CSI processing method, apparatus, and communication system

Abstract

Embodiments of the present application provide a CSI processing method, an apparatus, and a communication system. The method comprises: a terminal device sends first information to a network device, the first information being at least used for reporting processing complexity and / or storage capacity related to an AI / ML functionality / model; the terminal device receives second information from the network device, the second information being used for configuring and / or triggering a CSI report related to the AI / ML functionality / model; and the terminal device calculates / determines, at least on the basis of the first information, a processing complexity metric and / or storage capacity metric related to the CSI report.

Description

CSI processing methods, devices, and communication systems Technical Field

[0001] The embodiments of this application relate to the field of communication technology. Background Technology

[0002] The application of artificial intelligence (AI) and machine learning (ML) technologies (AI / ML) in the NR air interface can optimize the performance of wireless communication systems, improve network efficiency and user experience. It is one of the important study items (SI) in 3GPP Release 18 (Rel-18) and was transformed into a work item (WI) in 3GPP Release 19 (Rel-19). The related standardization work has just begun.

[0003] Currently, the AI / ML use cases being discussed at the 3GPP RAN WG1 (hereinafter referred to as RAN1) meeting include beam management, positioning, CSI prediction, and CSI compression. The terminal device (UE) uses a model deployed on the terminal side for inference and / or performance monitoring. The results of inference and / or monitoring can be reported to the network (NW) or base station (gNB) in the form of a CSI report within the CSI framework.

[0004] It should be noted that the above introduction to the technical background is only for the purpose of providing a clear and complete explanation of the technical solutions of this application and facilitating understanding by those skilled in the art. It should not be assumed that these technical solutions are known to those skilled in the art simply because they have been described in the background section of this application. Summary of the Invention

[0005] The inventors discovered that AI / ML functions / models are deployed on separate hardware, and the hardware's processing and storage capabilities are limited. The current CSI framework cannot handle the processing complexity and storage capacity issues related to AI / ML functions.

[0006] To address at least one of the above-mentioned problems, embodiments of this application provide a CSI processing method, apparatus, and communication system.

[0007] According to one aspect of the embodiments of this application, a channel state information (CSI) processing method is provided, including:

[0008] The terminal device sends first information to the network device, the first information being used at least to report the processing complexity and / or storage capacity related to the AI / ML function / model;

[0009] The terminal device receives second information from the network device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and

[0010] The terminal device at least calculates / determines a processing complexity metric and / or a storage capacity metric related to the Channel State Information (CSI) report based on the first information.

[0011] According to another aspect of the embodiments of this application, a CSI processing apparatus is provided, comprising:

[0012] A transmitter that sends first information to a network device, the first information being used at least to report processing complexity and / or storage capacity related to AI / ML functionality / model;

[0013] A receiver that receives second information from a network device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and

[0014] The processor calculates / determines at least based on the first information a processing complexity metric and / or storage capacity metric related to the Channel State Information (CSI) report.

[0015] According to another aspect of the embodiments of this application, a channel state information (CSI) processing method is provided, including:

[0016] The network device receives first information sent by the terminal device, the first information being used at least to report the processing complexity and / or storage capacity related to the AI / ML function / model;

[0017] The network device sends second information to the terminal device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model;

[0018] The first information is used by the terminal device to calculate / determine processing complexity and / or storage capacity metrics related to the Channel State Information (CSI) report.

[0019] According to another aspect of the embodiments of this application, a CSI processing apparatus is provided, comprising:

[0020] A receiver receives first information sent by a terminal device, the first information being used at least to report processing complexity and / or storage capacity related to AI / ML functionality / model;

[0021] A transmitter that sends second information to the terminal device, the second information being used to configure and / or trigger a channel state information (CSI) report related to the AI / ML function / model;

[0022] The first information is used by the terminal device to calculate / determine processing complexity metrics and / or storage capacity metrics related to the Channel State Information (CSI) report.

[0023] According to another aspect of the embodiments of this application, a communication system is provided, comprising:

[0024] A network device receives first information from a terminal device, the first information being used at least to report processing complexity and / or storage capacity related to an AI / ML function / model; sends second information to the terminal device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and

[0025] The terminal device calculates / determines at least based on the first information a processing complexity metric and / or storage capacity metric related to the Channel State Information (CSI) report.

[0026] One of the beneficial effects of the embodiments of this application includes: calculating / measuring the processing complexity and / or storage capacity of CSI reports associated with AI / ML functions / models, thereby enabling the handling of processing complexity and / or storage capacity issues related to AI / ML functions within the CSI framework, improving the performance and efficiency of AI / ML, and optimizing the performance of wireless communication systems.

[0027] Specific embodiments of this application are disclosed in detail with reference to the following description and accompanying drawings, indicating how the principles of this application can be adopted. It should be understood that the embodiments of this application are not limited in scope. Within the spirit and scope of the appended claims, embodiments of this application include many changes, modifications, and equivalents.

[0028] Features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, combined with features in other embodiments, or substituted for features in other embodiments.

[0029] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, whole, step, or component, but does not exclude the presence or addition of one or more other features, wholes, steps, or components. Attached Figure Description

[0030] The elements and features described in one drawing or embodiment of this application may be combined with elements and features shown in one or more other drawings or embodiments. Furthermore, in the drawings, similar reference numerals denote corresponding parts in several drawings and can be used to indicate corresponding parts used in more than one embodiment.

[0031] Figure 1 is a schematic diagram of a communication system according to an embodiment of this application;

[0032] Figure 2 is a schematic diagram of a CSI processing method according to an embodiment of this application;

[0033] Figure 3 is a schematic diagram of AI / ML for CSI prediction according to an embodiment of this application;

[0034] Figure 4 is a schematic diagram of CSI reporting according to an embodiment of this application;

[0035] Figure 5 is an example diagram of the first information in an embodiment of this application;

[0036] Figure 6 is a schematic diagram of a CSI processing method according to an embodiment of this application;

[0037] Figure 7 is a schematic diagram of a CSI processing device according to an embodiment of this application;

[0038] Figure 8 is a schematic diagram of a CSI processing device according to an embodiment of this application;

[0039] Figure 9 is a schematic diagram of a terminal device according to an embodiment of this application;

[0040] Figure 10 is a schematic diagram of a network device according to an embodiment of this application. Detailed Implementation

[0041] Referring to the accompanying drawings, the foregoing and other features of this application will become apparent from the following description. Specific embodiments of this application are specifically disclosed in the description and drawings, illustrating partial implementations in which the principles of this application may be employed. It should be understood that this application is not limited to the described embodiments; rather, it includes all modifications, variations, and equivalents falling within the scope of the appended claims.

[0042] In the embodiments of this application, the terms "first," "second," etc., are used to distinguish different elements by name, but do not indicate the spatial arrangement or chronological order of these elements, and these elements should not be limited by these terms. The term "and / or" includes any one or more of the terms listed in association and all combinations thereof. The terms "comprising," "including," "having," etc., refer to the presence of the stated features, elements, components, or assemblies, but do not exclude the presence or addition of one or more other features, elements, components, or assemblies.

[0043] In the embodiments of this application, the singular forms "a," "the," etc., including the plural forms, should be broadly understood as "a kind" or "a class" rather than limited to the meaning of "an." Furthermore, the term "the" should be understood to include both the singular and plural forms, unless the context explicitly indicates otherwise. Additionally, the term "according to" should be understood as "at least partially based on…," and the term "based on" should be understood as "at least partially based on…," unless the context explicitly indicates otherwise.

[0044] In the embodiments of this application, the term "communication network" or "wireless communication network" may refer to a network that conforms to any of the following communication standards, such as New Radio (NR), Long Term Evolution (LTE), LTE-Advanced (LTE-A), Wideband Code Division Multiple Access (WCDMA), High-Speed ​​Packet Access (HSPA), etc.

