, The one or more AI tasks include at least one model training task, and the method further includes: Send model parameter information or model parameter gradient information to the base station.

28. A communication method, characterized in that, include: Through the first protocol layer, second task configuration information is sent to the terminal device. The second task configuration information is used to indicate the configuration information of one or more artificial intelligence (AI) tasks. Wherein, for each of the one or more AI tasks, the configuration information of the AI ​​task is used to indicate the task type and task content in the AI ​​task; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

29. The method according to claim 28, characterized in that, The method further includes: Send third task configuration information to the base station, wherein the third task configuration information is used to indicate the configuration information of one or more AI tasks; The configuration information of each AI task in the one or more AI tasks indicated by the third task configuration information is used to indicate one or more of the following for each AI task: task ID, task type, task content, task execution subject, and task status, wherein the execution subject of the AI ​​task includes the centralized unit (CU) or the distributed unit (DU) of the base station.

30. The method according to claim 29, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status; wherein, the execution subject of the AI ​​task includes one or more terminal devices.

31. The method according to any one of claims 29-30, characterized in that, The task status includes activated or deactivated, or the task status includes activated, deactivated, or released.

32. The method according to any one of claims 29-30, characterized in that, The method further includes: The CU of the base station receives a first interface establishment request message, wherein the first interface establishment request message is used to indicate one or more of the following: Message type; The ID of the CU; The capability information of the CU; The configuration information of the CU; and The status information of the CU.

33. The method according to any one of claims 29-30, characterized in that, The method further includes: The base station receives a second interface establishment request message from its DU, wherein the second interface establishment request message indicates one or more of the following: Message type; The ID of the DU; The capability information of the DU; The configuration information of the DU; and The status information of the DU.

34. The method according to any one of claims 29-30, characterized in that, The terminal device receives information from the base station, and the terminal device information includes one or more of the following: the terminal device's capability information, the terminal device's configuration information, and the terminal device's status information.

35. The method according to claim 29 or 30, characterized in that, The third task configuration information indicates that the one or more AI tasks include at least one data collection task, and the method further includes: The collected data is received from the base station.

36. The method according to claim 29 or 30, characterized in that, The method further includes: The inference results are published to the base station.

37. The method according to any one of claims 28-30, characterized in that, The method further includes at least one data collection task among the one or more AI tasks indicated by the second task configuration information: The collected data is received from the terminal device through the first protocol layer.

38. The method according to any one of claims 28-30, characterized in that, The method further includes: The inference results are published to the terminal device through the first protocol layer.

39. The method according to any one of claims 28-30, characterized in that, The method further includes: (The second task configuration information indicates that the one or more AI tasks include at least one model training task.) The model parameter information or model parameter gradient information is received from the terminal device through the first protocol layer.

40. The method according to any one of claims 28-30, characterized in that, The first protocol layer is the Artificial Intelligence Control (AIC) layer above the Packet Data Convergence Protocol (PDCP) layer. The first protocol layer is the AIC layer above the Radio Resource Control (RRC) layer, or... The first protocol layer is the application layer.

41. A communication method, characterized in that, include: Receive third task configuration information from the wireless intelligent controller RIC, the third task configuration information being used to indicate the configuration information for one or more artificial intelligence (AI) tasks; For each of the one or more AI tasks, the configuration information of the AI ​​task is used to indicate the task type and task content in the AI ​​task; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

42. The method according to claim 41, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status; wherein, the execution subject of the AI ​​task includes the centralized unit (CU) of the base station or the distributed unit (DU) of the base station.

43. The method according to claim 42, characterized in that, For at least one of the one or more AI tasks, where the executor of the at least one AI task is the DU, the method further includes: Send information about each AI task in the at least one AI task to the DU.

44. The method according to claim 42 or 43, characterized in that, The method further includes: Send a first interface establishment request message to the RIC, wherein the first interface establishment request message is used to indicate one or more of the following: Message type; The ID of the CU; The capability information of the CU; The configuration information of the CU; and The status information of the CU.

45. The method according to claim 44, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: The collected data is sent to the RIC.

46. ​​The method according to claim 45, characterized in that, The data is received from the DU of the base station.

47. The method according to claim 46, characterized in that, The method further includes: Receive inference results from the RIC.

