Data processing method and apparatus

Abstract

Embodiments of the present application provide a data processing method and apparatus, comprising: receiving first indication information, the first indication information being used for indicating N tasks and M models corresponding to the N tasks, N being less than M, and N and M being positive integers greater than 1; performing switching evaluation on the basis of the first indication information, the switching evaluation being used for evaluating the execution quality of the N tasks on the basis of different sets, and the sets being used for indicating N models used for executing the N tasks among the M models; and determining a first set, the first set being one set among the different sets, and the execution quality corresponding to the first set satisfying an evaluation condition. In this way, a plurality of execution modes for executing the N tasks are evaluated, so as to determine the first set that meets the evaluation condition. The evaluation condition may be a QoS requirement. Subsequently, the execution of the N tasks on the basis of the first set that meets the evaluation condition can effectively ensure the execution quality, and avoid failing to meet the QoS requirement of a wireless communication network when executing a plurality of tasks.

Description

A data processing method and apparatus

[0001] This application claims priority to Chinese Patent Application No. 202510107193.9, filed on January 22, 2025, entitled “A Data Processing Method and Apparatus”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This disclosure relates to the field of communication technology, and in particular to a data processing method and apparatus. Background Technology

[0003] With the continuous development of artificial intelligence (AI) and machine learning (ML) technologies, the application of AI / ML technologies in wireless communication networks can effectively improve the intelligence level and automation capabilities of wireless networks. Network devices in wireless networks not only support traditional wireless communication connections but can also provide AI / ML services, i.e., intrinsic intelligence within the wireless network.

[0004] In practical implementation, terminal devices or network devices in a wireless communication network can execute AI / ML tasks. For example, terminal devices or network devices can perform time-domain channel estimation or frequency-domain channel estimation based on corresponding AI / ML models. Typically, the same AI / ML task can correspond to one or more AI / ML models, where each AI / ML model can have a different model architecture, different execution quality, or meet other specific requirements.

[0005] In practical applications, multiple AI / ML models corresponding to the same AI / ML task can be switched between each other to meet the Quality of Service (QoS) requirements of wireless network systems for AI / ML services. For example, a terminal device can switch between multiple AI / ML models corresponding to the same AI / ML task to ensure that the execution quality of the AI / ML task meets QoS requirements. QoS requirements can be requirements for the execution time of the AI / ML task, the bandwidth occupied by the AI / ML task, or other requirements. However, during the execution of AI / ML tasks, each AI / ML task corresponds to one or more AI / ML models. Even if the execution quality of each AI / ML task meets the QoS requirements, the overall execution quality of multiple AI / ML tasks may still fail to meet the QoS requirements of the wireless communication network for AI / ML services. Summary of the Invention

[0006] This disclosure provides a data processing method and apparatus to meet the QoS requirements of a wireless communication network when performing multiple AI / ML tasks.

[0007] In a first aspect, this disclosure provides a data processing method that can be used on a terminal side, for example, executed by a terminal device. The terminal device can be a device or apparatus with a chip, or a device or apparatus with integrated circuits, or a chip, chip system, module, or control unit in the aforementioned device or apparatus; specific details are not limited in this application. It should be noted that in this application, when referring to a terminal device, it can refer to the terminal device itself, or to the chip, functional module, or integrated circuit in the terminal device that performs the method provided in this application; specific details are not limited in this application. In the first aspect and its possible implementations, the method is described using the execution of the method by a terminal device as an example. The method includes: the terminal device receiving first indication information, wherein the first indication information is used to indicate N tasks and M models corresponding to the N tasks, where N is less than M, and N and M are positive integers greater than 1; the terminal device performing a switching evaluation based on the first indication information, the switching evaluation being used to evaluate the execution quality of the N tasks based on different sets, the sets indicating N models among the M models used to execute the N tasks; and the terminal device determining a first set, where the first set is one of the different sets, and the execution quality corresponding to the first set satisfies the evaluation conditions.

[0008] In this way, by evaluating multiple execution methods for executing N tasks (execution based on a set of tasks is considered one execution method), a first set that meets the evaluation criteria (QoS requirements) is determined. Subsequently, executing N tasks based on the first set that meets the evaluation criteria can effectively ensure execution quality and avoid failing to meet the QoS requirements of the wireless communication network when executing multiple tasks.

[0009] Based on the first aspect, in one possible implementation, the first indication information is specifically used to indicate the identifiers of N tasks and / or M models. This implementation illustrates a specific form of the first indication information, namely, that the first indication information is specifically used to indicate the identifiers of N tasks and / or M models, which is beneficial for the implementation of the solution.

[0010] Based on the first aspect, in one possible implementation, the first indication information is further used to indicate the priority of some or all of the M models, where the priority is determined based on the switching criteria. This implementation illustrates a specific form of the first indication information, namely, that the first indication information is further used to indicate the priority of some or all of the M models. Furthermore, during the switching evaluation process, using priority for switching evaluation can reduce the scale of the evaluation; for example, sets with excessively low priorities do not need to be evaluated, effectively improving the efficiency of the switching evaluation and facilitating the implementation of the solution.

[0011] Based on the first aspect, in one possible implementation, the first indication information is further used to indicate K sets, where K is a positive integer greater than 1. The K sets are determined based on the priorities of some or all of the M models, and the priorities are determined based on switching criteria. This implementation illustrates one form of the first indication information, namely, that the first indication information is also used to indicate K sets. Furthermore, during the switching evaluation process, using K sets for switching evaluation can reduce the scale of the evaluation; for example, the K sets do not include sets with excessively low priorities, effectively improving the efficiency of the switching evaluation and facilitating the implementation of the solution.

[0012] Based on the first aspect, in one possible implementation, the first indication information is further used to indicate the priority of all or some subsets among the J subsets. The subsets indicate at least one model among the M models, where J is a positive integer greater than 1, and the priority is determined based on the switching criteria. This implementation illustrates one form of the first indication information, where it is further used to indicate the priority of all or some subsets among the J subsets. Furthermore, during the switching evaluation process, using priority for switching evaluation can reduce the scale of the evaluation; for example, sets with excessively low priorities do not need to be evaluated, effectively improving the efficiency of the switching evaluation and facilitating the implementation of the solution. In addition, a group of models that need to be executed within the same set can form a subset; for example, a group of models that need to be executed sequentially can form a subset. Switching evaluation based on the set composed of subsets can reduce the scale of the evaluation, further improving the efficiency of the switching evaluation and facilitating the implementation of the solution.

[0013] Based on the first aspect, in one possible implementation, the switching evaluation is further used to evaluate the execution quality of N tasks based on different sequences, where the sequence indicates an ordered structure composed of N models out of M models used to execute the N tasks. The method also includes: determining a first sequence, which is one of the different sequences, and the execution quality corresponding to the first sequence satisfies the evaluation conditions. In this implementation, by evaluating multiple execution methods for executing N tasks (execution based on a sequence is considered one execution method), a first sequence that meets the evaluation conditions (QoS requirements) is determined. Subsequent execution of N tasks based on this first sequence can effectively guarantee execution quality and avoid failing to meet the QoS requirements of the wireless communication network when executing multiple tasks.

[0014] Based on the first aspect, in one possible implementation, the first indication information is further used to indicate the switching timing of all or some of the M models and / or the activation status of all or some of the N tasks. The switching timing indicates when to end the execution of the model, and includes the switching period or the operation that triggers the switching. The activation status indicates whether the task is active. This implementation illustrates one form of the first indication information, namely, that the first indication information is further used to indicate the switching timing of all or some of the M models and / or the activation status of all or some of the N tasks, which is beneficial for the implementation of the solution.

[0015] Based on the first aspect, in one possible implementation, the first indication information is carried in Radio Resource Control (RRC) signaling, Media Access Control (MAC) Control Element (CE), or Downlink Control Information (DCI). This implementation illustrates a specific method of carrying the first indication information; that is, the first indication information can be carried in RRC signaling, MAC CE, or DCI, which is beneficial for the implementation of the solution.

[0016] Based on the first aspect, in one possible implementation, the switching criteria include one or more of the following: different models based on the same model structure have the same priority; or different models associated with the same data have the same priority; or models corresponding to different tasks at the same network level have the same priority; or different models with the same execution quality level have the same priority; or the model corresponding to the first network level has a higher priority than the model corresponding to the second network level, and the first network level is lower than the second network level; or the model corresponding to the first execution quality level has a lower priority than the model corresponding to the second execution quality level, and the first execution quality level is lower than the second execution quality level, where the execution quality level includes the level of execution model duration. This implementation illustrates a specific form of the switching criteria. Under the constraints of the switching criteria, the efficiency of switching evaluation is significantly improved, which is beneficial to the implementation of the solution.

[0017] Based on the first aspect, one possible implementation further includes sending a second instruction message, which instructs the terminal device on its ability to perform a task or model. In this implementation, the second instruction message helps prevent the terminal device from performing tasks or models that are not within its capabilities, thus facilitating the implementation of the solution and improving the efficiency of the handover evaluation.

[0018] Based on the first aspect, one possible implementation method further includes: obtaining M models based on the identifiers of N tasks or the identifiers of M models, wherein the M models are sent by a network device or an over-the-top OTT server. This illustrates a specific implementation method for a terminal device to obtain M models, which is beneficial for the implementation of the solution.

[0019] Secondly, this disclosure provides a data processing method that can be used on the network side, for example, executed by a network device. The network device can be a device or apparatus with a chip, or a device or apparatus with integrated circuits, or a chip, chip system, module, or control unit in the aforementioned device or apparatus; specific details are not limited in this application. It should be noted that in this application, when referring to a network device, it can refer to the network device itself, or to the chip, functional module, or integrated circuit within the network device that performs the method provided in this application; specific details are not limited in this application. In the first aspect and its possible implementations, the method is described using the execution of the method by a network device as an example. The method includes: the network device determining first indication information, the first indication information indicating N tasks and M models corresponding to the N tasks, where N is less than M, and N and M are positive integers greater than 1; the network device sending the first indication information, the first indication information used for switching evaluation, the switching evaluation used to evaluate the execution quality of the N tasks based on different sets, the sets indicating N models among the M models used to execute the N tasks; and determining a first set, the first set being one of the different sets, the execution quality corresponding to the first set satisfying the evaluation conditions.

[0020] Based on the second aspect, in one possible implementation, the first indication information is specifically used to indicate the identifiers of N tasks and / or the identifiers of M models.

[0021] Based on the second aspect, in one possible implementation, the first indication information is also used to indicate the priority of some or all of the M models, and the method further includes: determining the priority of some or all of the M models according to the switching criteria.

[0022] Based on the second aspect, in one possible implementation, the first indication information is also used to indicate K sets, where K is a positive integer greater than 1. The method further includes: determining the priority of some or all of the M models according to the switching criteria; and determining K sets according to the priority of some or all of the M models.

[0023] Based on the second aspect, in one possible implementation, the first indication information is also used to indicate the priority of all or some subsets among the J subsets, the subsets are used to indicate at least one model among the M models, and J is a positive integer greater than 1; the method further includes: determining the priority of all or some subsets among the J subsets according to the switching criteria.

[0024] Based on the second aspect, in one possible implementation, the switching evaluation is also used to evaluate the execution quality of N tasks based on different sequences, where the sequence indicates an ordered structure composed of N models out of M models used to execute the N tasks; the method further includes: determining a first sequence, which is one of the different sequences, and the execution quality corresponding to the first sequence satisfies the evaluation conditions.

[0025] Based on the second aspect, in one possible implementation, the first indication information is also used to indicate the switching timing of all or some of the models in the M models and / or the activation status of all or some of the tasks in the N tasks. The switching timing is used to indicate the timing of ending the execution of the model. The switching timing includes the switching period or the operation that triggers the switching. The activation status is used to indicate whether the task is activated.

[0026] Based on the second aspect, in one possible implementation, the first indication information is carried in Radio Resource Control (RRC) signaling, Media Access Control (MAC) control element (CE), or Downlink Control Information (DCI).

[0027] Based on the second aspect, in one possible implementation, the switching criteria include one or more of the following: different models based on the same model structure have the same priority; or different models associated with the same data have the same priority; or models corresponding to different tasks at the same network level have the same priority; or different models with the same execution quality level have the same priority; or the priority of the model corresponding to the first network level is higher than the priority of the model corresponding to the second network level, and the first network level is lower than the second network level; or the priority of the model corresponding to the first execution quality level is lower than the priority of the model corresponding to the second execution quality level, and the first execution quality level is lower than the second execution quality level, where the execution quality level includes the level of the execution model duration.

[0028] Based on the second aspect, in one possible implementation, the method further includes: receiving second instruction information, the second instruction information being used to instruct the terminal device to perform a task or model; and determining first instruction information, including: determining the first instruction information based on the second instruction information.

[0029] Based on the second aspect, one possible implementation method also includes: sending M models.

[0030] Thirdly, this disclosure provides a data processing method that can be used on the network side, for example, executed by a network device. The network device can be a device or apparatus with a chip, or a device or apparatus with integrated circuits, or a chip, chip system, module, or control unit in the aforementioned device or apparatus; specific details are not limited in this application. It should be noted that in this application, when referring to a network device, it can refer to the network device itself, or to the chip, functional module, or integrated circuit within the network device that performs the method provided in this application; specific details are not limited in this application. In the first aspect and its possible implementations, the method is described using the execution of the method by a network device as an example. The method includes: the network device determining fifth indication information, wherein the fifth indication information is used to indicate R tasks and S models corresponding to the R tasks, where R is less than S, and R and S are positive integers greater than 1; the network device performing a switching evaluation based on the fifth indication information, the switching evaluation being used to evaluate the execution quality of the R tasks based on different sets, where the sets indicate R models among the S models used to execute the R tasks; and the terminal device determining a second set, which is one of the different sets, and the execution quality corresponding to the second set satisfies the evaluation conditions.