[0045] Furthermore, communication between devices in a communication system can be carried out according to communication protocols at any stage, including but not limited to the following communication protocols: 1G (generation), 2G, 2.5G, 2.75G, 3G, 4G, 4.5G and 5G, New Radio (NR), future 6G, etc., and / or other currently known or future communication protocols.

[0046] In the embodiments of this application, the term "network device" refers, for example, to a device in a communication system that connects a terminal device to a communication network and provides services to that terminal device. Network devices may include, but are not limited to, the following devices: base station (BS), access point (AP), transmission reception point (TRP), broadcast transmitter, mobile management entity (MME), gateway, server, radio network controller (RNC), base station controller (BSC), etc.

[0047] Base stations can include, but are not limited to: NodeBs (or NBs), evolved NodeBs (eNodeBs or eNBs), and 5G base stations (gNBs), IAB hosts, etc. They can also include Remote Radio Heads (RRHs), Remote Radio Units (RRUs), relays, or low-power nodes (e.g., femeto, pico, etc.). The term "base station" can encompass some or all of their functions, and each base station can provide communication coverage to a specific geographic area. The term "cell" can refer to a base station and / or its coverage area, depending on the context in which the term is used.

[0048] In the embodiments of this application, the terms "User Equipment" (UE) or "Terminal Equipment" (TE) refer, for example, to a device that accesses a communication network and receives network services through a network device. A terminal device can be fixed or mobile, and may also be referred to as a mobile station (MS), terminal, subscriber station (SS), access terminal (AT), station, etc.

[0049] The terminal device may include, but is not limited to, the following devices: cellular phone, personal digital assistant (PDA), wireless modem, wireless communication device, handheld device, machine-type communication device, laptop computer, cordless phone, smartphone, smartwatch, digital camera, etc.

[0050] For example, in scenarios such as the Internet of Things (IoT), terminal devices can also be machines or devices for monitoring or measurement, such as including but not limited to: machine-type communication (MTC) terminals, vehicle communication terminals, device-to-device (D2D) terminals, machine-to-machine (M2M) terminals, and so on.

[0051] Furthermore, the terms "network side" or "network equipment side" refer to one side of the network, which can be a base station or include one or more network devices as described above. The terms "user side," "terminal side," or "terminal equipment side" refer to the side of the user or terminal, which can be a UE or include one or more terminal devices as described above. Unless otherwise specified, "equipment" can refer to either network equipment or terminal equipment.

[0052] The following examples illustrate the scenarios of embodiments of this application, but this application is not limited thereto.

[0053] Figure 1 is a schematic diagram of a communication system according to an embodiment of this application, illustrating the case of a terminal device and a network device as examples. As shown in Figure 1, the communication system 100 may include a network device 101 and terminal devices 102 and 103. For simplicity, Figure 1 only illustrates the case of two terminal devices and one network device, but the embodiments of this application are not limited to this.

[0054] In this embodiment of the application, network device 101 and terminal devices 102 and 103 can transmit existing services or services that can be implemented in the future. For example, these services may include, but are not limited to: enhanced mobile broadband (eMBB), massive machine-type communication (mMTC), and ultra-reliable and low-latency communication (URLLC), etc.

[0055] It is worth noting that Figure 1 shows that both terminal devices 102 and 103 are within the coverage area of ​​network device 101, but this application is not limited to this. Both terminal devices 102 and 103 may be outside the coverage area of ​​network device 101, or one terminal device 102 may be within the coverage area of ​​network device 101 while the other terminal device 103 may be outside the coverage area of ​​network device 101.

[0056] In the embodiments of this application, higher-layer signaling may be, for example, Radio Resource Control (RRC) signaling; for example, referred to as an RRC message, including MIB, system information, dedicated RRC messages; or referred to as an RRC information element. Higher-layer signaling may also be, for example, Medium Access Control (MAC) signaling; or referred to as a MAC control element. However, this application is not limited to these.

[0057] In embodiments of this application, the terms "AI / ML model" and "AI / ML function" may be used interchangeably in certain unambiguous cases. For example, "functionality" may be used to refer to a set of configurations corresponding to one or more AI / ML models. Table 1 shows an exemplary description of functionality:

[0058] [Revised according to Detailed Rules 26, May 2025] Table 1

[0059] As indicated by the above agreement, activated AI / ML functionalities in RAN1 can be based on the CSI framework. Within the current protocol's CSI framework, CSI reporting involves a series of processes, including but not limited to: receiving CSI reporting configuration information, CSI-RS measurement, CSI processing, and sending a CSI report. Specifically, after receiving the CSI reporting configuration information, the UE, upon receiving a set of CSI-RS signals, processes them into a CSI report according to the requirements of the CSI reporting configuration information, and then sends the CSI report to the network side.

[0060] From the perspective of UE processing capabilities, the complexity that a UE can process simultaneously is limited, and the UE's storage space is also limited. For different configurations of CSI reports (corresponding to the higher-layer information element CSI-ReportConfig), parameter O is used. CPU To measure processing complexity, it represents the number of CSI processing units (CPUs) required to process a CSI report within the processing time. At the same point in time, the O(n) of all CSI reports being processed corresponds to... CPU The sum does not exceed the maximum value N. CPU , where N CPUThe specific value depends on the UE's capabilities.

[0061] For example, the UE indicates / reports the (maximum) number N of CSI calculations / processes that can be performed simultaneously within a single component carrier via the higher-layer parameter simultaneousCSI-ReportsPerCC (or simultaneousCSI-SubReportsPerCC-r18). CPU If the UE supports simultaneous N... CPU If a CSI is calculated / processed, then it is said to have N CPU Each CSI Processing Unit is used to process CSI reports.

[0062] For example, if L CPUs are used for CSI calculations / processing related to CSI reports during a certain OFDM symbol period, then L CPUs are said to be occupied. In this case, N CPUs are still available during that OFDM symbol period. CPU -L CPUs are unoccupied. If the remaining N... CPU Within -L unused OFDM symbol cycles, N CSI reports begin to occupy their respective CPUs (where each CSI report has n = 0, ..., N-1, corresponding to the CPU occupied). CPUs, of which The value of is related to the number of CSI-RS associated with the CSI report and its time-domain characteristics. If the UE does not need to update NM of the lowest priority CSI reports, the specific priority definition can be found in the relevant protocol. Here, M (satisfying 0 ≤ M ≤ N) satisfies the inequality... The largest integer value.

[0063] Regarding the storage capacity issue when the UE processes CSI reports, the current standard does not have a clear limit on similar processing complexity, but it will count the referenced CSI-RS resources and CSI-RS ports.

[0064] The deployment of AI / ML functions / models on new hardware differs from existing CSI processing. The way to measure the processing complexity of AI / ML functions will also differ from existing CSI processing. We cannot directly use the CPU processing within the existing CSI framework; we need to design new methods to measure AI / ML function complexity or new CPU processing methods. Furthermore, the new hardware has limited storage capacity and may not be able to load all models corresponding to / associated with AI / ML functions simultaneously. Therefore, the storage capacity of the models corresponding to / associated with AI / ML functions needs to be considered within the CSI framework.

[0065] An AI / ML-based CSI report will be associated with at least one AI / ML function, and an AI / ML function will be associated with at least one AI / ML model. For example, for two AI / ML-based CSI reports r1 and r2, r1 is associated with AI / ML function f1, and f1 is associated with AI / ML model m1; r2 is associated with AI / ML function f2, and f2 is associated with AI / ML model m2.

[0066] The AI / ML functions f1 and f2 can be the same or different. For example, if r1 is used to report beam management-related results and r2 is used to report CSI compression-related results, then f1 and f2 will be different. If r1 is used to report CSI compression model inference results and r2 is used to report CSI compression model performance monitoring results, and r1 and r2 are related in the CSI reporting settings, then f1 and f2 will be the same.

[0067] If f1 and f2 are the same, m1 and m2 can be the same or different, depending on factors such as model deployment and network configuration. If f1 and f2 are not different, m1 and m2 can be the same or different, again depending on factors such as model deployment and network configuration.