48. A communication method, characterized in that, include: The system receives second task configuration information from the wireless intelligent controller (RIC) through the first protocol layer. The second task configuration information is used to indicate the configuration information of one or more artificial intelligence (AI) tasks. Wherein, for each of the one or more AI tasks, the configuration information of the AI ​​task is used to indicate the task type and task content in the AI ​​task; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

49. The method according to claim 48, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status; wherein, the execution subject of the AI ​​task includes one or more terminal devices.

50. The method according to claim 48, characterized in that, The method further includes: sending terminal device information to the RIC via a base station, wherein the terminal device information includes one or more of the following: the terminal device's capability information, the terminal device's configuration information, and the terminal device's status information.

51. The method according to claim 48 or 50, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: The collected data is sent to the RIC through the first protocol layer.

52. The method according to claim 51, characterized in that, The method further includes: The inference results are received from the RIC through the first protocol layer.

53. The method according to claim 52, characterized in that, The one or more AI tasks include at least one model training task, and the method further includes: The model parameter information or model parameter gradient information is sent to the RIC through the first protocol layer.

54. The method according to claim 53, characterized in that, The first protocol layer is the Artificial Intelligence Control (AIC) layer above the Packet Data Convergence Protocol (PDCP) layer. The first protocol layer is the AIC layer above the Radio Resource Control (RRC) layer, or... The first protocol layer is the application layer.

55. A communication method, characterized in that, include: The system sends one or more artificial intelligence (AI) task information to the terminal device through the application layer, AI control (AIC) layer, radio resource control (RRC) signaling, system information block (SIB), master information block (MIB), paging message, media access control (MAC) control element (CE), or physical layer information. The AIC layer is located above the packet data aggregation layer protocol (PDCP) layer, or the AIC layer is located above the RRC layer. For each of the one or more AI tasks, the information of the AI ​​task is used to indicate the task type and task content of the AI ​​task below; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

56. The method according to claim 55, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status.

57. The method according to claim 56, characterized in that, The task status includes activated or deactivated, or the task status includes activated, deactivated, or released.

58. The method according to claim 55 or 56, characterized in that, The method further includes: The terminal device receives information from the terminal device, wherein the information of the terminal device includes one or more of the following: the capability information of the terminal device, the configuration information of the terminal device, and the status information of the terminal device.

59. The method according to claim 55 or 56, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: Receive the collected data from the terminal device.

60. The method according to claim 55 or 56, characterized in that, The method further includes: The inference results are sent to the terminal device.

61. The method according to claim 55 or 56, characterized in that, The one or more AI tasks include at least one model training task, and the method further includes: Receive model parameter information or model parameter gradient information from the terminal device.

62. A communication method, characterized in that, include: Information about one or more artificial intelligence (AI) tasks is received from the base station through the application layer, AI control (AIC) layer, radio resource control (RRC) signaling, system information block (SIB), master information block (MIB), paging messages, media access control (MAC) control element (CE), or physical layer information. The AIC layer is located above the packet data aggregation layer protocol (PDCP) layer, or the AIC layer is located above the RRC layer. For each of the one or more AI tasks, the information of the AI ​​task is used to indicate the task type and task content of the AI ​​task below; When the task type is data collection, the task content indicates one or more of the following: data measurement type, measurement conditions, and measurement result reporting method; The task type is when the reasoning result is published, and the task content indicates the reasoning result; The task type is defined when the model is published, and the task content indicates model information; or, The task type is during model training, and the task content indicates one or more of the following: conditions for reporting model parameter information or reporting model parameter gradient information, information of the reference neural network, and neural network training dataset.

63. The method according to claim 62, characterized in that, The configuration information of the AI ​​task is also used to indicate one or more of the following: task identifier ID, task execution subject, and task status.

64. The method according to claim 62, characterized in that, The method further includes: Send terminal device information to the base station, wherein the terminal device information includes one or more of the following: the terminal device's capability information, the terminal device's configuration information, and the terminal device's status information.

65. The method according to claim 62 or 64, characterized in that, The one or more AI tasks include at least one data collection task, and the method further includes: The collected data is sent to the base station.

66. The method according to claim 65, characterized in that, The method further includes: Receive inference results from the base station.

67. The method according to claim 66, characterized in that, The one or more AI tasks include at least one model training task, and the method further includes: Send model parameter information or model parameter gradient information to the base station.

68. An apparatus, characterized in that, It includes a processor and a memory, the memory and the processor being coupled together, the processor being configured to perform the method according to any one of claims 1-67.

69. A computer-readable storage medium, characterized in that, Includes instructions that, when executed on a computer, cause the computer to perform the method described in any one of claims 1-67.

Images