[0031] Based on the third aspect, in one possible implementation, the fifth instruction information is specifically used to indicate the identifiers of R tasks and / or the identifiers of S models.

[0032] Based on the third aspect, in one possible implementation, the fifth indication information is also used to indicate the priority of some or all of the S models, the priority being determined based on the switching criteria.

[0033] Based on the third aspect, in one possible implementation, the fifth indication information is also used to indicate T sets, where T is a positive integer greater than 1. The T sets are determined based on the priority of some or all of the S models, and the priority is determined based on the switching criteria.

[0034] Based on the third aspect, in one possible implementation, the fifth indication information is also used to indicate the priority of all or part of the subsets in the U subsets, the subsets are used to indicate at least one model in the S models, U is a positive integer greater than 1, and the priority is determined based on the switching criteria.

[0035] Based on the third aspect, in one possible implementation, the switching evaluation is also used to evaluate the execution quality of R tasks based on different sequences, where the sequence indicates an ordered structure composed of R models out of S models used to execute the R tasks; the method further includes: determining a second sequence, which is one of the different sequences, and the execution quality corresponding to the second sequence satisfies the evaluation conditions.

[0036] Based on the third aspect, in one possible implementation, the fifth indication information is also used to indicate the switching timing of all or some of the S models and / or the activation status of all or some of the R tasks. The switching timing is used to indicate the timing of ending the execution of the model. The switching timing includes the switching period or the operation that triggers the switching. The activation status is used to indicate whether the task is activated.

[0037] Based on the third aspect, in one possible implementation, the switching criteria include one or more of the following: different models based on the same model structure have the same priority; or different models associated with the same data have the same priority; or models corresponding to different tasks at the same network level have the same priority; or different models with the same execution quality level have the same priority; or the priority of the model corresponding to the first network level is higher than the priority of the model corresponding to the second network level, and the first network level is lower than the second network level; or the priority of the model corresponding to the first execution quality level is lower than the priority of the model corresponding to the second execution quality level, and the first execution quality level is lower than the second execution quality level, where the execution quality level includes the level of the execution model duration.

[0038] Based on the third aspect, in one possible implementation, the method further includes: determining sixth indication information, which is used to indicate the network device's ability to perform tasks or models; and determining fifth indication information, including: determining the fifth indication information based on the sixth indication information.

[0039] Fourthly, this disclosure provides a communication device, which can be a terminal device, a device, module, or chip within the terminal device, or a device compatible with the terminal device. In one design, the communication device may include modules corresponding to the methods / operations / steps / actions described in the first aspect. These modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the communication device may include a processing module and a communication module.

[0040] One example:

[0041] The processing module is used to perform a switching evaluation based on the first indication information. The switching evaluation is used to evaluate the execution quality of N tasks based on different sets. The set is used to indicate N models among M models used to execute N tasks. The first set is determined as one of the different sets, and the execution quality corresponding to the first set meets the evaluation conditions.

[0042] The communication module is used to receive first indication information, which indicates N tasks and M models corresponding to the N tasks, where N is less than M and N and M are positive integers greater than 1.

[0043] In one possible design, the first indication information is specifically used to indicate the identifiers of N tasks and / or M models.

[0044] In one possible design, the first indication information is also used to indicate the priority of some or all of the M models, the priority being determined based on switching criteria.

[0045] In one possible design, the first indication information is also used to indicate K sets, where K is a positive integer greater than 1. The K sets are determined based on the priority of some or all of the M models, and the priority is determined based on the switching criteria.

[0046] In one possible design, the first indication information is also used to indicate the priority of all or some subsets of J subsets, the subsets being used to indicate at least one model among M models, where J is a positive integer greater than 1, and the priority is determined based on switching criteria.

[0047] In one possible design, the switching evaluation is also used to evaluate the execution quality of N tasks based on different sequences, where the sequence indicates an ordered structure composed of N models out of M models used to execute the N tasks; the processing module is also used to determine a first sequence, which is one of the different sequences, and the execution quality corresponding to the first sequence satisfies the evaluation conditions.

[0048] In one possible design, the first indication information is also used to indicate the switching timing of all or some of the M models and / or the activation status of all or some of the N tasks. The switching timing is used to indicate when to end the execution of the model. The switching timing includes the switching period or the operation that triggers the switching. The activation status is used to indicate whether the task is active.

[0049] In one possible design, the first indication information is carried in Radio Resource Control (RRC) signaling, Media Access Control (MAC) control element (CE), or Downlink Control Information (DCI).

[0050] In one possible design, the switching criteria include one or more of the following: different models based on the same model structure have the same priority; or different models associated with the same data have the same priority; or models corresponding to different tasks at the same network level have the same priority; or different models with the same execution quality level have the same priority; or the model corresponding to the first network level has a higher priority than the model corresponding to the second network level, and the first network level is lower than the second network level; or the model corresponding to the first execution quality level has a lower priority than the model corresponding to the second execution quality level, and the first execution quality level is lower than the second execution quality level, where the execution quality level includes the level of the execution model duration.

[0051] In one possible design, the communication module is also used to send a second instruction message, which is used to instruct the terminal device on its ability to perform a task or model.

[0052] In one possible design, the communication module is also used to obtain M models based on the identifiers of N tasks or the identifiers of M models, wherein the M models are sent by a network device or an over-the-top OTT server.

[0053] Fifthly, this disclosure provides a communication device, which may be a network device, a device, module, or chip within a network device, or a device compatible with a network device. In one design, the communication device may include modules corresponding to the methods / operations / steps / actions described in the second aspect. These modules may be hardware circuits, software, or a combination of hardware circuits and software. In another design, the communication device may include a processing module and a communication module.

[0054] One example:

[0055] The communication module is used to send first indication information, which is used to perform a switching evaluation. The switching evaluation is used to evaluate the execution quality of N tasks based on different sets, and the sets are used to indicate N models among M models used to execute the N tasks.

[0056] The processing module is used to determine the first indication information, which indicates N tasks and M models corresponding to the N tasks, where N is less than M and N and M are positive integers greater than 1; and to determine the first set, which is one of the different sets, and the execution quality corresponding to the first set meets the evaluation conditions.

[0057] In one possible design, the first indication information is specifically used to indicate the identifiers of N tasks and / or M models.

[0058] In one possible design, the first indication information is also used to indicate the priority of some or all of the M models; the processing module is also used to determine the priority of some or all of the M models according to the switching criteria.

[0059] In one possible design, the first indication information is also used to indicate K sets, where K is a positive integer greater than 1; the processing module is also used to determine the priority of some or all of the M models according to the switching criteria; and to determine the K sets according to the priority of some or all of the M models.

[0060] In one possible design, the first indication information is also used to indicate the priority of all or some subsets among the J subsets, the subsets being used to indicate at least one model among the M models, where J is a positive integer greater than 1; the processing module is also used to determine the priority of all or some subsets among the J subsets according to the switching criteria.

[0061] In one possible design, the switching evaluation is also used to evaluate the execution quality of N tasks based on different sequences, where the sequence indicates an ordered structure composed of N models out of M models used to execute the N tasks; the processing module is also used to determine a first sequence, which is one of the different sequences, and the execution quality corresponding to the first sequence satisfies the evaluation conditions.

[0062] In one possible design, the first indication information is also used to indicate the switching timing of all or some of the M models and / or the activation status of all or some of the N tasks. The switching timing is used to indicate when to end the execution of the model. The switching timing includes the switching period or the operation that triggers the switching. The activation status is used to indicate whether the task is active.

[0063] In one possible design, the first indication information is carried in Radio Resource Control (RRC) signaling, Media Access Control (MAC) control element (CE), or Downlink Control Information (DCI).

[0064] In one possible design, the switching criteria include one or more of the following: different models based on the same model structure have the same priority; or different models associated with the same data have the same priority; or models corresponding to different tasks at the same network level have the same priority; or different models with the same execution quality level have the same priority; or the model corresponding to the first network level has a higher priority than the model corresponding to the second network level, and the first network level is lower than the second network level; or the model corresponding to the first execution quality level has a lower priority than the model corresponding to the second execution quality level, and the first execution quality level is lower than the second execution quality level, where the execution quality level includes the level of the execution model duration.

[0065] In one possible design, the communication module is further configured to receive second instruction information, which instructs the terminal device to perform a task or model; the processing module is specifically configured to determine the first instruction information based on the second instruction information.

[0066] In one possible design, the communication module is also used to send M models.

[0067] Sixthly, this disclosure provides a communication device, the communication device including a processor for implementing the method described in the first aspect above. The processor is coupled to a memory for storing instructions and data, and when the processor executes the instructions stored in the memory, it can implement the method described in the first aspect above. Optionally, the communication device may further include a memory; the communication device may also include a communication interface for communicating with other devices. Exemplarily, the communication interface may be a transceiver, circuit, bus, module, pin, or other type of communication interface.

[0068] In one possible device, the communication apparatus includes:

[0069] Memory, used to store instructions;

[0070] The processor is used to perform a switching evaluation based on the first instruction information. The switching evaluation is used to evaluate the execution quality of N tasks based on different sets. The sets are used to indicate N models among M models used to execute the N tasks. The processor determines a first set, which is one of the different sets, and the execution quality corresponding to the first set satisfies the evaluation conditions.

[0071] The communication interface is used to receive first indication information, which indicates N tasks and M models corresponding to the N tasks, where N is less than M and N and M are positive integers greater than 1.

[0072] In a seventh aspect, this disclosure provides a communication device, the communication device including a processor for implementing the method described in the second aspect above. The processor is coupled to a memory for storing instructions and data, and when the processor executes the instructions stored in the memory, it can implement the method described in the second aspect above. Optionally, the communication device may further include a memory; the communication device may also include a communication interface for communicating with other devices. Exemplarily, the communication interface may be a transceiver, circuit, bus, module, pin, or other type of communication interface.

[0073] In one possible device, the communication apparatus includes:

[0074] Memory, used to store instructions;

[0075] A communication interface is used to send first indication information. The first indication information is used to perform a switching evaluation. The switching evaluation is used to evaluate the execution quality of N tasks based on different sets. The sets are used to indicate N models among M models used to execute N tasks.

[0076] The processor is used to determine first indication information, which indicates N tasks and M models corresponding to the N tasks, where N is less than M and N and M are positive integers greater than 1; and to determine a first set, which is one of a different set, and the execution quality corresponding to the first set satisfies the evaluation conditions.

[0077] Eighthly, this disclosure provides a communication system including a terminal device and a network device. Specifically, the interaction between the terminal device and the network device can be understood with reference to the following:

[0078] The network device determines the first indication information, which is used to indicate N tasks and M models corresponding to the N tasks, where N is less than M and N and M are positive integers greater than 1;

[0079] The network device sends the first instruction information to the terminal device;

[0080] The terminal device performs a handover assessment based on the first instruction information. The handover assessment is used to evaluate the execution quality of N tasks based on different sets. The sets are used to indicate N models out of M models used to execute the N tasks.

[0081] The terminal device determines a first set, which is one of a different set, and the execution quality corresponding to the first set meets the evaluation criteria.

[0082] The network devices determine the first set.

[0083] The solution implemented on the terminal device side can be understood with reference to the design described in the first aspect. Similarly, the solution implemented on the network device side can be understood with reference to the design described in the second aspect, and this disclosure will not elaborate further on these aspects.

[0084] Ninthly, this disclosure provides a communication system including a communication device as described in the fourth or sixth aspect; and a communication device as described in the fifth or seventh aspect.

[0085] In a tenth aspect, this disclosure also provides a computer program that, when run on a computer, causes the computer to perform the method provided in any one of the first to third aspects described above.

[0086] In an eleventh aspect, this disclosure also provides a computer program product, including instructions that, when executed on a computer, cause the computer to perform the method provided in any one of the first to third aspects.

[0087] In a twelfth aspect, this disclosure also provides a computer-readable storage medium storing a computer program or instructions that, when executed on a computer, cause the computer to perform the method provided in any one of the first to third aspects.

[0088] In a thirteenth aspect, this disclosure also provides a chip for reading a computer program stored in a memory and executing the method provided in any one of the first to third aspects, or the chip includes circuitry for executing the method provided in any one of the first to third aspects.

[0089] In a fourteenth aspect, this disclosure also provides a chip system including a processor for supporting the implementation of the methods provided in any of the first to third aspects. In one possible design, the chip system further includes a memory for storing programs and data necessary for the device. The chip system may be composed of chips or may include chips and other discrete devices.