[0068] Multiple AI / ML-based CSI reports may have the same AI / ML functions, or they may be all different, or they may have some of the same functions.

[0069] For a set of (R) AI / ML-based CSI reports Their corresponding AI / ML functions constitute a multiset. in Any two AI / ML functions f i and f j(i,j = 0,1,…,R-1, and i≠j) can be the same or different; AI / ML functions in E, among which Any two AI / ML models m i and m j (i,j = 0, 1, ..., R-1, and i ≠ j) can be the same or different. Based on whether the AI / ML models corresponding to the AI / ML functions are the same, we can... Divide into multiple submultisets: Each submultiset all elements They all correspond to the same AI / ML model, that is

[0070] The CSI framework has been illustrated above, but this application is not limited thereto. The above content can be considered part of the embodiments of this application and can be combined with one or more of the following embodiments. In the following description, "CSI processing unit for AI / ML," "CPU for / based on AI / ML," "AI / ML based CSI Processing Unit for AI / ML," and "AI / ML based CPU" can all be abbreviated as "APU," and these expressions are equivalent and can be used interchangeably.

[0071] In the embodiments of this application, one or more AI / ML models can be configured and run in network devices and / or terminal devices. AI / ML models can be used for various signal processing functions in wireless communication, such as CSI prediction, CSI compression, beam prediction, positioning management, etc.; this application is not limited thereto.

[0072] First aspect of the embodiments

[0073] This application provides a CSI processing method, which is described from the perspective of the terminal device.

[0074] Figure 2 is a schematic diagram of a CSI processing method according to an embodiment of this application. As shown in Figure 2, the method includes:

[0075] 201. The terminal device sends first information to the network device, the first information being used at least to report the processing complexity and / or storage capacity related to the AI / ML function / model;

[0076] 202, the terminal device receives second information from the network device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and

[0077] 203, the terminal device at least calculates / determines a processing complexity metric and / or a storage capacity metric related to the channel state information (CSI) report based on the first information.

[0078] It is worth noting that Figure 2 above is only an illustrative description of the embodiments of this application, but this application is not limited thereto. For example, the execution order between various operations can be appropriately adjusted, and other operations can be added or some operations can be removed. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the description in Figure 2 above.

[0079] In some embodiments, a functionality refers to an AI / ML feature / feature group enabled by a configuration, wherein the configuration is supported based on conditions indicated by UE capabilities.

[0080] For example, an AL / ML function can be one or more functions, or one or more logical models, or one or more sub-functions, or one or more features, or one or more feature groups.

[0081] For example, the function could be to use AI / ML for spatial beam prediction, or to use AI / ML for temporal beam prediction, or to use AI / ML for CSI prediction, or to use AI / ML for direct positioning, or to use AI / ML for assisted positioning, or to use AI / ML for CSI compression, and so on.

[0082] In some embodiments, the AI / ML function / model can be used for CSI prediction. One or more reference signals are used for measurement, and the measurement results are input into the AI / ML function / model. One or more CSIs are used in the output of the AI / ML function / model for inference.

[0083] For ease of description, CSI prediction based on AI / ML functionality / model will be referred to as model inference or inference operation, training data collection based on AI / ML functionality / model will be referred to as training data collection (training data collection can also use non-AI / ML methods), and performance monitoring based on AI / ML functionality / model will be referred to as performance monitoring.

[0084] In some embodiments, the network device may transmit configuration information for one or more reference signals, such as CSI-RS configuration information, etc. This application is not limited thereto; further details regarding specific configuration information can be found in related technologies. The configuration information may include configuration information for training data collection, and / or configuration information for model inference, and / or configuration information for performance monitoring.

[0085] Figure 3 is a schematic diagram of AI / ML for CSI prediction according to an embodiment of this application. As shown in Figure 3, one or more reference signals can be received and measured by a terminal device. The measurement results can be used as input to AI / ML, and CSI or beamforming can be used by the terminal device as output to AI / ML, for example, the measurement results can be used as labeled data or ground truth data for AI / ML. For details regarding AI / ML, please refer to related technologies, which will not be elaborated here.

[0086] Figure 4 is a schematic diagram of CSI reporting according to an embodiment of this application, illustratively illustrated using a terminal device configured with AI / ML. As shown in Figure 4, the method includes:

[0087] 401. The terminal device receives CSI reporting configuration information associated with AI / ML from the network device; for example, the configuration information includes a set of reference signal resources for measurement.

[0088] 402. The network device sends a reference signal (e.g., CSI-RS) to the terminal device. The UE receives the CSI-RS signal and performs channel measurement.

[0089] 403. The terminal device performs inference and / or monitoring based on AI / ML; that is, the terminal device inputs the measurement results into the AI / ML function / model; for example, the measurement results of CSI-RS are used as the input of AI / ML, and CSI or beamforming is used for prediction (or inference).

[0090] 404. After processing the inference / monitoring results, the terminal device reports them to the network device via a CSI report.

[0091] It is worth noting that Figure 4 above is only an illustrative description of the embodiments of this application, but this application is not limited thereto. For example, the execution order between various operations can be appropriately adjusted, and other operations can be added or some operations can be removed. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the description in Figure 4 above.

[0092] The above illustration demonstrates CSI reporting based on AI / ML, but this application is not limited to this. This application can be applied to CSI prediction, as well as CSI compression, beam management, positioning, and so on.

[0093] In some embodiments, the first information includes at least one of the following related to the AI / ML functionality / model: the maximum AI / ML processing complexity metric N supported by the terminal device. CPU,AI The maximum AI / ML storage capacity metric supported by the terminal device. AI,total List of AI / ML supported functionalities for terminal devices.

[0094] In some embodiments, the maximum AI / ML processing complexity is measured as floating-point operations per second (FLOPS), or as a quantized / normalized FLOPS, or as the number of AI / ML-based CPUs (CSI Processing Units); the maximum AI / ML storage capacity is measured as a floating-point number, or as a quantized / normalized value.

[0095] For example, the maximum AI / ML processing complexity can be measured in terms of floating-point operations per second (FLOPS); FLOPS can be in floating-point form or in quantized form.

[0096] For example, the maximum AI / ML processing complexity can also be measured by the number of AI / ML-based CSI Processing Units (AI / ML based CPU, also referred to as APU in this paper), which can be denoted as N. CPU,AI .

[0097] For example, the maximum AI / ML storage capacity metric supported in the first piece of information can be expressed as a floating-point number, or it can be expressed as a quantized number, or it can be expressed as a normalized number, which can be denoted as S. AI,total .

[0098] In some embodiments, the AI / ML function list includes at least one AI / ML function. The AI / ML function includes the following information:

[0099] The use case information corresponding to the AI / ML function, and / or

[0100] The processing complexity metric corresponding to the AI / ML function, and / or

[0101] The storage capacity metric corresponding to the AI / ML function, and / or

[0102] The inference latency of the AI / ML function, and / or

[0103] The model loading time of the AI / ML function, and / or

[0104] The network configuration information corresponding to the AI / ML function.

[0105] The above illustrative examples illustrate the information included in AI / ML functions. This application is not limited to these examples; for instance, it may include other information, one of these information examples, or any combination of these information examples.

[0106] In some examples, the use case information includes at least one of the following: beam management use case, positioning use case, CSI prediction use case, and CSI compression use case.

[0107] For example, each AI / ML function in the list of supported functionalities in the first information can include use case information, such as beam management, positioning, CSI prediction, CSI compression, or other use cases.

[0108] In some examples, the processing complexity metric is floating-point operations per second (FLOPS), or quantized / normalized floating-point operations per second (FLOPS), or the number of AI / ML-based CPUs (CSI Processing Units); the storage capacity metric is a floating-point number, or a quantized / normalized value.

[0109] For example, for each AI / ML function i in the list of supported functionalities in the first information, a corresponding AI / ML complexity metric can be included. This complexity metric can have the same form as the maximum AI / ML processing complexity metric; both can be in the form of FLOPS, or both can be in the form of the number of CPUs based on AI / ML, and can be denoted as...

[0110] For example, each AI / ML function i in the list of supported functionalities in the first information can include a corresponding AI / ML storage capacity metric. This storage capacity metric can have the same form as the maximum storage capacity metric, either a floating-point number or a quantized form, and can be denoted as...

[0111] In some examples, the network configuration information includes at least one of the following: the number of antenna ports or CSI-RS ports, the number of subbands, and feedback overhead. This application is not limited to these; for example, it may also include other configuration information.

[0112] For example, each AI / ML function i in the list of supported functionalities in the first information may include network configuration information. This network configuration information includes, but is not limited to: the number of antenna ports or CSI-RS ports, the number of subbands, feedback overhead, etc., denoted as follows: Different use cases may require different configuration information; for example, this information can affect the correspondence between AL / ML functions and models.

[0113] Figure 5 is an example diagram of the first information according to an embodiment of this application, illustrating an example of the first information. As shown in Figure 5, in the first information, for example, the maximum AI / ML processing complexity metric is set in the form of the number of CPUs used for AI / ML, and takes the value N. CPU,AI =20; The maximum AI / ML storage capacity metric is in a normalized form, taking a value of 30 storage capacity units, i.e., S. AI,total =30.

[0114] As shown in Figure 5, in the first information, for example, there are 64 AI / ML functions in the AI / ML function list.

[0115] For AI / ML function 0:

[0116] - Used for CSI prediction use cases;

[0117] -AI / ML processing complexity is

[0118] -AI / ML storage capacity is 3 capacity units, that is

[0119] - Model inference latency is 1.5 time slots, and model loading time is 3 OFDM symbols;

[0120] - Other network configuration information corresponding to this AI / ML function includes: number of ports. Number of sub-bands

[0121] For AI / ML function 1:

[0122] - For CSI compression use cases;

[0123] -AI / ML processing complexity is

[0124] -AI / ML storage capacity is in 7 units, i.e.

[0125] - Model inference latency is 1.5 time slots, and model loading time is 5 OFDM symbols;

[0126] - Other network configuration information corresponding to this AI / ML function includes: number of ports. Number of sub-bands Feedback cost is

[0127] For AI / ML feature 2:

[0128] - For CSI compression use cases;

[0129] -AI / ML processing complexity is

[0130] -AI / ML storage capacity is 8 capacity units, that is

[0131] - Model inference latency is 2 time slots, and model loading time is 7 OFDM symbols;

[0132] - Other network configuration information corresponding to this AI / ML function includes: number of ports. Number of sub-bands Feedback cost is

[0133] Other AI / ML functions in the list will not be elaborated upon further. The parameter values ​​in the above examples are for illustrative purposes only; other values ​​may be used in actual systems.

[0134] The above provides an illustrative explanation of the first information, etc. The following will explain the CSI report update, etc.

[0135] In some embodiments, the terminal device determines the updated information of the Channel State Information (CSI) report based at least on a calculated / determined processing complexity metric and / or storage capacity metric.

[0136] In some embodiments, the terminal device determines the update information of the Channel State Information (CSI) report based at least on a calculated / determined processing complexity metric and / or storage capacity metric, as well as third information.

[0137] For example, the third information may include at least one of the following: CSI report priority associated with AI / ML, number of CSI-RS resources, and number of ports. This application is not limited to this, and may include other information as well.

[0138] In some embodiments, the UE calculates the complexity metric and storage capacity metric of all currently activated and to-be-activated AI / ML functions based on the parameters in the first information, requiring that the sum of the complexity metrics of the activated and to-be-activated AI / ML functions does not exceed N. CPU,AI Furthermore, the sum of the storage capacity metrics for both activated and to-be-activated AI / ML functions must not exceed S. AI,total If the AI / ML complexity metric exceeds N... CPU,AI Or storage capacity metric exceeding S AI,total If so, no new AI / ML function will be activated before the currently activated AI / ML function is deactivated.

[0139] The following further explains the metrics for processing complexity and / or storage capacity. Different AI / ML functions can correspond to / be associated with different models, or they can correspond to / be associated with the same model.

[0140] In some embodiments, the complexity metric in the AI / ML feature list included in the first information can be used directly. Calculate the sum of complexity metrics for activated and to-be-activated AI / ML features, and directly use the storage capacity metrics from the AI / ML feature list contained in the first information. Calculate the sum of storage capacity metrics for activated and to-be-activated AI / ML functions. It can be set by default that AI / ML functions with the same use case correspond to / are associated with the same model, or specific parameters can be set to explicitly indicate which AI / ML functions correspond to / are associated with the same model.

[0141] In some embodiments, the terminal device calculates / determines processing complexity metrics and / or storage capacity metrics related to the Channel State Information (CSI) report, including:

[0142] When L is present in a certain OFDM symbol period AI If an APU is used for CSI calculations / processing related to CSI reports, then it is said to have L AI If one APU is occupied, then N APUs are available within that OFDM symbol period. CPU,AI -L AI One APU is unoccupied. When L... AI,S If a storage capacity unit is used for AI / ML related storage, then within this OFDM symbol period, there are still S... AI,total -L AI,S N storage capacity units are unused. CPU,AI -LAI One APU is not occupied, S AI,total -L AI,S Within an OFDM symbol period where no storage capacity unit is used, there are N AI Each AI / ML-based CSI report begins to occupy its corresponding APU and its corresponding storage capacity unit (where each CSI report has n = 0, ..., N). AI -1, corresponding to the occupation One APU, and occupy One storage capacity unit, of which and The value of N is determined by the parameters in the first information, then the UE does not need to update N. AI -M AI The lowest priority CSI report will be selected; specific priority definitions are not required here. Among them, M... AI (satisfies 0≤M) AI ≤N AI To satisfy the inequality and The largest integer value.

[0143] Table 2 is an example of an embodiment of this application:

[0144] [Revised according to Rule 26, May 2025] Table 2

[0145] Table 2 illustrates an example of an embodiment of this application, but is not limited thereto. For example, the above high-level parameters can be adjusted according to actual needs and are not limited to the description in Table 2; furthermore, terms can also be appropriately replaced, for example, "storage" and "memory" can be interchanged, and "APU" can also be replaced with other terms.

[0146] In some embodiments, the terminal device calculates / determines processing complexity metrics and / or storage capacity metrics related to the Channel State Information (CSI) report, including:

[0147] The terminal device calculates the sum of the processing complexity metric of the first AI / ML function to be activated and the processing complexity metric of one or more second AI / ML functions that have been activated, and calculates the sum of the storage capacity metric of the first AI / ML function and the storage capacity metric of the one or more second AI / ML functions.

[0148] If the sum of the processing complexity metrics does not exceed the maximum AI / ML processing complexity metric, and the sum of the storage capacity metrics does not exceed the maximum AI / ML storage capacity metric, the terminal device determines to activate the first AI / ML function / model and updates the occupied processing complexity metric and storage capacity metric.

[0149] In one example, suppose the maximum AI / ML processing complexity metric in the first message received by the UE is N. CPU,AI =20, maximum AI / ML storage capacity metric is S AI,total =30. The complexity metric used by the AI / ML functions currently activated in the UE is N. CPU,AI,used =15, the measured AI / ML storage capacity already occupied (S) AI,used = 25 storage capacity units.

[0150] For example, the CSI report to be activated or triggered corresponds to / is associated with AI / ML function j1. In the first information, the O configured for AI / ML function j1... CPU,AI =3, S AI =2. Because N CPU,AI,used +O CPU,AI =15+3=18<20, and S AI,used +S AI =25+2=27<30, the model corresponding to / associated with AI / ML function j1 ​​can be loaded. Based on the above conditions, and if other conditions are met at the same time (if there are other conditions), the corresponding CSI report can be updated.