[0090] The effects of the solutions provided in any of the second to fourteenth aspects above can be referenced in the corresponding descriptions in the first aspect. Attached Figure Description

[0091] Figure 1 is a schematic diagram of a possible application framework in a communication system;

[0092] Figure 2 is a schematic diagram of another possible application framework in a communication system;

[0093] Figure 3a is a schematic diagram of a communication system applicable to an embodiment of this application;

[0094] Figure 3b is a schematic diagram of another communication system applicable to an embodiment of this application;

[0095] Figure 4 is a schematic diagram of an AI / ML application framework;

[0096] Figure 5 is a flowchart illustrating the data processing method provided in this disclosure;

[0097] Figure 6 is a schematic diagram of the structure of a communication device provided in this disclosure;

[0098] Figure 7 is a schematic diagram of another communication device provided in this disclosure. Detailed Implementation

[0099] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0100] The technical solutions provided in this application can be applied to various communication systems, such as: 5th generation (5G) or new radio (NR) systems, long term evolution (LTE) systems, LTE frequency division duplex (FDD) systems, LTE time division duplex (TDD) systems, wireless local area network (WLAN) systems, satellite communication systems, future communication systems such as 6th generation (6G) mobile communication systems, or integrated systems of multiple systems. The technical solutions provided in this application can also be applied to device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, machine-type communication (MTC), and Internet of Things (IoT) communication systems or other communication systems.

[0101] In a communication system, one network element can send signals to or receive signals from another network element. These signals can include information, signaling, or data. The term "network element" can also be replaced by an entity, network entity, device, communication equipment, communication module, node, communication node, etc. This disclosure uses a network element as an example. For instance, a communication system can include at least one terminal device and at least one network device. The network device can send downlink signals to the terminal device, and / or the terminal device can send uplink signals to the network device. It is understood that the terminal device in this disclosure can be replaced by a first network element, and the network device can be replaced by a second network element, both performing the corresponding data processing methods described in this disclosure.

[0102] In the embodiments of this application, the terminal device may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent, or user apparatus.

[0103] Terminal devices can be devices that provide voice / data, such as handheld devices with wireless connectivity, in-vehicle devices, etc. Currently, examples of terminals include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in self-driving vehicles, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), handheld devices with wireless communication capabilities, computing devices or other processing devices connected to wireless modems, wearable devices, terminal devices in 5G networks, or future public land mobile communication networks. Terminal devices in a network (PLMN), etc., are not limited to this in the embodiments of this application.

[0104] By way of example and not limitation, in this embodiment, the terminal device can also be a wearable device. Wearable devices, also known as wearable smart devices, are a general term for devices that utilize wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices that are worn directly on the body or integrated into the user's clothing or accessories. Wearable devices are not merely hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are feature-rich, large in size, and can achieve complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those that focus on a specific type of application function and require the use of other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign detection.

[0105] In this embodiment, the device for implementing the functions of the terminal device can be the terminal device itself, or it can be any device capable of supporting the terminal device in implementing those functions, such as a chip system. This device can be installed in or used in conjunction with the terminal device. In this embodiment, the chip system can be composed of chips or may include chips and other discrete components. This embodiment only uses the terminal device as an example to illustrate the device for implementing the functions of the terminal device, and does not constitute a limitation on the solution of this embodiment.

[0106] The network device in this application embodiment can be a device for communicating with a terminal device. This network device can also be called an access network device or a wireless access network device, such as a base station. In this application embodiment, the network device can refer to a radio access network (RAN) node (or device) that connects the terminal device to the wireless network. A base station can broadly encompass, or be replaced by, various names including: NodeB, evolved NodeB (eNB), next-generation NodeB (gNB), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master station, auxiliary station, motor slide retainer (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), radio unit (RU), positioning node, etc. A base station can be a macro base station, micro base station, relay node, donor node, or similar entities, or combinations thereof. A base station can also refer to a communication module, modem, or chip installed within the aforementioned equipment or apparatus. A base station can also be a mobile switching center, equipment performing base station functions in D2D, V2X, and M2M communications, network-side equipment in 6G networks, and equipment performing base station functions in future communication systems. A base station can support networks using the same or different access technologies. Optionally, a RAN node can also be a server, wearable device, vehicle, or in-vehicle equipment. For example, the access network equipment in vehicle-to-everything (V2X) technology can be a roadside unit (RSU). The embodiments of this application do not limit the specific technologies or equipment forms used in the network equipment.

[0107] Base stations can be fixed or mobile. For example, a helicopter or drone can be configured to act as a mobile base station, and one or more cells can move depending on the location of the mobile base station. In other examples, a helicopter or drone can be configured as a device to communicate with another base station.

[0108] In some deployments, the network devices mentioned in the embodiments of this application may be devices including CU, DU, or CU and DU, or devices with control plane CU nodes (central unit-control plane (CU-CP)) and user plane CU nodes (central unit-user plane (CU-UP)) and DU nodes. For example, the network devices may include gNB-CU-CP, gNB-CU-UP, and gNB-DU.

[0109] In some deployments, multiple RAN nodes collaborate to assist terminals in achieving wireless access, with different RAN nodes each implementing some of the base station's functions. For example, RAN nodes can be CUs, DUs, CU-CPs, CU-UPs, or RUs. CUs and DUs can be configured separately or included in the same network element, such as a BBU. RUs can be included in radio frequency equipment or radio frequency units, such as RRUs, AAUs, or RRHs.

[0110] RAN nodes can support one or more types of fronthaul interfaces, each corresponding to a DU and RU with different functions. If the fronthaul interface between the DU and RU is a common public radio interface (CPRI), the DU is configured to implement one or more baseband functions, and the RU is configured to implement one or more radio frequency functions. If the fronthaul interface between the DU and RU is another type of interface, relative to CPRI, some downlink and / or uplink baseband functions, such as, for downlink, precoding, digital beamforming (BF), or one or more of inverse fast Fourier transform (IFFT) / cyclic prefix addition (CP), are moved from the DU to the RU; and for uplink, digital beamforming (BF), or one or more of fast Fourier transform (FFT) / cyclic prefix removal (CP), are moved from the DU to the RU. In one possible implementation, the interface can be an enhanced common public radio interface (eCPRI). Under the eCPRI architecture, the segmentation between DU and RU differs, corresponding to different categories (Cat) of eCPRI, such as eCPRI Cat A, B, C, D, E, F.

[0111] Taking eCPRI Cat A as an example, for downlink transmission, the DU is configured to implement one or more functions before and after layer mapping (i.e., coding, rate matching, scrambling, modulation, and layer mapping), while other functions after layer mapping (e.g., resource element (RE) mapping, digital beamforming (BF), or one or more functions of inverse fast Fourier transform (IFFT) / adding cyclic prefix (CP)) are moved to the RU. For uplink transmission, the DU is configured to implement one or more functions before and after demapping (i.e., decoding, rate matching de-matching, descrambling, demodulation, inverse discrete Fourier transform (IDFT), channel equalization, and demapping), while other functions after demapping (e.g., digital BF or one or more functions of fast Fourier transform (FFT) / removing CP) are moved to the RU. It is understandable that the functional descriptions of the DU and RU corresponding to various types of eCPRI can be found in the eCPRI protocol, and will not be elaborated here.

[0112] In one possible design, the processing unit in the BBU used to implement baseband functions is called the baseband high (BBH) unit, and the processing unit in the RRU / AAU / RRH used to implement baseband functions is called the baseband low (BBL) unit.

[0113] In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meaning. For example, in an ORAN system, CU can also be called O-CU (open CU), DU can also be called O-DU, CU-CP can also be called O-CU-CP, CU-UP can also be called O-CU-UP, and RU can also be called O-RU. Any of the units among CU (or CU-CP, CU-UP), DU, and RU in this application can be implemented through software modules, hardware modules, or a combination of software modules and hardware modules.

[0114] In this embodiment, the apparatus for implementing the functions of a network device can be a network device itself; it can also be an apparatus capable of supporting the network device in implementing those functions, such as a chip system, hardware circuit, software module, or a hardware circuit plus a software module. This apparatus can be installed in the network device or used in conjunction with the network device. In this embodiment, the example of a network device being used to implement the functions of a network device is provided only and does not constitute a limitation on the solutions described in this embodiment.

[0115] Network devices and / or terminal devices can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; and they can also be deployed in the air on airplanes, balloons, and satellites. This application does not limit the scenario in which the network devices and terminal devices are located. Furthermore, terminal devices and network devices can be hardware devices, or software functions running on dedicated hardware or general-purpose hardware, such as virtualization functions instantiated on a platform (e.g., a cloud platform), or entities that include dedicated or general-purpose hardware devices and software functions. This application does not limit the specific form of the terminal devices and network devices.

[0116] In wireless communication networks, such as mobile communication networks, the services supported by the networks are becoming increasingly diverse, leading to increasingly diverse requirements. For example, networks need to support ultra-high speeds, ultra-low latency, and / or massive connectivity. This characteristic makes network planning, network configuration, and / or resource scheduling increasingly complex. Furthermore, as network functions become more powerful, such as supporting higher spectrum levels, supporting higher-order multiple-input multiple-output (MIMO) technologies, supporting beamforming, and / or supporting beam management, network energy efficiency has become a hot research topic. These new requirements, new scenarios, and new characteristics bring unprecedented challenges to network planning, operation, and efficient operation. To meet these challenges, artificial intelligence technology can be introduced into wireless communication networks to achieve network intelligence.

[0117] To support AI / ML technologies in wireless networks, AI / ML nodes may also be introduced into the network.

[0118] Optionally, AI / ML nodes can be deployed in one or more of the following locations within the communication system: access network devices, terminal devices, or core network devices, etc. Alternatively, AI / ML nodes can be deployed independently, for example, in a location other than any of the aforementioned devices, such as in the host or cloud server of an over-the-top (OTT) system. AI / ML nodes can communicate with other devices in the communication system, which can be, for example, one or more of the following: network devices, terminal devices, or core network elements, etc.

[0119] It is understood that this application does not limit the number of AI / ML nodes. For example, when there are multiple AI / ML nodes, these nodes can be divided based on function, such as different AI / ML nodes being responsible for different functions.

[0120] It can also be understood that AI / ML nodes can be independent devices, or they can be integrated into the same device to implement different functions. Alternatively, they can be network elements in hardware devices, software functions running on dedicated hardware, or virtualization functions instantiated on a platform (e.g., a cloud platform). This application does not limit the specific form of the aforementioned AI / ML nodes.

[0121] AI / ML nodes can be AI / ML network elements or AI / ML modules.

[0122] Figure 1 illustrates a possible application framework in a communication system. As shown in Figure 1, network elements in the communication system are connected via interfaces (e.g., NG, Xn) or over-the-air interfaces. These network element nodes, such as core network equipment, access network nodes (RAN nodes), terminals, or one or more devices in the OAM, are equipped with one or more AI / ML modules (only one is shown in Figure 1 for clarity). The access network node can be a single RAN node or can include multiple RAN nodes, for example, including CU and DU. The CU and / or DU can also be equipped with one or more AI / ML modules. Optionally, the CU can be further divided into CU-CP and CU-UP. One or more AI / ML models are configured in CU-CP and / or CU-UP.

[0123] The AI / ML module is used to implement corresponding AI / ML functions. AI / ML modules deployed in different network elements can be the same or different. Depending on the parameter configuration, the AI / ML module can implement different functions. The AI / ML module model can be configured based on one or more of the following parameters: structural parameters (e.g., at least one of the following: number of neural network layers, neural network width, inter-layer connections, neuron weights, neuron activation function, or bias in the activation function), input parameters (e.g., type and / or dimension of input parameters), or output parameters (e.g., type and / or dimension of output parameters). The bias in the activation function can also be referred to as the neural network bias.

[0124] An AI / ML module can have one or more models. A model can infer an output that includes one or more parameters. The learning, training, or inference processes of different models can be deployed on different nodes or devices, or they can be deployed on the same node or device.

[0125] Figure 2 illustrates a possible application framework in a communication system. As shown in Figure 2, the communication system includes a RAN intelligent controller (RIC). For example, the RIC can be the AI / ML modules 117 and 118 shown in Figure 1, used to implement AI / ML related functions. The RIC includes near-real-time RICs (near-RT RICs) and non-real-time RICs (non-RT RICs). Non-real-time RICs primarily process non-real-time information, such as data that is not sensitive to latency, with latency in the order of seconds. Near-real-time RICs primarily process near-real-time information, such as data that is relatively sensitive to latency, with latency in the order of tens of milliseconds.

[0126] The near real-time RIC is used for model training and inference. For example, it can be used to train an AI / ML model and then use that model for inference. The near real-time RIC can obtain network-side and / or terminal-side information from RAN nodes (e.g., CU, CU-CP, CU-UP, DU, and / or RU) and / or terminals. This information can be used as training data or inference data. Optionally, the near real-time RIC can deliver inference results to RAN nodes and / or terminals. Optionally, inference results can be exchanged between CUs and DUs, and / or between DUs and RUs. For example, the near real-time RIC delivers inference results to a DU, which then forwards them to an RU.

[0127] The non-real-time RIC is also used for model training and inference. For example, it can be used to train AI / ML models and then use those models for inference. The non-real-time RIC can obtain network-side and / or terminal-side information from RAN nodes (e.g., CU, CU-CP, CU-UP, DU, and / or RU) and / or terminals. This information can be used as training data or inference data, and the inference results can be delivered to RAN nodes and / or terminals. Optionally, inference results can be exchanged between CU and DU, and / or between DU and RU; for example, the non-real-time RIC delivers the inference results to the DU, which then forwards them to the RU.

[0128] The near real-time RIC and non-real-time RIC can also be set up as separate network elements. Optionally, the near real-time RIC and non-real-time RIC can also be part of other devices. For example, the near real-time RIC can be set in the RAN node (e.g., in CU, DU), while the non-real-time RIC can be set in the OAM, cloud server, core network device, or other network device.