[0151] For example, the CSI report to be activated or triggered corresponds to / is associated with AI / ML function j2. In the first information, the O configured for AI / ML function j2... CPU,AI =6, S AI =2. Because N CPU,AI,used +O CPU,AI =15+6=21>20, then the model corresponding to / associated with AI / ML function j2 cannot be loaded.

[0152] For example, the CSI report to be activated or triggered corresponds to / is associated with AI / ML function j3. In the first information, the O configured for AI / ML function j3... CPU,AI =3, S AI =6. Because S AI,used +S AI =25+6=31>30, then the model corresponding to / associated with AI / ML function j3 cannot be loaded.

[0153] For example, the CSI report to be activated or triggered corresponds to / is associated with AI / ML function j4. In the first information, the O configured for AI / ML function j4... CPU,AI =6, S AI =6. Because N CPU,AI,used +O CPU,AI =15+6=21>20, and S AI,used +S AI =25+6=31>30, then the model corresponding to / associated with AI / ML function j4 cannot be loaded.

[0154] In some embodiments, different AI / ML functions may correspond to / be associated with different models. In other embodiments, different AI / ML functions may correspond to / be associated with the same model. For example, AI / ML functions with the same use case may be associated with / be associated with the same model by default, or specific parameters may be set to explicitly indicate which AI / ML functions correspond to / be associated with the same model.

[0155] For example, use the complexity metric from the list of AI / ML features included in the first piece of information. Calculate the sum of complexity metrics for both activated and to-be-activated AI / ML functions, using a second storage capacity metric. Calculate the sum of storage capacity metrics for AI / ML functions that are already activated and those that are to be activated.

[0156] The second storage capacity metric For based on The adjusted value, Specific values ​​are as follows:

[0157] - There are N within a certain OFDM symbol period AI There are N CSI reports related to AI / ML functions pending processing. AI A multiset consisting of AI / ML functions associated with a single CSI report. In the context, if there is a set of P (P≥1) AI / ML functions p0,…p P-1 The resulting sub-multiset C corresponds to / is associated with the same AI / ML model, and the storage capacity occupied by this model is... Then take the sum of the storage capacity occupied by these P AI / ML functions as S. P ,Right now Alternatively, it can be expressed as follows: for the first occurrence of element p0 in the submultiset C, take... In a multiset C, other elements are taken... (p k ∈C, and p k ≠p0);

[0158] -like If there are multiple submultisets, and the AI / ML functions in each submultiset correspond to / are associated with the same AI / ML model, then the above operation is performed on the storage capacity occupied by the AI / ML functions in each submultiset.

[0159] In some embodiments, the terminal device calculates / determines processing complexity metrics and / or storage capacity metrics related to the Channel State Information (CSI) report, including:

[0160] When L is present in a certain OFDM symbol period AI If an APU is used for CSI calculations / processing related to CSI reports, then it is said to have L AI If one APU is occupied, then N APUs are available within that OFDM symbol period. CPU,AI -L AI One APU is unoccupied. When L... AI,S If a storage capacity unit is used for AI / ML related storage, then within this OFDM symbol period, there are still S... AI,total -L AI,S N storage capacity units are not occupied. CPU,AI -L AI One APU is not occupied, S AI,total -L AI,S Within an OFDM symbol period where no storage capacity unit is used, there are N AI Each AI / ML-based CSI report begins to occupy its corresponding APU and its corresponding storage capacity unit (where each CSI report has n = 0, ..., N). AI -1, corresponding to the occupation One APU, and occupy One storage capacity unit, of which and The value of is related to the specific model deployed by the UE and the capabilities of the UE, and Determined by the parameters in the first information, Based on the first information (Adjusted value). About The specific value of NAI The CSI report corresponds to / is associated with a multiset of AI / ML functions. In the context, if there is a set of P (P≥1) AI / ML functions p0,…p P-1 The resulting submultiset C corresponds to / is associated with the same AI / ML model, and the storage capacity occupied by this model is... Then take the sum of the storage capacity occupied by these P AI / ML functions as S. P ,Right now Alternatively, it can be expressed as follows: for the first occurrence of element p0 in the submultiset C, take... For other elements in the submultiset C, take (p k ∈C, and p k ≠p0). If Given multiple submultisets, where each AI / ML function in a submultiset corresponds to / is associated with the same AI / ML model, determine the storage capacity required for each AI / ML function within the submultiset. Perform the above operations. Then the UE does not need to update N. AI -M AI The lowest priority CSI report; the specific priority definition is not required here, where M AI (satisfies 0≤M) AI ≤N AI To satisfy the inequality and The largest integer value.

[0161] In some embodiments, the terminal device calculates / determines processing complexity metrics and / or storage capacity metrics related to the Channel State Information (CSI) report, including:

[0162] The terminal device calculates the sum of the processing complexity metric of the first AI / ML function to be activated and the processing complexity metric of one or more second AI / ML functions / models that have been activated.

[0163] If the sum of the processing complexity metrics does not exceed the maximum AI / ML processing complexity metric, the terminal device determines to activate the first AI / ML function / model and updates the occupied processing complexity metric.

[0164] In some examples, it is assumed that the maximum AI / ML processing complexity metric in the first message received by the UE is N. CPU,AI =20, maximum AI / ML storage capacity metric is S AI,total =30. In the AI / ML function list in the first information, AI / ML function i1 and AI / ML function i2 correspond to / are associated with the same model, O CPU,AI,i1 =2,S AI,,i1 =3; AI / ML function i3 corresponds to / is associated with a different model than AI / ML function i1 or AI / ML function i2, O CPU,AI,i3 =2,S AI,,i3 =3; where AI / ML function i1 has been activated, while AI / ML functions i2 and i3 have not been activated. The complexity metric occupied by the currently activated AI / ML functions is N. CPU,AI,used =15, the measured AI / ML storage capacity already occupied (S) AI,used =25.

[0165] For example, the activated or triggered CSI report 1 corresponds to / is associated with AI / ML function i3, due to N CPU,AI,used +O CPU,AI,i1 =15+2=17<20, and S AI,used +S AI,,i1 =25+3=28<30, the model corresponding to / associated with AI / ML function i3 can be loaded. Based on the above conditions, and if other conditions are met at the same time (if there are other conditions), the corresponding CSI report can be updated.

[0166] After loading the model corresponding to / associated with AI / ML function i3, the complexity metric already occupied by the currently activated AI / ML function is updated to N. CPU,AI,used =15+2=17, the occupied AI / ML storage capacity metric is updated to S AI,used =25+2=27.

[0167] Based on this, if CSI reporting 2 is activated or triggered again, corresponding to / associated with AI / ML function i2, since AI / ML function i2 and AI / ML function i1 correspond to / associate with the same model, only the occupied AI / ML complexity metric needs to be updated, and the occupied AI / ML storage capacity metric does not need to be updated. That is, determine N. CPU,AI,used +O CPU,AI,i3 =17+2=19<20, and S AI,used +0 = 28 < 30, which means that the model corresponding to / associated with AI / ML function i2 can be loaded and the corresponding CSI reporting process can be executed.

[0168] In some implementation examples, AI / ML functions for the same use case are defaulted to / associated with the same model, or the first parameter indicates the AI / ML function that corresponds to / associates with the same model, or the second parameter indicates the AI / ML function that corresponds to / associates with a different model.

[0169] For example, AI / ML functions with the same use case can be associated with the same model by default, or specific parameters can be set to explicitly or implicitly indicate which AI / ML functions are associated with the same model, and / or indicate which AI / ML functions are associated with different models.

[0170] The above embodiments are merely illustrative examples of embodiments of this application, but this application is not limited thereto, and appropriate modifications can be made based on the above embodiments. For example, the above embodiments can be used alone, or one or more of the above embodiments can be combined.

[0171] As can be seen from the above embodiments, the processing complexity and / or storage capacity of the CSI report associated with AI / ML functions / models are calculated / measured. Thus, the processing complexity and / or storage capacity issues related to AI / ML functions can be handled within the CSI framework, thereby improving the performance and efficiency of AI / ML and optimizing the performance of the wireless communication system.

[0172] Second aspect of the embodiments

[0173] This application provides a CSI processing method, described from the perspective of a network device. The embodiments of the second aspect can be combined with the embodiments of the first aspect, and the content identical to that of the embodiments of the first aspect will not be repeated.

[0174] Figure 6 is a schematic diagram of a CSI processing method according to an embodiment of this application. As shown in Figure 6, the method includes:

[0175] 601, The network device receives first information sent by the terminal device, the first information being used at least to report the processing complexity and / or storage capacity related to the AI / ML function / model;

[0176] 602, the network device sends second information to the terminal device, the second information being used to configure and / or trigger a channel state information (CSI) report related to the AI / ML function / model;

[0177] The first information is used by the terminal device to calculate / determine processing complexity metrics and / or storage capacity metrics related to the Channel State Information (CSI) report.

[0178] It is worth noting that Figure 6 above is only an illustrative description of the embodiments of this application, but this application is not limited thereto. For example, the execution order between various operations can be appropriately adjusted, and other operations can be added or some operations can be removed. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the description in Figure 6 above.

[0179] In some embodiments, the first information includes at least one of the following related to the AI / ML functionality / model: the maximum AI / ML processing complexity metric N supported by the terminal device. CPU,AI The maximum AI / ML storage capacity metric supported by the terminal device. AI,total List of AI / ML supported functionalities for terminal devices.

[0180] In some embodiments, the AI / ML function list includes at least one AI / ML function.

[0181] The AI / ML function includes the following information:

[0182] The use case information corresponding to the AI / ML function, and / or

[0183] The processing complexity metric corresponding to the AI / ML function, and / or

[0184] The storage capacity metric corresponding to the AI / ML function, and / or

[0185] The inference latency of the AI / ML function, and / or

[0186] The model loading time of the AI / ML function, and / or

[0187] The network configuration information corresponding to the AI / ML function.

[0188] In some embodiments, when L is present within a certain OFDM symbol period AI If an APU is used for CSI calculations / processing related to CSI reports, then it is said to have L AI If one APU is occupied, then N APUs are available within that OFDM symbol period. CPU,AI -L AI One APU is unoccupied. When L... AI,S If a storage capacity unit is used for AI / ML related storage, then within this OFDM symbol period, there are still S...AI,total -L AI,S N storage capacity units are unused. CPU,AI -L AI One APU is not occupied, S AI,total -L AI,S Within an OFDM symbol period where no storage capacity unit is used, there are N AI Each AI / ML-based CSI report begins to occupy its corresponding APU and its corresponding storage capacity unit (where each CSI report has n = 0, ..., N). AI -1, corresponding to the occupation One APU, and occupy One storage capacity unit, of which and The value of N is determined by the parameters in the first information, then the UE does not need to update N. AI -M AI The lowest priority CSI report will be selected; specific priority definitions are not required here. Among them, M... AI (satisfies 0≤M) AI ≤N AI To satisfy the inequality and The largest integer value.

[0189] In some embodiments, a second storage capacity metric is used. Calculate the sum of storage capacity metrics for activated and to-be-activated AI / ML features. For based on The adjusted value, regarding The specific values,

[0190] N has a certain OFDM symbol period AI There are N CSI reports related to AI / ML functions pending processing. AI A multiset consisting of AI / ML functions corresponding to / associated with each CSI report. In the context, if there is a set of P (P≥1) AI / ML functions p0,…p P-1 The resulting sub-multiset C corresponds to / is associated with the same AI / ML model, and the storage capacity occupied by this model is... Then take the sum of the storage capacity occupied by these P AI / ML functions as S. P ,Right now Alternatively, it can be expressed as follows: for the first occurrence of element p0 in the submultiset C, take... Take other elements in set C (p k ∈C, and p k ≠p0). If If there are multiple submultisets, and the AI / ML functions in each submultiset correspond to / are associated with the same AI / ML model, then the above operation is performed on the storage capacity occupied by the AI / ML functions in each submultiset.

[0191] In some embodiments, the network device may send configuration information, etc., to the terminal device. The network device may receive feedback information and / or report information sent by the terminal device. For example, the terminal device may report inference results and / or performance monitoring results and / or training data collection results to the network device, but this application is not limited thereto.

[0192] The above embodiments are merely illustrative examples of embodiments of this application, but this application is not limited thereto, and appropriate modifications can be made based on the above embodiments. For example, the above embodiments can be used alone, or one or more of the above embodiments can be combined.

[0193] As can be seen from the above embodiments, the processing complexity and / or storage capacity of the CSI report associated with AI / ML functions / models are calculated / measured. Thus, the processing complexity and / or storage capacity issues related to AI / ML functions can be handled within the CSI framework, thereby improving the performance and efficiency of AI / ML and optimizing the performance of the wireless communication system.

[0194] Third aspect of the embodiments

[0195] This application provides a CSI processing apparatus. This apparatus may be, for example, a terminal device, or one or more components or parts configured within a terminal device; details identical to those in the first and second aspects will not be repeated.

[0196] Figure 7 is a schematic diagram of a CSI processing device according to an embodiment of the present application. As shown in Figure 7, the CSI processing device 700 according to an embodiment of the present application includes: a transmitter 701, a receiver 702 and a processor 703.

[0197] The transmitter 701 sends first information to the network device, the first information being used at least to report processing complexity and / or storage capacity related to the AI / ML function / model; the receiver 702 receives second information from the network device, the second information being used to configure and / or trigger a channel state information (CSI) report related to the AI / ML function / model; and the processor 703 calculates / determines at least the processing complexity metric and / or storage capacity metric related to the channel state information (CSI) report based on the first information.

[0198] In some embodiments, the processor 703 determines the updated information of the channel state information (CSI) report based at least on a computed / determined processing complexity metric and / or storage capacity metric.

[0199] In some embodiments, the first information includes at least one of the following related to the AI / ML functionality / model: the maximum AI / ML processing complexity metric N supported by the terminal device. CPU,AI The maximum AI / ML storage capacity metric supported by the terminal device. AI,total List of AI / ML supported functionalities for terminal devices.

[0200] In some embodiments, the maximum AI / ML processing complexity is measured in floating-point operations per second (FLOPS), or in quantized / normalized floating-point operations per second (FLOPS), or in the number of AI / ML-based CPUs (CSI Processing Units).

[0201] The maximum AI / ML storage capacity metric is a floating-point number, or a quantized / normalized value.

[0202] In some embodiments, the AI / ML function list includes at least one AI / ML function.

[0203] In some embodiments, an AI / ML function includes the following information:

[0204] The use case information corresponding to the AI / ML function, and / or

[0205] The processing complexity metric corresponding to the AI / ML function, and / or

[0206] The storage capacity metric corresponding to the AI / ML function, and / or

[0207] The inference latency of the AI / ML function, and / or

[0208] The model loading time of the AI / ML function, and / or

[0209] The network configuration information corresponding to the AI / ML function.

[0210] In some embodiments, the use case information includes at least one of the following: beam management use case, positioning use case, CSI prediction use case, and CSI compression use case.

[0211] In some embodiments, the processing complexity metric is floating-point operations per second (FLOPS), or floating-point operations per second after quantization / normalization (FLOPS), or the number of AI / ML-based CPUs (CSI Processing Units).

[0212] The storage capacity metric is a floating-point number, or a quantized / normalized value.

[0213] In some embodiments, the network configuration information includes at least one of the following: the number of antenna ports or CSI-RS ports, the number of subbands, and feedback overhead.