[0129] Figure 3a is a schematic diagram of a communication system applicable to the data processing method of this application embodiment. As shown in Figure 3a, the communication system 100 may include at least one network device, such as network device 110 shown in Figure 3a; the communication system 100 may also include at least one terminal device, such as terminal device 120 and terminal device 130 shown in Figure 3a. Network device 110 and terminal devices (such as terminal device 120 and terminal device 130) can communicate via a wireless link. The communication devices in this communication system, for example, network device 110 and terminal device 120, can communicate via multi-antenna technology.

[0130] Figure 3b is a schematic diagram of another communication system applicable to the data processing method of this application embodiment. Compared with the communication system 100 shown in Figure 3a, the communication system 200 shown in Figure 3b further includes an AI / ML network element 140. The AI / ML network element 140 is used to perform AI / ML related operations, such as building a training dataset or training an AI / ML model.

[0131] In one possible implementation, network device 110 can send data related to the training of the AI / ML model to AI / ML network element 140, which then constructs a training dataset and trains the AI / ML model. For example, the data related to the training of the AI / ML model may include data reported by the terminal device. AI / ML network element 140 can send the results of operations related to the AI / ML model to network device 110, which then forwards them to the terminal device. For example, the results of operations related to the AI / ML model may include at least one of the following: a trained AI / ML model, model evaluation results, or test results. Exemplarily, a portion of the trained AI / ML model may be deployed on network device 110, and another portion on the terminal device. Alternatively, the trained AI / ML model may be deployed on network device 110. Or, the trained AI / ML model may be deployed on the terminal device.

[0132] It should be understood that Figure 3b is only used as an example of the AI / ML network element 140 being directly connected to the network device 110. In other scenarios, the AI / ML network element 140 can also be connected to the terminal device. Alternatively, the AI / ML network element 140 can be connected to both the network device 110 and the terminal device simultaneously. Alternatively, the AI / ML network element 140 can also be connected to the network device 110 through a third-party network element. This application embodiment does not limit the connection relationship between the AI / ML network element and other network elements.

[0133] AI / ML network element 140 can also be set as a module in network devices and / or terminal devices, for example, in network device 110 or terminal device shown in Figure 3a.

[0134] It should be noted that Figures 3a and 3b are simplified schematic diagrams for ease of understanding. For example, the communication system may also include other devices, such as wireless relay devices and / or wireless backhaul devices, which are not shown in Figures 3a and 3b. In practical applications, the communication system may include multiple network devices or multiple terminal devices. This application does not limit the number of network devices and terminal devices included in the communication system.

[0135] To facilitate understanding of the solutions in the embodiments of this application, the terms that may be involved in the embodiments of this application are explained below.

[0136] (1) AI / ML tasks

[0137] AI / ML tasks refer to specific functions achieved through AI / ML technologies, such as classification, regression, clustering, natural language processing, and computer vision tasks. In wireless communication networks, AI / ML tasks can specifically include tasks such as channel state information (CSI) feedback enhancement, beam management, and location accuracy enhancement.

[0138] It should be noted that in the embodiments of this application, AI / ML task can also be replaced with terms such as AI task, ML task, AI / ML function (AI / ML-enabled feature), AI function, ML function, task, function, etc. In actual applications, AI / ML task can also be replaced with other terms, and there is no limitation on this.

[0139] The following section introduces some AI / ML tasks applied to wireless communication networks.

[0140] CSI feedback enhancement refers to the task of improving or optimizing the CSI feedback process. In practical applications, network devices need to acquire the downlink channel CSI and, based on the CSI, determine the resources, modulation and coding scheme (MCS), and precoding configurations for scheduling downlink data channels of terminal devices. CSI can be understood as a type of channel information, reflecting channel characteristics and quality. Channel information can also be called channel response. By introducing AI / ML techniques into the CSI feedback process, it is possible to reduce feedback overhead, improve feedback accuracy, and predict channel conditions. For example, due to the sparsity of the channel in the frequency or time domain, compressed sensing methods can reduce the amount of information required for CSI feedback, thereby reducing CSI feedback overhead; by learning past CSI information, future channel conditions can be predicted, allowing for advance resource scheduling and beamforming.

[0141] Beam management refers to the task of improving or optimizing the communication process using beamforming technology. In practical applications, beamforming technology can concentrate signal energy, improving signal coverage and received power. The goal of beam management is to predict the optimal beam direction, improve beam selection accuracy, and reduce communication overhead and latency while ensuring communication quality. For example, beam management can specifically be a time-domain beam prediction task, that is, by introducing AI / ML technology and using past beam information or CSI, predicting the optimal beam direction within a certain period of the future, thereby adjusting the beam in advance, reducing beam switching latency, and improving communication continuity and stability. Besides this, beam management can also be other tasks, without limitation.

[0142] Location accuracy enhancement tasks refer to tasks that improve or optimize the accuracy of the positioning process. In practical applications, the location of terminal devices can be predicted based on parameters such as channel information. By introducing AI / ML technology, it is possible to achieve the goal of improving positioning accuracy in various complex scenarios, such as in non-line-of-sight (NLOS) environments. For example, the location of terminal devices can be predicted with higher accuracy based on channel information using AI / ML technology; or, higher-accuracy parameters can be determined using AI / ML technology, and then the location of terminal devices can be determined with higher accuracy using these parameters. Besides these, location accuracy enhancement tasks can also be other tasks, without limitation.

[0143] It should be noted that the above-mentioned AI / ML tasks are only illustrative examples. In actual applications, wireless communication networks may also include other AI / ML tasks.

[0144] (2) AI / ML model

[0145] An AI / ML model is an algorithm or computer program that implements AI / ML functions; or, an AI / ML model is an algorithm or computer program used to perform AI / ML tasks. Generally, an AI / ML model represents the mapping relationship between the model's input and output. Types of AI / ML models can include neural networks, linear regression models, decision tree models, support vector machines (SVMs), Bayesian networks, Q-learning models, or other AI / ML models. When applied to wireless communication networks, AI / ML models can specifically be models for implementing CSI feedback enhancement tasks, models for implementing beam management tasks, models for implementing location accuracy enhancement, etc.

[0146] It should be noted that in the embodiments of this application, AI / ML model can also be replaced with terms such as AI model, ML model, model, AI / ML algorithm, AI / ML computer program, etc. In practical applications, AI / ML model can also be replaced with other terms, and there is no limitation on this.

[0147] Furthermore, an AI / ML task can correspond to one or more AI / ML models.

[0148] For example, different AI / ML models can be used to perform the AI / ML task in different specific scenarios. For instance, different AI / ML models can be trained on different training datasets for different specific scenarios.

[0149] Alternatively, different AI / ML models can be used to perform the AI / ML task under different specific configurations. For example, different AI / ML models can have different accuracies, different inference latency, or be based on different model structures. Here, the inference accuracy of an AI / ML model refers to how close its predictions are to the actual results, and the inference latency refers to the time between the AI / ML model receiving input data and obtaining the output data.

[0150] Alternatively, different AI / ML models can be applied to different devices performing the AI / ML task; for example, the AI / ML model applicable to network devices is different from the AI / ML model applicable to terminal devices.

[0151] Alternatively, an AI / ML task can correspond to more AI / ML models, without limitation.

[0152] In addition, in the embodiments of this application, the AI / ML task may also include multiple subtasks of the task implemented by AI / ML technology, and each subtask may be implemented by one or more AI / ML models corresponding to the subtask, for example, it may be multiple serial subtasks for implementing a specific task.

[0153] (3) Model Structure

[0154] Model structure refers to the internal organization and components of an AI / ML model, defining the process by which the model receives input, processes information, and generates output. Regarding the components of an AI / ML model, the model structure can include neural networks, decision trees, support vector machines (SVMs), etc. In terms of the connection methods or data flow methods within the AI / ML model, the model structure can include feedforward neural networks (FNNs), recurrent neural networks (RNNs), Transformers (self-attention models), etc. Regarding the hierarchical structure of the AI / ML model, the model structure can include deep neural networks (DNNs) and convolutional neural networks (CNNs), etc.

[0155] It should be noted that the examples of different model structures mentioned above are for illustrative purposes only. In actual applications, there are many more model structures, which are not limited here.

[0156] (4) Quality of Service (QoS)

[0157] In wireless communication networks, Quality of Service (QoS) is a technology that ensures the quality of service and aims to provide better service capabilities for network communication services. For example, QoS can ensure the performance, reliability, security, and compliance with other QoS metrics by prioritizing critical traffic and allocating resources appropriately. These metrics may include bandwidth, latency, jitter, packet loss rate, robustness, and scalability.

[0158] In this application embodiment, QoS requirements refer to the requirements that need to be met in order to provide better service capabilities and ensure the quality of service of the network. These requirements may include, for example, bandwidth requirements, latency requirements, jitter requirements, packet loss rate requirements, and robustness requirements in a wireless communication network. Generally, QoS requirements can be specifically reflected in quantifiable parameters as the value range or threshold of each parameter. For example, bandwidth requirements in a wireless communication network can specifically be the value range or threshold of bandwidth. It should be noted that QoS requirements can also be replaced by other terms such as QoS objectives, QoS conditions, and QoS thresholds, without limitation.

[0159] Furthermore, applicable to wireless communication networks including AI / ML technologies, QoS can include Quality of AI Service (QoAIS) to provide better service capabilities for AI / ML services and ensure the quality of service for AI / ML services. Based on this, in the embodiments of this application, QoS requirements can also include QoAIS requirements, that is, the requirements that need to be met in order to provide better service capabilities and improve the quality of service for AI / ML services.

[0160] For example, in a wireless communication network incorporating AI / ML technology, QoS requirements can be performance requirements. These could include latency requirements (the duration of the AI / ML service), jitter requirements (the stability of the AI / ML service's duration), or bandwidth requirements (the bandwidth consumed during the AI / ML service). QoS requirements can also be resource requirements, such as storage resource requirements (the storage resources consumed during the AI / ML service) or computing resource requirements (the computing resources consumed by the AI / ML service). Furthermore, QoS requirements can also be other requirements that the AI / ML service needs to meet, such as accuracy requirements for AI / ML tasks, energy consumption requirements for AI / ML services, security requirements, and privacy requirements; there are no limitations on these.

[0161] It should be noted that QoS requirements may include one or more of the above requirements, and are not limited thereto. Furthermore, in the embodiments of this application, satisfying QoS requirements means satisfying all the requirements included in QoS requirements.

[0162] (5) Training dataset and inference data:

[0163] In the field of AI / ML, ground truth usually refers to data that is considered accurate or real.

[0164] The training dataset is used to train AI / ML models. It can include the input to the AI / ML model, or it can include both the input and the target output. Specifically, the training dataset includes one or more training data points, which can include training samples input to the AI / ML model, or the target output of the AI / ML model. The target output can also be referred to as the label, sample label, or labeled sample. The label is the ground truth value.

[0165] In the field of communications, training datasets can include simulation data collected through simulation platforms, experimental data collected from experimental scenarios, or measured data collected in actual communication networks. Because the geographical environment and channel conditions where the data is generated vary—for example, indoor / outdoor conditions, movement speed, frequency bands, or antenna configurations—the collected data can be categorized during acquisition. For instance, data with the same channel propagation environment and antenna configuration can be grouped together.

[0166] Model training essentially involves learning certain features from training data. In training AI / ML models (such as neural network models), the goal is to make the model's output as close as possible to the desired predicted value. This is achieved by comparing the network's current predictions with the target value and updating the weight vector of each layer based on the difference. (Of course, there's usually an initialization process before the first update, where parameters are pre-configured for each layer.) For example, if the network's prediction is too high, the weight vector is adjusted to predict a lower value. This adjustment continues until the AI / ML model can predict the target value or a value very close to it. Therefore, it's necessary to predefine "how to compare the difference between the predicted and target values," which is the loss function or objective function. These are important equations used to measure the difference between the predicted and target values. Taking the loss function as an example, a higher output value (loss) indicates a greater difference. Therefore, training an AI / ML model becomes a process of minimizing this loss, making the loss function value less than a threshold, or making the loss function value meet the target requirements. For example, if the AI / ML model is a neural network, adjusting the model parameters of the neural network includes adjusting at least one of the following parameters: the number of layers of the neural network, the width, the weights of the neurons, or the parameters in the activation function of the neurons.

[0167] Inference data can be used as input to a trained AI / ML model for inference. During the model inference process, the inference data is input into the AI / ML model, and the corresponding output, which is the inference result, is obtained.

[0168] (6) Design of AI / ML models:

[0169] The design of AI / ML models mainly includes the data collection phase (e.g., collecting training data and / or inference data), the model training phase, and the model inference phase. It can also further include the application of the inference results.

[0170] Figure 4 illustrates an AI / ML application framework.

[0171] In the aforementioned data collection phase, the data source provides both training and inference data. In the model training phase, the AI / ML model is obtained by analyzing or training the training data (trAI / MLning data) provided by the data source. The AI / ML model represents the mapping relationship between the model's input and output. Learning the AI / ML model through model training nodes is equivalent to learning the mapping relationship between the model's input and output using the training data. In the model inference phase, the AI / ML model trained in the model training phase is used to perform inference based on the inference data provided by the data source, yielding the inference result. This phase can also be understood as: inputting inference data into the AI / ML model, obtaining the output through the AI / ML model, which is the inference result. This inference result can indicate the configuration parameters used (executed) by the execution object, and / or the operations performed by the execution object. In the inference result application phase, the inference result is published. For example, the inference result can be uniformly planned by the actor entity, which can send the inference result to one or more execution objects (e.g., network devices or terminal devices) for execution. For example, the executing entity can also provide feedback on the model's performance to the data source, which facilitates subsequent model updates and training.

[0172] It is understood that a communication system may include network elements with artificial intelligence (AI) capabilities. The AI / ML model design-related steps described above can be performed by one or more network elements with AI capabilities. In one possible design, AI / ML functions (such as AI / ML modules or AI / ML entities) can be configured within existing network elements in the communication system to implement AI / ML-related operations, such as AI / ML model training and / or inference. For example, this existing network element could be a network device or a terminal device. Alternatively, in another possible design, an independent network element can be introduced into the communication system to perform AI / ML-related operations, such as training an AI / ML model. This independent network element can be called an AI / ML network element or an AI / ML node, etc., and this application embodiment does not limit the use of this name. Exemplarily, the AI / ML network element can be directly connected to network devices in the communication system, or it can be indirectly connected to network devices through a third-party network element. The third-party network element can be any core network element, such as an authentication management function (AMF) network element or a user plane function (UPF) network element, an operation administration and mAI / MLntenance (OAM) network element, a cloud server, or other network elements, without limitation. For example, this independent network element can be deployed on one or more of the following: the network device side, the terminal device side, or the core network side. Optionally, it can be deployed on a cloud server. For example, the communication system shown in Figure 3b introduces an AI / ML network element 140.

[0173] The training processes of different models can be deployed on different devices or nodes, or on the same device or node. Similarly, the inference processes of different models can be deployed on different devices or nodes, or on the same device or node. Taking a terminal device completing the model training phase as an example, after training the corresponding encoder and decoder, the terminal device sends the model parameters of the decoder to the network device. Taking a network device completing the model training phase as an example, after training the corresponding encoder and decoder, the network device can send the model parameters of the encoder to the terminal device and the model parameters of the decoder to the network device. Then, the model inference phase corresponding to the encoder is performed on the terminal device, and the model inference phase corresponding to the decoder is performed on the network device.

[0174] The model parameters can include one or more of the following: model structure parameters (e.g., number of layers, and / or weights), model input parameters (e.g., input dimension, number of input ports), or model output parameters (e.g., output dimension, number of output ports). The input dimension refers to the size of an input data set; for example, if the input data is a sequence, the corresponding input dimension indicates the length of the sequence. The number of input ports refers to the quantity of input data. Similarly, the output dimension refers to the size of an output data set; for example, if the output data is a sequence, the corresponding output dimension indicates the length of the sequence. The number of output ports refers to the quantity of output data.

[0175] The following further describes the AI / ML model switching process involved in the embodiments of this application, taking the execution of a certain AI / ML task by a terminal device as an example. In practical applications, network devices or other devices executing AI / ML tasks and switching AI / ML models can refer to similar descriptions below.

[0176] During the execution of an AI / ML task on a terminal device, this task corresponds to multiple AI / ML models. Based on QoS requirements and the current scenario, the terminal device can select and switch to one of the multiple AI / ML models to ensure that the execution quality of the AI / ML task meets QoS requirements and is adapted to the current scenario. For example, QoS requirements may include latency requirements, and the execution latency of the AI / ML model should meet these requirements (which can be within the range of the latency requirement). Alternatively, QoS requirements may include both latency and storage resource requirements, meaning that not only should the execution latency of the AI / ML model meet these latency requirements, but the storage space required by the AI / ML model during execution should also meet these storage resource requirements.

[0177] However, problems may arise when a terminal device urgently needs to execute multiple AI / ML tasks. For example, after completing one of multiple AI / ML tasks, the terminal device continues to execute the next. Based on QoS requirements and the current scenario, the terminal device can select and switch to one of the multiple AI / ML models included in the next AI / ML task, and the execution quality of the AI / ML task using that model meets QoS requirements and is suitable for the current scenario. However, even if the execution quality of each AI / ML task meets QoS requirements, the overall execution quality of executing multiple AI / ML tasks may still fail to meet those QoS requirements.

[0178] For example, QoS requirements may include latency requirements. Even if the execution latency of each AI / ML model meets the range indicated by the latency requirement, the overall execution time of multiple AI / ML tasks may still fail to meet the latency requirement. Alternatively, even if the sum of the execution latencies of each AI / ML model meets the range indicated by the latency requirement, other time-consuming processes outside of task execution (such as time spent acquiring or switching inference data) may still prevent the overall execution time of multiple AI / ML tasks from meeting the latency requirement. Or, QoS requirements may include jitter requirements. Even if the jitter of the execution latency of each AI / ML model (e.g., the difference between the maximum and minimum execution latency) meets the range indicated by the jitter requirement, the overall jitter of multiple AI / ML tasks may still fail to meet the jitter requirement.

[0179] Based on this, in the embodiments provided in this application, the terminal device can receive first indication information, wherein the first indication information can be used to indicate N tasks and M models corresponding to the N tasks (for example, specifically, the identifiers of the N tasks and the identifiers of the M models), and N is less than M, and N and M are positive integers greater than 1. Furthermore, the terminal device can perform a switching evaluation based on the first indication information, that is, evaluate the execution quality of the N tasks based on different sets, wherein the aforementioned sets can indicate the N models among the M models used to execute the N tasks. Based on this, the terminal device can determine a first set among the different sets, and the execution quality corresponding to the first set satisfies the evaluation conditions, wherein the evaluation conditions can be, for example, QoS requirements.

[0180] In this way, by evaluating multiple execution methods for executing N tasks (execution based on a set of tasks is considered one execution method), a first set that meets the evaluation criteria (QoS requirements) is determined. Subsequently, executing N tasks based on the first set that meets the evaluation criteria can effectively ensure execution quality and avoid failing to meet the QoS requirements of the wireless communication network when executing multiple tasks.

[0181] To make the technical solution of this application clearer and easier to understand, the following description is based on Figure 5 and an example of its application in the communication system shown in Figure 1.

[0182] Referring to Figure 5, a data processing method provided by an embodiment of this application is illustrated. This data processing method can be applied to the aforementioned communication system, such as an FDD communication scenario. Optionally, this data processing method can also be used in a TDD communication scenario, which is not limited in this disclosure.

[0183] It should be understood that in this application, the indication includes direct indication (also known as explicit indication) and implicit indication. Direct indication information A refers to information A being included; implicit indication information A refers to information A being indicated through the correspondence between information A and information B, and through direct indication information B. The correspondence between information A and information B can be predefined, pre-stored, pre-burned, or pre-configured.

[0184] It should be understood that in this application, information C is used to determine information D, including both situations where information D is determined solely based on information C and situations where it is determined based on information C and other information. Furthermore, information C can also be used to determine information D indirectly, for example, where information D is determined based on information E, and information E is determined based on information C.

[0185] Furthermore, in the embodiments of this application, "network element A sends information A to network element B" can be understood as network element B being the destination of information A or an intermediate network element in the transmission path between the destination and network element B, which may include sending information directly or indirectly to network element B. "Network element B receives information A from network element A" can be understood as network element A being the source of information A or an intermediate network element in the transmission path between the source and network element A, which may include receiving information directly or indirectly from network element A. Information may undergo necessary processing between the source and destination, such as format changes, but the destination can understand the valid information from the source. Similar expressions in this application can be understood in a similar way and will not be elaborated further here.

[0186] As shown in Figure 5, the data processing method includes the following steps:

[0187] Please refer to Figure 5, which is a flowchart illustrating a data processing method provided in an embodiment of this application. The method includes the following steps.

[0188] Step S501: The network device determines the first indication information, which is used to indicate N AI / ML tasks and M AI / ML models corresponding to the N AI / ML tasks, where N is less than M and N and M are positive integers greater than 1.

[0189] The network device can determine first indication information, which can indicate N AI / ML tasks and M AI / ML models corresponding to the N AI / ML tasks. Here, N is less than M, and N and M are positive integers greater than 1. That is, each of the N AI / ML tasks can correspond to one or more AI / ML models. Different AI / ML models can be used to execute the AI / ML task in different specific scenarios. Alternatively, different AI / ML models can be used to execute the AI / ML task under different specific configurations. Alternatively, different AI / ML models can be applicable to different devices executing the AI / ML task. Alternatively, different AI / ML models can also be AI / ML models trained based on other conditions; this is not limited.

[0190] The first indication information is used to indicate N AI / ML tasks and M AI / ML models corresponding to the N AI / ML tasks. Specifically, the first indication information can indicate the identifiers of the N AI / ML tasks and the identifiers of the M AI / ML models.

[0191] Alternatively, the first indication information can indicate the identifiers of N AI / ML tasks, so that the terminal device can determine the identifiers of M AI / ML models based on the correspondence between the identifiers of each AI / ML task and the identifiers of one or more corresponding AI / ML models, that is, to indirectly indicate the identifiers of M AI / ML models.

[0192] Alternatively, the first indication information can indicate the identifiers of M AI / ML models, so that the terminal device can determine the identifiers of N AI / ML tasks based on the correspondence between the identifiers of each AI / ML model and the identifiers of the AI / ML tasks it is used to implement, that is, to indirectly indicate the identifiers of N AI / ML tasks.

[0193] Alternatively, the first indication information may also indicate an index, so that the terminal device can determine the identifiers of the N AI / ML tasks and / or the M AI / ML models based on the correspondence between the index and the identifiers of the N AI / ML tasks and / or the identifiers of the M AI / ML models, that is, to indirectly indicate the identifiers of the N AI / ML tasks and / or the identifiers of the M AI / ML models.

[0194] It should be noted that the above-mentioned correspondences may be indicated by or carried in the first instruction information, or the above-mentioned correspondences may be pre-configured on the terminal device, without limitation.

[0195] This embodiment provides several implementation examples of network devices determining N AI / ML tasks and M AI / ML models.

[0196] As a first implementation example, the network device can determine N AI / ML tasks to be executed based on the needs of the current AI / ML tasks, and then determine M AI / ML models based on the correspondence between AI / ML tasks and AI / ML models.

[0197] Optionally, the requirement for executing the aforementioned AI / ML task may be, for example, a predefined AI / ML task execution cycle or execution time period. Alternatively, the requirement for executing the AI / ML task may be a triggering condition that prompts the terminal device to execute the AI / ML task, such as a specific scenario where the current scene meets the requirements for executing the AI / ML task. Other than these, the requirement for executing the aforementioned AI / ML task may also be other cases, which are not limited here.

[0198] As a second implementation example, the network device can receive a second instruction information sent by the terminal device, which is used to indicate the terminal device's ability to perform AI / ML tasks or AI / ML models. Then, the network device can determine N AI / ML tasks and M AI / ML models based on the second instruction information.

[0199] For example, the second indication information may specifically indicate one or more of the following: AI / ML performance parameters of the terminal device, network parameters of the terminal device, and scene information of the terminal device. Alternatively, the second indication information may also indicate other information, without limitation. Furthermore, the network device can match the AI / ML tasks or AI / ML models that the terminal device can execute based on the terminal device's ability to perform AI / ML tasks or AI / ML models.

[0200] Specifically, performance parameters may include the computing power of the terminal device, the accuracy of the terminal device in processing inference data, the latency parameters of the terminal device in processing data or executing AI / ML tasks (e.g., the duration of the terminal device executing a specific task or a single basic computation), or the throughput of the terminal device. Network parameters may include signal-to-noise ratio (SNR), adaptive modulation and coding (AMC) parameters, channel characteristics, the number of antennas of the terminal device, or scheduling information (e.g., time scheduling information or frequency scheduling information).

[0201] For example, the second indication information received by the network device indicates the precision of the processing and inference data of the terminal device. Therefore, the network device determines that the obtained AI / ML task or AI / ML model can be executed at the aforementioned precision. It should be noted that the parameter information required to execute the AI / ML task or AI / ML model can be pre-configured in the network device, or can be obtained or determined by the network device; there is no limitation on this.

[0202] In addition, since some AI / ML tasks or models require the cooperation of network devices with terminal devices for execution—for example, the network device can send the data required to execute the aforementioned AI / ML tasks or models (including signals, datasets, initial data, intermediate data, etc.), or the network device and the terminal device can jointly execute the aforementioned AI / ML tasks or models—the network device can also determine one or more of the following: AI / ML performance parameters, network parameters, and scene information. Furthermore, based on the determined information and combined with the second instruction information, the network device determines N AI / ML tasks and M AI / ML models.

[0203] It should be noted that the above implementation method for determining N AI / ML tasks and M AI / ML models by network devices is only an example. In actual applications, network devices can also determine N AI / ML tasks and M AI / ML models based on other methods.

[0204] Furthermore, the first indication information can also indicate the priority of some or all of the M AI / ML models, where the priority can be used to indicate the importance of the AI / ML model in achieving the goal of QoS requirements. For example, QoS requirements can be a range of duration values; therefore, the lower the latency of an AI / ML model, the better the execution quality of the AI / ML task performed using that AI / ML model meets the QoS requirements, and correspondingly, the higher the priority of that AI / ML model.

[0205] As one possible implementation, network devices can determine the priority of AI / ML models based on handover criteria. It is understood that handover criteria are one or more rules defined based on the goal of achieving QoS requirements. Optionally, handover criteria may include one or more of the following:

[0206] As a first optional criterion, different AI / ML models based on the same model structure have the same priority. For example, the terminal device can optimize one or more specific model structures, such as specifically optimizing the transformer model structure, to achieve better execution quality when executing AI / ML models based on these model structures. Optionally, since the optimization levels of different model structures vary among terminal devices, the one or more AI / ML models corresponding to the more optimized model structure have higher priority.