[0214] In some embodiments, the processor 703 determines the update information of the channel state information (CSI) report based at least on a computed / determined processing complexity metric and / or storage capacity metric, as well as third information.

[0215] In some embodiments, the third information includes at least one of the following: CSI report priority associated with AI / ML, number of CSI-RS resources, and number of ports.

[0216] In some embodiments, the processor 703 calculates / determines processing complexity metrics and / or storage capacity metrics associated with the channel state information (CSI) report, including:

[0217] The processor 703 calculates the sum of the processing complexity metric of the first AI / ML function to be activated and the processing complexity metric of one or more second AI / ML functions that have been activated, and calculates the sum of the storage capacity metric of the first AI / ML function and the storage capacity metric of the one or more second AI / ML functions.

[0218] In some embodiments, if the sum of the processing complexity metrics does not exceed the maximum AI / ML processing complexity metric and the sum of the storage capacity metrics does not exceed the maximum AI / ML storage capacity metric, the processor 703 determines to activate the first AI / ML function / model and updates the occupied processing complexity metric and storage capacity metric.

[0219] In some embodiments, the processor 703 calculates / determines processing complexity metrics and / or storage capacity metrics associated with the channel state information (CSI) report, including:

[0220] The processor 703 calculates the sum of the processing complexity metric of the first AI / ML function to be activated and the processing complexity metric of one or more second AI / ML functions that have been activated.

[0221] In some embodiments, if the sum of the processing complexity metrics does not exceed the maximum AI / ML processing complexity metric, the processor 703 determines to activate the first AI / ML function / model and updates the occupied processing complexity metric.

[0222] In some embodiments, the processor 703 calculates / determines processing complexity metrics and / or storage capacity metrics associated with the channel state information (CSI) report, including:

[0223] When L is present in a certain OFDM symbol period AI If an APU is used for CSI calculations / processing related to CSI reports, then it is said to have L AI If one APU is occupied, then N APUs are available within that OFDM symbol period. CPU,AI -L AI One APU is unoccupied. When L... AI,S If a storage capacity unit is used for AI / ML related storage, then within this OFDM symbol period, there are still S... AI,total -L AI,S N storage capacity units are unused. CPU,AI -L AI One APU is not occupied, S AI,total -L AI,S Within an OFDM symbol period where no storage capacity unit is used, there are N AIEach AI / ML-based CSI report begins to occupy its corresponding APU and its corresponding storage capacity unit (where each CSI report has n = 0, ..., N). AI -1, corresponding to the occupation One APU, and occupy One storage capacity unit, of which and The value of N is determined by the parameters in the first information, then the UE does not need to update N. AI -M AI The lowest priority CSI report will be selected; specific priority definitions are not required here. Among them, M... AI (satisfies 0≤M) AI ≤N AI To satisfy the inequality and The largest integer value.

[0224] In some embodiments, a second storage capacity metric is used. Calculate the sum of storage capacity metrics for activated and to-be-activated AI / ML features. For based on The adjusted value, regarding The specific values,

[0225] N has a certain OFDM symbol period AI There are N CSI reports related to AI / ML functions pending processing. AI A multiset consisting of AI / ML functions corresponding to / associated with each CSI report. In the context, if there is a set of P (P≥1) AI / ML functions p0,…p P-1 The resulting sub-multiset C corresponds to / is associated with the same AI / ML model, and the storage capacity occupied by this model is... Then take the sum of the storage capacity occupied by these P AI / ML functions as S. P ,Right now Alternatively, it can be expressed as follows: for the first occurrence of element p0 in the submultiset C, take... Take other elements in set C (p k ∈C, and p k ≠p0). If If there are multiple submultisets, and the AI / ML functions in each submultiset correspond to / are associated with the same AI / ML model, then the above operation is performed on the storage capacity occupied by the AI / ML functions in each submultiset.

[0226] The above embodiments are merely illustrative examples of embodiments of this application, but this application is not limited thereto, and appropriate modifications can be made based on the above embodiments. For example, the above embodiments can be used alone, or one or more of the above embodiments can be combined.

[0227] It is worth noting that the above description only covers the components or modules relevant to this application, but this application is not limited thereto. The CSI processing apparatus may also include other components or modules, and for details regarding these components or modules, please refer to related technologies.

[0228] Furthermore, for simplicity, Figure 7 only illustrates the connection relationships or signal flow between the various components or modules, but those skilled in the art should understand that various related technologies such as bus connections can be used. The aforementioned components or modules can be implemented using hardware facilities such as processors, memory, transmitters, and receivers; this application does not limit this implementation.

[0229] As can be seen from the above embodiments, the processing complexity and / or storage capacity of the CSI report associated with AI / ML functions / models are calculated / measured. Thus, the processing complexity and / or storage capacity issues related to AI / ML functions can be handled within the CSI framework, thereby improving the performance and efficiency of AI / ML and optimizing the performance of the wireless communication system.

[0230] Fourth aspect of the embodiment

[0231] This application provides a CSI processing apparatus. This apparatus may be, for example, a network device, or one or more components or parts configured within a network device; details identical to those in the embodiments of the first to third aspects will not be repeated.

[0232] Figure 8 is another schematic diagram of a CSI processing device according to an embodiment of this application. As shown in Figure 8, the CSI processing device 800 includes a receiver 801 and a transmitter 802, and may also include a processor 803.

[0233] Receiver 801 receives first information sent by terminal device, the first information being used at least to report processing complexity and / or storage capacity related to AI / ML functionality / model; transmitter 802 sends second information to terminal device, the second information being used to configure and / or trigger channel state information (CSI) reporting related to the AI / ML functionality / model; wherein, the first information is used by terminal device to calculate / determine processing complexity and / or storage capacity metrics related to the channel state information (CSI) report.

[0234] In some embodiments, the first information includes at least one of the following related to the AI / ML functionality / model: the maximum AI / ML processing complexity metric N supported by the terminal device. CPU,AI The maximum AI / ML storage capacity metric supported by the terminal device. AI,total List of AI / ML supported functionalities for terminal devices.

[0235] The above embodiments are merely illustrative examples of embodiments of this application, but this application is not limited thereto, and appropriate modifications can be made based on the above embodiments. For example, the above embodiments can be used alone, or one or more of the above embodiments can be combined.

[0236] It is worth noting that the above description only covers the components or modules relevant to this application, but this application is not limited thereto. The CSI processing apparatus may also include other components or modules, and for details regarding these components or modules, please refer to related technologies.

[0237] Furthermore, for simplicity, Figure 8 only illustrates the connection relationships or signal flow between the various components or modules, but those skilled in the art should understand that various related technologies such as bus connections can be used. The aforementioned components or modules can be implemented using hardware facilities such as processors, memory, transmitters, and receivers; this application does not limit this implementation.

[0238] As can be seen from the above embodiments, the processing complexity and / or storage capacity of the CSI report associated with AI / ML functions / models are calculated / measured. Thus, the processing complexity and / or storage capacity issues related to AI / ML functions can be handled within the CSI framework, thereby improving the performance and efficiency of AI / ML and optimizing the performance of the wireless communication system.

[0239] Fifth aspect of the embodiment

[0240] This application also provides a communication system, which can be referred to FIG1. ​​The contents that are the same as those in the embodiments of the first to fourth aspects will not be repeated.

[0241] In some embodiments, the communication system 100 may include at least:

[0242] A network device receives first information, which is at least used to report processing complexity and / or storage capacity related to an AI / ML function / model; sends second information, which is used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and

[0243] The terminal device calculates / determines at least based on the first information a processing complexity metric and / or storage capacity metric related to the Channel State Information (CSI) report.

[0244] This application also provides a terminal device, but the application is not limited thereto and may also include other devices.

[0245] Figure 9 is a schematic diagram of a terminal device according to an embodiment of this application. As shown in Figure 9, the terminal device 900 may include a processor 910 and a memory 920; the memory 920 stores data and programs and is coupled to the processor 910. It is worth noting that this figure is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunications functions or other functions.