[0207] As a second optional criterion, different AI / ML models that correlate the same data have the same priority. It should be noted that the aforementioned data can include ordinary data, signals, signaling, datasets, data packets, or other data that can be used to transmit information. This data is not limited to inference data or intermediate or result data generated or relied upon during model computation. Furthermore, correlated data can mean that the data involved in executing different AI / ML models is the same or overlaps, or that the data involved in one AI / ML model can be easily determined based on the data involved in another AI / ML model (e.g., the data involved in one AI / ML model is an offset of the data involved in another AI / ML model). For example, if different AI / ML models that correlate the same data do not require time or bandwidth to determine (or acquire, generate, parse, etc.) the aforementioned same data during execution, the execution quality of the different AI / ML models executed by the terminal device will be higher. Optionally, the more different AI / ML models that correlate the same data, the higher the priority of these different AI / ML models.

[0208] As a third optional criterion, AI / ML models for different AI / ML tasks corresponding to the same network layer have the same priority. The network layer can be, for example, a layer within the Open System Interconnection (OSI) reference model, or a layer within other protocol stacks; there is no limitation on this. For example, different AI / ML tasks may correspond to functions at different network layers, but AI / ML models for different AI / ML tasks corresponding to the same network layer can have the same priority.

[0209] As a fourth optional criterion, the AI / ML model corresponding to the first network layer has a higher priority than the AI / ML model corresponding to the second network layer, where the first network layer is lower than the second network layer. That is, the lower the network layer, the more fundamental the functionality required by the AI / ML task corresponding to that network layer, and correspondingly, the higher the priority of the AI / ML model for that AI / ML task.

[0210] As a fifth optional criterion, different AI / ML models with the same execution quality level have the same priority. It should be noted that the execution quality of an AI / ML model refers to the execution quality of the AI / ML task performed by the terminal device using that AI / ML model, or the average execution quality of multiple executions of the AI / ML task by the terminal device using that AI / ML model. Execution quality can include the duration of AI / ML task execution, the bandwidth used for execution, and the storage or computing resources used for execution, etc. Furthermore, different AI / ML models with the same or similar execution quality have the same execution quality level, or different AI / ML models with execution quality within the same execution quality range have the same execution quality level.

[0211] As a sixth optional criterion, the priority of the AI / ML model corresponding to the first execution quality level is lower than the priority of the AI / ML model corresponding to the second execution quality level, where the first execution quality level is lower than the second execution quality level. That is, the lower the execution quality level of the AI / ML model, the worse the execution quality of the AI / ML task performed using that AI / ML model, and the less likely it is to achieve the QoS requirement goal; correspondingly, the lower the priority of the AI / ML model.

[0212] It should be noted that the switching criteria can include more criteria, and there is no limitation on this. For example, based on QoS requirements, different devices executing AI / ML models will have different priorities for the AI / ML models. For instance, different terminal devices may execute the same AI / ML model with different priorities, or terminal devices and network devices may execute the same AI / ML model with different priorities.

[0213] In addition, if the handover criteria include multiple optional criteria, during the process of determining the priority of the AI / ML model based on the handover criteria, the network device can also determine the priority of the AI / ML model based on the weight values ​​of different criteria. The weight values ​​indicate the degree to which different criteria influence the priority of the AI / ML model. For example, based on the optional first criterion, the priority of the AI / ML model is a first value; based on the optional second criterion, the priority of the AI / ML model is a second value. Therefore, the priority of the AI / ML model determined by the network device is the sum of the product of the first value and the weight value of the first criterion, and the product of the second value and the weight value of the second criterion. Furthermore, the weight values ​​of different criteria can be determined based on QoS requirements or other information.

[0214] Optionally, the switching criterion is a computer program or algorithm that enables the network device to determine the priority of all or some of the M AI / ML models based on the switching criterion. Alternatively, the switching criterion is a correspondence between AI / ML models and priorities, which can be pre-configured in the network device.

[0215] Optionally, the network device can send handover criteria to the terminal device. For example, first indication information can be used to indicate the handover criteria, or the handover criteria can be carried in other signals that the network device can send to the terminal device. Alternatively, the handover criteria can be pre-configured in the terminal device. Furthermore, the terminal device can determine the priority of all or some of the AI / ML models based on the handover criteria and the identifiers of the M AI / ML models.

[0216] As one possible implementation, the network device can determine K sets based on the priorities of all or part of the determined AI / ML models, where K is a positive integer greater than 1. First indication information can be used to indicate these K sets.

[0217] In this embodiment of the application, the set is used to indicate the N AI / ML models among the M AI / ML models used to perform N AI / ML tasks, that is, the N AI / ML models selected from the M AI / ML models to perform the N AI / ML tasks.

[0218] Optionally, the set may specifically indicate the identifiers of N AI / ML models. Alternatively, the set may specifically indicate an index, enabling the terminal device to determine the identifiers of the N AI / ML models based on the correspondence between the index and the identifiers of the N AI / ML models. Alternatively, the set may specifically indicate the identifiers of multiple subsets, each subset indicating at least one AI / ML model among the M AI / ML models.

[0219] Optionally, the first indication information can indirectly indicate the identifiers of M AI / ML models (the union of the identifiers of the AI / ML models indicated by the K sets) and the identifiers of N AI / ML tasks by indicating K sets.

[0220] It should be noted that in the embodiments of this application, the term "set" can also be replaced with terms such as "group" or "cluster," or other terms, without limitation.

[0221] The following are some implementation examples of how network devices determine K sets based on the priority of all or part of the AI / ML models.

[0222] As a first implementation example, for all or part of the N AI / ML tasks, the network device excludes several AI / ML models with lower priority from the multiple AI / ML models corresponding to the AI / ML tasks (for example, it could be X AI / ML models with lower priority, where X is a positive integer), and then combines the remaining AI / ML models to obtain K sets.

[0223] As a second implementation example, M AI / ML models can be combined into Y sets, where Y is a positive integer greater than K. Furthermore, the network device can select the K sets with the highest sum of priorities from the Y sets, or the network device can select the K sets with the highest average priority (the average priority of the AI / ML models with priorities in the set), thus obtaining the aforementioned K sets.

[0224] It should be noted that there are other ways to determine the K sets for network devices, and this is a limitation.

[0225] As one possible implementation, the first indication information can also indicate the priority of all or part of the J subsets, where J is a positive integer greater than 1. The subset can indicate at least one AI / ML model among the M AI / ML models. For example, the subset can indicate at least one AI / ML model among the M AI / ML models that needs to be executed sequentially. For instance, the AI / ML models corresponding to constellation design, sparse pilot, pilotless design, and channel estimation tasks can be combined into a subset; or, the AI / ML models corresponding to beam management, channel prediction, CSI feedback compression, AI / ML positioning, and AI / ML scheduling tasks can be combined into a subset.

[0226] Optionally, the subset may specifically indicate the identifier of at least one AI / ML model among the M AI / ML models. Alternatively, the subset may specifically indicate an index, such that the terminal device determines the identifier of at least one AI / ML model based on the correspondence between the index and the identifier of the aforementioned at least one AI / ML model.

[0227] Optionally, the first indication information can indirectly indicate the identifiers of M AI / ML models (the union of the identifiers of the AI / ML models indicated by the J subsets) and the identifiers of N AI / ML tasks by indicating J subsets.

[0228] It should be noted that in the embodiments of this application, the term "subset" can also be replaced with terms such as "subcombination" or "subcluster," or other terms, without limitation.

[0229] The following are several implementation examples of how network devices determine the priority of all or some subsets of J subsets based on handover criteria.

[0230] As a first implementation example, the network device can directly determine the priority of all or some subsets based on handover criteria. For example, if different subsets indicate the same overall execution quality level of the AI / ML model, then the different subsets have the same priority; if the overall execution quality level of the AI / ML model indicated by the first subset is lower than that indicated by the second subset, then the priority of the first subset is lower than that of the second subset.

[0231] As a second implementation example, the network device can determine the priority of all or some of the AI / ML models among the M AI / ML models according to the handover criteria. Furthermore, the priority of the subset can be the sum of the priorities of at least one AI / ML model indicated by the subset, or the priority of the subset can be the average of the priorities of at least one AI / ML model indicated by the subset (the average of the priorities of the AI / ML models with priorities among the at least one AI / ML model).

[0232] It should be noted that there are other ways for network devices to determine the priority of all or some subsets among J subsets, and this is subject to limitation.

[0233] As one possible implementation, the network device can determine Q sequences based on the priorities of all or part of the determined AI / ML models, where Q is a positive integer greater than 1. First indication information can be used to indicate the aforementioned Q sequences.

[0234] In this embodiment, the sequence indicates the ordered arrangement of N AI / ML models from M AI / ML models used to perform N AI / ML tasks. That is, N AI / ML models are selected in an orderly manner from the M AI / ML models to perform the N AI / ML tasks. For example, in the sequence, the order of the AI / ML models corresponding to the first network layer precedes the order of the AI / ML models corresponding to the second network layer, where the first network layer is lower than the second network layer. Therefore, the terminal device can execute the AI / ML tasks corresponding to the lower network layers first, according to the sequence indication; that is, the terminal device implements the lower-level functions first.

[0235] Optionally, the set may specifically indicate the identifiers of the ordered N AI / ML models. Alternatively, the set may specifically indicate an index, such that the terminal device determines the identifiers of the ordered N AI / ML models based on the correspondence between the index and the identifiers of the ordered N AI / ML models.

[0236] Optionally, the first indication information can indirectly indicate the identifiers of M AI / ML models (the union of the identifiers of the AI / ML models indicated by the Q sets) and the identifiers of N AI / ML tasks by indicating Q sequences.

[0237] It should be noted that in the embodiments of this application, the term "sequence" can also be replaced with terms such as "order," "queue," or "linked list," or other terms, without limitation.

[0238] Furthermore, the network device can determine Q sequences based on the priorities of all or some of the AI / ML models. For example, the network device can first determine multiple sets as described above, and then determine Q sequences based on the priorities of all or some of the N AI / ML models indicated by each set, where the priorities of the ordered N AI / ML models indicated by each sequence decrease sequentially. It should be noted that if the ordered N AI / ML models indicated by a sequence include AI / ML models without priorities, the position of the AI / ML models without priorities in the sequence can be arbitrary or a pre-indicated position (e.g., at the end of the sequence). Other implementations exist for the network device to determine the Q sequences, which are limited here.

[0239] Optionally, the first indication information may also indicate the activation status of all or some of the N AI / ML tasks, or the first indication information may indicate the activation status of the AI / ML tasks corresponding to all or some of the M AI / ML models. The activation status indicates whether an AI / ML task is activated, so that when an AI / ML task is activated, the terminal device executes the AI / ML task using the AI / ML model corresponding to that AI / ML task.

[0240] Optionally, the first indication information may also indicate the network layers corresponding to all or some of the N AI / ML tasks, or the first indication information may indicate the network layers corresponding to the AI / ML tasks of all or some of the M AI / ML models.

[0241] Optionally, the first indication information may also indicate the switching timing of all or some of the M AI / ML models. The switching timing indicates when the terminal device ends the execution of the AI / ML model. For example, the switching timing may be the cycle of AI / ML model switching or the time period used to execute the AI / ML model; or, the switching timing may be an operation or triggering condition that triggers the terminal device to switch, such as output result data or a specific scenario where the current scene meets the triggering condition indication. Other than these, the switching timing may also be other cases, which are not limited thereto.

[0242] It should be noted that the aforementioned first indication information is used to indicate information such as the identifiers of N AI / ML tasks and / or the identifiers of M AI / ML models (including but not limited to the priority of all or some of the M AI / ML models). Specifically, the first indication information may carry information such as the identifiers of the aforementioned N AI / ML tasks and / or the identifiers of the M AI / ML models, and there is no limitation on this.

[0243] Step S502: The terminal device receives the first instruction information.

[0244] After the network device determines the first indication information, it can send the first indication information to the terminal device, so that the terminal device can receive the first indication information to carry out the subsequent handover evaluation process.

[0245] As one possible implementation, the first indication information can be carried in a message sent by the network device to the terminal device. Optionally, all or part of the information in the first indication information can be carried in multiple messages. For example, the first indication information can be carried in radio resource control (RRC) signaling, or in a media access control layer control element (MAC CE) or downlink control information (DCI).

[0246] Optionally, the specific message in which the first indication information is carried can be based on the network layer corresponding to the AI / ML task indicated by the first indication information. For example, if the network layers corresponding to the AI / ML tasks indicated by the first indication information are all high-level, then the first indication information can be carried in RRC signaling or MAC CE. If the network layers corresponding to the AI / ML tasks indicated by the first indication information include not only high-level network layers but also the physical layer, then the first indication information can be carried in RRC signaling, MAC CE, or DCI.

[0247] The following sections describe the specific implementation methods of carrying the first indication information in RRC signaling, MAC CE, or DCI.

[0248] Firstly, the first indication information can be carried in the RRC signaling. Optionally, the RRC signaling can be configured as follows.

[0249] Among them, AITasksSchTypes indicates the network level of the AI / ML task, AITasksSchPeriod indicates the switching time of the AI / ML task, AITasksSchPeriod indicates the activation status of the AI / ML task, and AITasks priority indicates the priority of the AI / ML task.

[0250] `nn-ConfigurationInformation` indicates N AI / ML tasks; the `model structure info` included in `nn-ConfigurationInformation` indicates information about the AI / ML model structure; the `model candidates` included in `model structure info` indicates M AI / ML models, which may include, for example, Model 1-1 (model 1 corresponding to AI / ML task 1), Model 1-2, Model 2-1, and Model 2-2. In practical applications, `model candidates` can also indicate more AI / ML models.

[0251] Additionally, INTEGER indicates that the parameter is of integer type. SEQUENCE indicates that the parameter is of sequence type.

[0252] Secondly, the first instruction information can be carried in the MAC CE. Optionally, the MAC CE can refer to the configuration indicated in Table 1 below.

[0253] Table 1

[0254] ...

[0255] In this table, R is an optional field in the MAC CE, and C1 to C7 indicate the activation status of up to seven AI / ML tasks indicated by the MAC CE. AITaskID1 (AITaskID2 is similar to other AITaskIDs) indicates the identifier of AI / ML task number 1, and Model1-1 (Model 1-2 is similar to other Models) indicates the identifier of AI / ML model number 1 corresponding to AI / ML task number 1. In practical applications, Table 1 can be expanded to include more AI / ML tasks and more AI / ML models.

[0256] Thirdly, the first instruction information can be carried in the DCI. Optionally, the fields in the DCI can be directly used to carry the identifiers of the N AI / ML tasks and / or the identifiers of the M AI / ML models indicated by the first instruction information, or the fields in the DCI can also directly carry other information indicated by the first instruction information, such as the priority of the AI / ML models.

[0257] Optionally, the first indication information can be implicitly carried in the DCI. For example, the first indication information can be scrambled into the cyclic redundancy check (CRC) code of the DCI, so that the terminal device can obtain the first indication information by extracting the scrambled code of the CRC code of the DCI.

[0258] In addition, the first instruction information can also be carried in the DCI in other ways, without limitation.

[0259] It should be noted that the specific implementation of the first indication information carried in RRC signaling, MAC CE or DCI is only an example. In actual application, the first indication information can be carried with reference to other configurations, or the first indication information can be carried in other messages sent by the network device to the terminal device. There is no limitation on this.

[0260] Step S503: The terminal device performs a switching evaluation based on the first instruction information. The switching evaluation is used to evaluate the execution quality of N AI / ML tasks based on different sets. The sets are used to indicate the N AI / ML models among the M AI / ML models used to execute the N AI / ML tasks.

[0261] After receiving the first instruction information, the terminal device can perform a handover assessment based on the first instruction information.

[0262] Specifically, the terminal device performs a switching evaluation, which involves assessing the execution quality of N AI / ML tasks based on different sets. For example, the terminal device can traverse different sets and execute N AI / ML tasks using the N AI / ML models indicated by the traversed sets to determine the execution quality of the N AI / ML tasks. This execution quality is understood to be the execution quality corresponding to that set. Furthermore, the terminal device can determine whether the execution quality corresponding to the traversed sets meets the evaluation criteria to obtain the switching evaluation result.

[0263] The evaluation criteria can be, for example, QoS requirements, or other conditions, without limitation.

[0264] The description of the set in this embodiment can be found in step S501, and will not be repeated here. The specific description of the QoS requirements in this embodiment can be found in the previous introduction to QoS terminology, and will not be repeated here.

[0265] For example, the execution quality of N AI / ML tasks can specifically be the duration of executing the N AI / ML tasks; or, the execution quality can be the bandwidth (maximum bandwidth) used during the execution of the N AI / ML tasks; or, the execution quality can be the storage resources used during the execution of the N AI / ML tasks (e.g., the size of the storage space used); or, the execution quality can be the computing resources used during the execution of the N AI / ML tasks (e.g., the processor utilization rate of the terminal device); or, the execution quality can be the overall accuracy of executing the N AI / ML tasks; or, the execution quality can be the energy consumption during the execution of the N AI / ML tasks. Alternatively, the execution quality of the N AI / ML tasks can also be other indicators or parameters, which are not limited thereto.

[0266] Optionally, the terminal device can execute N AI / ML tasks multiple times based on a traversed set to determine the average execution quality corresponding to that set. Alternatively, the execution quality can be the fluctuation in duration when executing N AI / ML tasks multiple times (e.g., the difference between the maximum and minimum duration).

[0267] Optionally, the execution quality of the N AI / ML tasks may include one or more of the above, and there is no limitation on this.

[0268] It should be noted that the specific metrics for performance quality determined in the handover assessment can be the same as those indicated by the assessment conditions. For example, if the assessment condition is QoS requirements in a wireless communication network, and the QoS requirements indicate a threshold for the duration of AI / ML services, then the performance quality determined by the handover assessment is the duration of executing N AI / ML tasks. Furthermore, the performance quality meeting the assessment conditions could be, for example, that the duration of executing N AI / ML tasks is less than or equal to the threshold of the duration indicated by the QoS requirements.

[0269] There are several ways for a terminal device to determine whether the execution quality corresponding to the traversed set meets the evaluation conditions (QoS requirements).

[0270] As the first implementation method, QoS requirements can be sent to the terminal device or pre-deployed in the terminal device. Then, the terminal device can determine whether the execution quality corresponding to the traversed set meets the QoS requirements based on the QoS requirements.

[0271] It should be noted that QoS requirements can be sent to the terminal device by the core network equipment, or by the network equipment shown in Figure 5, or by other network equipment, or by a third-party server via the network equipment. Furthermore, before sending QoS requirements to the terminal device, the network equipment can also receive these QoS requirements through the core network equipment.

[0272] In the second implementation, the terminal device can send the execution quality corresponding to the traversed set to the network device. Then, the network device, based on QoS requirements, determines whether the execution quality corresponding to the traversed set meets the QoS requirements and sends a third indication to the terminal device. This third indication is used to indicate whether the execution quality corresponding to the set meets the QoS requirements.

[0273] It should be noted that the above implementation method is only an example. In actual applications, other implementation methods can be used to determine whether the execution quality corresponding to the traversed set meets the evaluation conditions (QoS requirements), and there are no limitations on this.

[0274] Before executing the N AI / ML tasks indicated by the first instruction information, the terminal device may obtain model files of M AI / ML models according to the first instruction information in order to execute the aforementioned N AI / ML tasks. Alternatively, the terminal may obtain the N AI / ML models indicated by a certain set while executing the N AI / ML tasks based on a certain set. Alternatively, if all or some of the aforementioned M AI / ML models have been pre-deployed on the terminal device, the terminal device may obtain the remaining undeployed AI / ML models in order to execute the aforementioned N AI / ML tasks.

[0275] It should be noted that the M AI / ML models acquired by the terminal device can be sent by the network device shown in Figure 5; or, they can be sent by other network devices; or, they can be sent by the core network device; or, they can be sent by an over-the-top (OTT) server, without any limitation.

[0276] Optionally, some of the M AI / ML models determined by the network device in step S501 may be unavailable. For example, the OTT server may not have stored some of the M AI / ML models. In this case, the terminal device can retrieve the missing AI / ML models from other devices, or the terminal device can perform a switching evaluation based on the remaining downloadable AI / ML models.

[0277] Additionally, if the terminal device performs a handover assessment based on the remaining downloadable AI / ML models, the terminal device also needs to send a fourth indication to the network device. This fourth indication indicates that the handover assessment is based on the remaining downloadable AI / ML models. This fourth indication may, for example, indicate the identifier of the remaining downloadable AI / ML models, the identifier of the missing AI / ML models, or other information; there are no limitations on this.

[0278] Furthermore, the terminal device can determine different sets to be traversed based on the M AI / ML models, so as to perform switching evaluations based on different sets. For example, according to the permutation and combination method, the M AI / ML models can be completely combined into Y sets, where Y is the number of combinations, and Y is greater than K.

[0279] Alternatively, the terminal device can determine different sequences to be traversed based on the M AI / ML models, allowing for switching evaluations based on different sequences. For example, according to the permutation and combination method, the M AI / ML models can be completely arranged into Z sequences, where Z is the number of permutations, and Z is greater than Q. Thus, compared to evaluating the set, the range of sequence-based evaluation is larger (generally, Z is greater than Y), the execution quality is more stable (the execution quality may differ under different orders within the set), and correspondingly, the accuracy of the evaluation results is higher.

[0280] Alternatively, the terminal device can determine different sets to be traversed based on priorities, such as K sets, so that switching evaluation can be performed based on these K sets. The specific implementation of determining the sets based on priorities can be found in the relevant description of step S501, and will not be repeated here. Thus, since the terminal device determines the K sets to be traversed based on priorities, compared to the complete Y sets that can be formed by M AI / ML models, the number of sets to be evaluated during the switching evaluation process is less, resulting in higher evaluation efficiency and lower evaluation overhead. It should be noted that the terminal device can determine the different sets to be traversed before traversal, or it can determine the sets to be traversed temporarily during the traversal process; there is no limitation on this.

[0281] Alternatively, the terminal device can determine different sequences to be traversed based on priority, such as Q sequences, so that switching evaluation can be performed based on different sequences. The specific implementation of determining the sequence based on priority can be found in the relevant description of step S501, and will not be repeated here. Therefore, since the terminal device determines the Q sequences to be traversed based on priority, compared to the Z sequences that can be completely arranged by M AI / ML models, the number of sequences to be evaluated during the switching evaluation process is less, resulting in higher evaluation efficiency and lower evaluation overhead. It should be noted that the terminal device can determine the different sequences to be traversed before traversal, or it can determine the sequences to be traversed temporarily during the traversal process; there is no limitation on this.

[0282] Alternatively, the terminal device can determine the different sets to be traversed based on the priority of all or some of the sub-combinations among the J sub-combinations. For details, please refer to the relevant explanations in steps S501 or S503 above, which will not be repeated here.

[0283] Alternatively, the first indication information may indicate K sets or Q sequences, and the terminal device can directly determine the K sets or Q sequences based on the first indication information.

[0284] Step S504a: The terminal device determines the first set, which is one of the different sets, and the execution quality corresponding to the first set meets the evaluation conditions.

[0285] Based on the handover assessment, the terminal device can determine the first set, and the execution quality corresponding to the first set meets the assessment conditions (QoS requirements). The first set is one of the different sets assessed during the handover assessment process.

[0286] Optionally, during the handover evaluation process, if the execution quality corresponding to the traversed set meets the evaluation conditions, the terminal device ends the handover evaluation, and the traversed set is the aforementioned first set.

[0287] Optionally, the terminal device can traverse all different sets to be traversed, and the first set can be the set whose execution quality best meets the evaluation criteria among the different sets to be traversed. It can be understood that, for example, the set that best meets the evaluation criteria can be the set whose execution quality differs the most from the threshold indicated by the QoS requirement compared to other sets.

[0288] Additionally, it should be noted that the terminal device can determine the first set based on the fourth indication information, that is, the fourth indication information indicates that the execution quality corresponding to the first set meets the evaluation conditions.

[0289] Alternatively, the terminal device determines the first set based on evaluation criteria. Optionally, after determining the first set, the terminal device may send an identifier of the first set and / or other information indicating the first set to the network device, with the aim of synchronizing the relevant information of the first set with the network device.

[0290] It should be noted that there is a correspondence between the first set and the M AI / ML models indicated by the first indication information. If the M AI / ML models change, the first set will also change accordingly.

[0291] Optionally, the terminal device can also determine a first sequence, which is one of a different sequence, and the execution quality corresponding to the first sequence meets the evaluation criteria. The relevant description of how the terminal device determines the first sequence can be found above and will not be repeated here.

[0292] In this way, by evaluating multiple execution methods for executing N AI / ML tasks (execution of AI / ML tasks based on a set is considered one execution method), a first set that meets the evaluation criteria (QoS requirements) is determined. Subsequently, executing N AI / ML tasks based on the first set that meets the evaluation criteria can effectively ensure execution quality and avoid failing to meet the QoS requirements of the wireless communication network when executing multiple AI / ML tasks.

[0293] Step S504b: The network device determines the first set.

[0294] Based on the handover assessment performed by the terminal device, the network device can determine the first set, and the execution quality corresponding to the first set meets the assessment conditions (QoS requirements). The first set is one of the different sets assessed during the handover assessment process.

[0295] Optionally, the network device can determine whether the execution quality corresponding to the traversed set meets the evaluation criteria based on the execution quality sent by the terminal device. Then, the network device can determine a first set that meets the evaluation criteria. Furthermore, after determining the first set, the network device can send a fourth indication message to the terminal device.

[0296] Alternatively, the network device may receive an identifier of the first set and / or other information used to indicate the first set sent by the terminal device to determine the first set.

[0297] Optionally, the network device can also determine a first sequence, which is one of a different sequence, and the execution quality corresponding to the first sequence meets the evaluation criteria. The relevant description of how the network device determines the first sequence can be found above and will not be repeated here.

[0298] In addition, the network device can also refer to step S504a to determine the relevant description of the first set, which will not be repeated here.

[0299] It should be noted that this embodiment uses a terminal device for handover evaluation as an example. In this embodiment or other embodiments, network devices can also be used for handover evaluation.

[0300] For example, the network device can determine fifth indication information, which indicates R tasks and S models corresponding to the R tasks, where R is less than S, and R and S are positive integers greater than 1. Then, based on the fifth indication information, the network device performs a handover evaluation, which assesses the execution quality of the R tasks based on different sets, where each set indicates one of the R models among the S models used to execute the R tasks. Based on this, the network device can determine a second set, which is one of the different sets, and the execution quality corresponding to the second set satisfies the evaluation conditions. The specific implementation of the above method can be found in steps S501 to S504b, and will not be repeated here.

[0301] Based on the same concept, referring to Figure 6, this disclosure provides a communication device 600, which includes a processing module 601 and a communication module 602. The communication device 600 can be a terminal device, or a communication device applied to or used in conjunction with a terminal device to implement a data processing method executed on the terminal device side; alternatively, the communication device 600 can be a network device, or a communication device applied to or used in conjunction with a network device to implement a data processing method executed on the network device side.

[0302] The communication module can also be called a transceiver module, transceiver, transceiver unit, or transceiver device. The processing module can also be called a processor, processing board, processing unit, or processing device. Optionally, the communication module is used to perform the sending and receiving operations on the terminal device side or network device side in the above method. The device in the communication module that implements the receiving function can be regarded as a receiving unit, and the device in the communication module that implements the sending function can be regarded as a sending unit. That is, the communication module includes a receiving unit and a sending unit.

[0303] When the communication device 600 is applied to a terminal device, the processing module 601 can be used to implement the processing functions of the terminal device in the example shown in FIG5, and the communication module 602 can be used to implement the sending and receiving functions of the terminal device in the example shown in FIG5. Alternatively, the communication device can be understood with reference to the fourth aspect of the invention and the possible designs in the fourth aspect.

[0304] When the communication device 600 is applied to a network device, the processing module 601 can be used to implement the processing functions of the network device in the example shown in FIG5, and the communication module 602 can be used to implement the sending and receiving functions of the network device in the example shown in FIG5. Alternatively, the communication device can be understood with reference to the fifth aspect of the invention and the possible designs in the fifth aspect.

[0305] Furthermore, it should be noted that the aforementioned communication module and / or processing module can be implemented through virtual modules. For example, the processing module can be implemented through software functional units or virtual devices, and the communication module can be implemented through software functions or virtual devices. Alternatively, the processing module or communication module can also be implemented through physical devices. For example, if the device is implemented using a chip / chip circuit, the communication module can be an input / output circuit and / or a communication interface, performing input operations (corresponding to the aforementioned receiving operation) and output operations (corresponding to the aforementioned sending operation); the processing module is an integrated processor, microprocessor, or integrated circuit.

[0306] The module division in this disclosure is illustrative and represents only one logical functional division. In actual implementation, other division methods are possible. Furthermore, the functional modules in the various examples of this disclosure can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0307] Based on the same technical concept, this disclosure also provides a communication device 700. For example, the communication device 700 may be a chip or a chip system. Optionally, in this disclosure, the chip system may be composed of chips or may include chips and other discrete devices.

[0308] The communication device 700 can be used to implement the function of any network element in the communication system described in the foregoing examples. The communication device 700 may include at least one processor 710. Optionally, the processor 710 is coupled to a memory, which may be located within the device, integrated with the processor, or located outside the device. For example, the communication device 700 may also include at least one memory 720. The memory 720 stores the computer programs, computer programs or instructions, and / or data necessary for implementing any of the above examples; the processor 710 may execute the computer programs stored in the memory 720 to complete the methods in any of the above examples.

[0309] The communication device 700 may also include a communication interface 730, through which the communication device 700 can interact with other devices. For example, the communication interface 730 may be a transceiver, circuit, bus, module, pin, or other type of communication interface. When the communication device 700 is a chip-based device or circuit, the communication interface 730 may also be an input / output circuit, capable of inputting information (or receiving information) and outputting information (or sending information). The processor may be an integrated processor, microprocessor, integrated circuit, or logic circuit, and the processor can determine the output information based on the input information.

[0310] The coupling in this disclosure refers to indirect coupling or communication connection between devices, units, or modules, which can be electrical, mechanical, or other forms, for information exchange between devices, units, or modules. The processor 710 may operate in conjunction with the memory 720 and the communication interface 730. This disclosure does not limit the specific connection medium between the processor 710, the memory 720, and the communication interface 730.

[0311] Optionally, referring to Figure 7, the processor 710, the memory 720, and the communication interface 730 are interconnected via a bus 740. The bus 740 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in Figure 7, but this does not indicate that there is only one bus or one type of bus.

[0312] In this disclosure, the processor can 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 this disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in this disclosure can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0313] In this disclosure, 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 this disclosure can also be a circuit or any other device capable of implementing a storage function for storing program instructions and / or data.

[0314] In one possible implementation, the communication device 700 can be applied to a network device. Specifically, the communication device 700 can be a network device itself, or it can be any device capable of supporting a network device and implementing the functions of the network device in any of the examples described above. The memory 720 stores computer programs (or instructions) and / or data that implement the functions of the network device in any of the examples described above. The processor 710 can execute the computer program stored in the memory 720 to perform the methods executed by the network device in any of the examples described above. Applied to a network device, the communication interface in the communication device 700 can be used to interact with a terminal device, sending information to or receiving information from the terminal device.

[0315] In another possible implementation, the communication device 700 can be applied to a terminal device. Specifically, the communication device 700 can be a terminal device or a device capable of supporting the terminal device and implementing the functions of the terminal device in any of the examples mentioned above. The memory 720 stores computer programs (or instructions) and / or data that implement the functions of the terminal device in any of the examples mentioned above. The processor 710 can execute the computer program stored in the memory 720 to complete the method executed by the terminal device in any of the examples mentioned above. Applied to a terminal device, the communication interface in the communication device 700 can be used to interact with network devices, sending information to or receiving information from network devices.

[0316] Since the communication device 700 provided in this example can be applied to a network device to complete the method executed by the network device, or applied to a terminal device to complete the method executed by the terminal device, the technical effects it can achieve can be referred to the above method example, and will not be repeated here.

[0317] Based on the above examples, this disclosure provides a communication system including a network device and a terminal device, wherein the network device and the terminal device can implement the data processing method provided in the example shown in Figure 5.

[0318] The technical solutions provided in this disclosure can be implemented, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented, in whole or in part, in the form of 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 this disclosure are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a terminal device, a network device, 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, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access 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.

[0319] In this disclosure, examples may be referenced to each other without logical contradiction. For example, methods and / or terms between method embodiments may be referenced to each other, functions and / or terms between device embodiments may be referenced to each other, and functions and / or terms between device examples and method examples may be referenced to each other.

[0320] Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from its scope. Therefore, if such modifications and variations fall within the scope of the claims of this disclosure and their equivalents, this disclosure is also intended to include such modifications and variations.

Claims

1. A data processing method, characterized by, The method is applied to a terminal device, and the method includes: Receive first indication information, the first indication information is used to indicate N tasks and M models corresponding to the N tasks, where N is less than M, and N and M are positive integers greater than 1; Based on the first indication information, a switching evaluation is performed. The switching evaluation is used to evaluate the execution quality of the N tasks based on different sets, where the sets are used to indicate N models among the M models used to execute the N tasks. A first set is determined, which is one of the different sets, and the execution quality corresponding to the first set meets the evaluation criteria.

2. The method of claim 1, wherein, The first indication information is specifically used to indicate the identifiers of the N tasks and / or the identifiers of the M models.

3. The method of claim 2, wherein, The first indication information is also used to indicate the priority of some or all of the M models, the priority being determined based on switching criteria.

4. The method of claim 2, wherein, The first indication information is also used to indicate K sets, where K is a positive integer greater than 1, and the K sets are determined based on the priority of some or all of the M models, the priority being determined based on switching criteria.

5. The method of claim 2, wherein, The first indication information is also used to indicate the priority of all or some subsets in the J subsets, the subsets being used to indicate at least one model among the M models, where J is a positive integer greater than 1, and the priority is determined based on switching criteria.

6. The method of claim 1, wherein, The switching evaluation is also used to evaluate the execution quality of the N tasks based on different sequences, the sequences being used to indicate an ordered structure composed of N models out of the M models used to execute the N tasks; The method further includes: A first sequence is determined, which is one of the different sequences, and the execution quality corresponding to the first sequence meets the evaluation criteria.

7. The method of claim 2, wherein, The first indication information is also used to indicate the switching timing of all or some of the M models and / or the activation status of all or some of the N tasks. The switching timing is used to indicate the timing of ending the execution of the model. The switching timing includes the switching period or the operation that triggers the switching. The activation status is used to indicate whether the task is activated.

8. The method according to any one of claims 1 to 7, characterized in that, The first indication information is carried in Radio Resource Control (RRC) signaling, Media Access Control (MAC) control element (CE), or Downlink Control Information (DCI).

9. The method of claim 3 or 4, wherein, The switching criteria include one or more of the following: Different models based on the same model structure have the same priority; or Different models that relate to the same data have the same priority; or Models for different tasks at the same network level have the same priority; or Different models with the same quality level have the same priority; or The priority of the model corresponding to the first network layer is higher than the priority of the model corresponding to the second network layer, and the priority of the first network layer is lower than that of the second network layer, or The priority of the model corresponding to the first execution quality level is lower than the priority of the model corresponding to the second execution quality level. The first execution quality level is lower than the second execution quality level. The execution quality level includes the level of execution time of the model.

10. The method of claim 1, wherein, The method further includes: Send a second instruction message, which is used to instruct the terminal device on its ability to perform a task or model.

11. The method of claim 2, wherein, The method further includes: Based on the identifiers of the N tasks or the identifiers of the M models, the M models are obtained, wherein the M models are sent by a network device or an over-the-top OTT server.

12. A data processing method, characterized by, The method is applied to a network device, and the method includes: First indication information is determined, which is used to indicate N tasks and M models corresponding to the N tasks, wherein N is less than M, and N and M are positive integers greater than 1; Send the first indication information, which is used to perform a switching evaluation. The switching evaluation is used to evaluate the execution quality of the N tasks based on different sets, where the sets are used to indicate N models among the M models used to execute the N tasks. A first set is determined, which is one of the different sets, and the execution quality corresponding to the first set meets the evaluation criteria.

13. The method of claim 12, wherein, The first indication information is specifically used to indicate the identifiers of the N tasks and / or the identifiers of the M models.

14. The method of claim 13, wherein, The first indication information is also used to indicate the priority of some or all of the M models, and the method further includes: Based on the switching criteria, the priority of some or all of the M models is determined.

15. The method of claim 13, wherein, The first indication information is further used to indicate K sets, where K is a positive integer greater than 1, and the method further includes: Based on the switching criteria, determine the priority of some or all of the M models; The K sets are determined based on the priority of some or all of the M models.

16. The method of claim 13, wherein, The first indication information is also used to indicate the priority of all or some subsets among the J subsets, the subsets being used to indicate at least one model among the M models, where J is a positive integer greater than 1; The method further includes: Based on the switching criteria, the priority of all or some subsets among the J subsets is determined.

17. The method of claim 12, wherein, The switching evaluation is also used to evaluate the execution quality of the N tasks based on different sequences, the sequences being used to indicate an ordered structure composed of N models out of the M models used to execute the N tasks; The method further includes: A first sequence is determined, which is one of the different sequences, and the execution quality corresponding to the first sequence meets the evaluation criteria.

18. The method of claim 13, wherein, The first indication information is also used to indicate the switching timing of all or some of the M models and / or the activation status of all or some of the N tasks. The switching timing is used to indicate the timing of ending the execution of the model. The switching timing includes the switching period or the operation that triggers the switching. The activation status is used to indicate whether the task is activated.

19. The method according to any one of claims 12 to 18, characterized in that, The first indication information is carried in Radio Resource Control (RRC) signaling, Media Access Control (MAC) control element (CE), or Downlink Control Information (DCI).

20. The method of claim 14 or 15, wherein, The switching criteria include one or more of the following: Different models based on the same model structure have the same priority; or Different models that relate to the same data have the same priority; or Models for different tasks at the same network level have the same priority; or Different models with the same quality level have the same priority; or The priority of the model corresponding to the first network layer is higher than the priority of the model corresponding to the second network layer, and the priority of the first network layer is lower than that of the second network layer, or The priority of the model corresponding to the first execution quality level is lower than the priority of the model corresponding to the second execution quality level. The first execution quality level is lower than the second execution quality level. The execution quality level includes the level of execution time of the model.

21. The method of claim 12, wherein, The method further includes: Receive a second instruction message, the second instruction message being used to instruct the terminal device on its ability to perform a task or model; The determination of the first indication information includes: The first instruction information is determined based on the second instruction information.

22. The method of claim 12, wherein, The method further includes: Send the M models.

23. A communications device, characterized by Used to implement the method as described in any one of claims 1-11.

24. A communications device, characterized by Used to implement the method as described in any one of claims 12-22.

25. A communications device, characterized by include: A processor coupled to a memory, the processor being configured to invoke computer program instructions stored in the memory to perform the method as described in any one of claims 1-11.

26. A communications device, characterized by include: A processor coupled to a memory, the processor being configured to invoke computer program instructions stored in the memory to perform the method as described in any one of claims 12-22.

27. A communication system, characterized by It includes the communication device as described in claim 23 or 25, and the communication device as described in claim 24 or 26.

28. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1-11 or the method as described in any one of claims 12-22.

29. A computer program product, characterised in that, Includes instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1-11 or the method as described in any one of claims 12-22.

Images