[0246] For example, processor 910 may be configured to execute a program to implement the CSI processing method as described in the embodiments of the first aspect. For example, processor 910 may be configured to perform the following control: send first information to a network device, the first information being used at least to report processing complexity and / or storage capacity related to the AI / ML function / model; receive second information from the network device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and calculate / determine at least based on the first information a processing complexity metric and / or storage capacity metric related to the Channel State Information (CSI) report.

[0247] As shown in Figure 9, the terminal device 900 may further include: a communication module 930, an input unit 940, a display 950, and a power supply 960. The functions of these components are similar to those in the prior art and will not be described in detail here. It is worth noting that the terminal device 900 does not necessarily include all the components shown in Figure 9; these components are not essential. Furthermore, the terminal device 900 may also include components not shown in Figure 9, which can be referred to in the prior art.

[0248] This application also provides a network device, such as a base station, but this application is not limited to this and may also include other network devices.

[0249] Figure 10 is a schematic diagram of the configuration of a network device according to an embodiment of this application. As shown in Figure 10, the network device 1000 may include: a processor 1010 (e.g., a central processing unit CPU) and a memory 1020; the memory 1020 is coupled to the processor 1010. The memory 1020 can store various types of data; in addition, it also stores an information processing program 1030, and executes the program 1030 under the control of the processor 1010.

[0250] For example, processor 1010 may be configured to execute a program to implement the CSI processing method as described in the embodiments of the second aspect. For example, processor 1010 may be configured to perform the following control: receiving first information sent by a terminal device, the first information being used at least to report processing complexity and / or storage capacity related to the AI / ML function / model; sending second information to the terminal device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; wherein the first information is used to calculate / determine a processing complexity metric and / or storage capacity metric related to the Channel State Information (CSI) report.

[0251] In addition, as shown in Figure 10, the network device 1000 may also include a transceiver 1040 and an antenna 1050, etc.; the functions of the above components are similar to those in the prior art, and will not be described in detail here. It is worth noting that the network device 1000 does not necessarily have to include all the components shown in Figure 10; furthermore, the network device 1000 may also include components not shown in Figure 10, which can be referred to in the prior art.

[0252] This application also provides a computer program, wherein when the program is executed in a terminal device, the program causes the terminal device to perform the CSI processing method described in the first aspect of the embodiment.

[0253] This application also provides a storage medium storing a computer program, wherein the computer program causes a terminal device to execute the CSI processing method described in the first aspect of the embodiment.

[0254] This application also provides a computer program, wherein when the program is executed in a network device, the program causes the network device to perform the CSI processing method described in the second aspect of the embodiment.

[0255] This application also provides a storage medium storing a computer program, wherein the computer program causes a network device to perform the CSI processing method described in the second aspect of the embodiment.

[0256] The apparatus and methods described above in this application can be implemented in hardware or in combination with software. This application relates to a computer-readable program that, when executed by a logic component, enables the logic component to implement the apparatus or components described above, or to implement the various methods or steps described above. This application also relates to storage media for storing the above programs, such as hard disks, magnetic disks, optical disks, DVDs, flash memory, etc.

[0257] The methods / apparatus described in conjunction with the embodiments of this application can be directly embodied in hardware, software modules executed by a processor, or a combination of both. For example, one or more and / or combinations of one or more functional block diagrams shown in the figures can correspond to various software modules in a computer program flow, or to various hardware modules. These software modules can correspond to the various steps shown in the figures, respectively. These hardware modules can be implemented, for example, using a field-programmable gate array (FPGA) to embed these software modules.

[0258] The software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. A storage medium can be coupled to the processor, enabling the processor to read information from and write information to the storage medium; or the storage medium can be an integral part of the processor. The processor and storage medium can reside in an ASIC. The software module can be stored in the memory of a mobile terminal or in a memory card that can be inserted into the mobile terminal. For example, if the device (such as a mobile terminal) uses a high-capacity MEGA-SIM card or a high-capacity flash memory device, the software module can be stored in the MEGA-SIM card or the high-capacity flash memory device.

[0259] One or more and / or one or more combinations of functional blocks described in the accompanying drawings can be implemented as a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, or any suitable combination thereof for performing the functions described herein. One or more and / or one or more combinations of functional blocks described in the accompanying drawings can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in communication with a DSP, or any other such configuration.

[0260] The present application has been described above with reference to specific embodiments. However, those skilled in the art should understand that these descriptions are exemplary and not intended to limit the scope of protection of the present application. Those skilled in the art can make various modifications and variations to the present application based on its spirit and principles, and these modifications and variations are also within the scope of the present application.

[0261] Regarding the implementation methods including the above embodiments, the following notes are also disclosed:

[0262] 1. A method for processing Channel State Information (CSI), comprising:

[0263] The terminal device sends first information to the network device, the first information being used at least to report the processing complexity and / or storage capacity related to the AI / ML function / model;

[0264] The terminal device receives second information from the network device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; and

[0265] The terminal device at least calculates / determines a processing complexity metric and / or a storage capacity metric related to the Channel State Information (CSI) report based on the first information.

[0266] 2. A method for processing Channel State Information (CSI), comprising:

[0267] The network device receives first information sent by the terminal device, the first information being used at least to report the processing complexity and / or storage capacity related to the AI / ML function / model;

[0268] The network device sends second information to the terminal device, the second information being used to configure and / or trigger a Channel State Information (CSI) report related to the AI / ML function / model; wherein, the first information is used to calculate / determine a processing complexity metric and / or storage capacity metric related to the Channel State Information (CSI) report.

[0269] 3. According to the method described in Appendix 2, the first information includes at least one of the following related to the AI / ML functionality / model: the maximum AI / ML processing complexity metric N supported by the terminal device. CPU,AI The maximum AI / ML storage capacity metric supported by the terminal device. AI,total List of AI / ML supported functionalities for terminal devices.

[0270] 4. The method according to Appendix 3, wherein the AI / ML function list includes at least one AI / ML function;

[0271] The AI / ML function includes the following information:

[0272] The use case information corresponding to the AI / ML function, and / or

[0273] The processing complexity metric corresponding to the AI / ML function, and / or

[0274] The storage capacity metric corresponding to the AI / ML function, and / or

[0275] The inference latency of the AI / ML function, and / or

[0276] The model loading time of the AI / ML function, and / or

[0277] The network configuration information corresponding to the AI / ML function.

[0278] 5. According to the method described in Appendix 2, wherein when L is present in a certain OFDM symbol period AI If an APU is used for CSI calculations / processing related to CSI reports, then it is said to have L AI If one APU is occupied, then N APUs are available within that OFDM symbol period. CPU,AI -L AI One APU is unoccupied; when L is available within a certain OFDM symbol period AI,SIf a storage capacity unit is used for AI / ML related storage, then within this OFDM symbol period, there are still S... AI,total -L AI,S N storage capacity units are unused; if N remain CPU,AI -L AI One APU is not occupied, S AI,total -L AI,S Within an OFDM symbol period where no storage capacity unit is used, there are N AI Each AI / ML-based CSI report begins to occupy its corresponding APU and its corresponding storage capacity unit (where each CSI report has n = 0, ..., N). AI -1, corresponding to the occupation One APU, and occupy One storage capacity unit, of which and The value of N is determined by the parameters in the first information, then the UE does not need to update N. AI -M AI The lowest priority CSI report; the specific priority definition is not required here, where M AI (satisfies 0≤M) AI ≤N AI To satisfy the inequality and The largest integer value.

[0279] 6. Using the second storage capacity metric according to the method described in Appendix 5. replace Among them, regarding The value of has N within a certain OFDM symbol period. AI There are N CSI reports related to AI / ML functions pending processing. AI A multiset consisting of AI / ML functions corresponding to / associated with each CSI report. In the context, if there is a set of P (P≥1) AI / ML functions p0,…p P-1 The resulting sub-multiset C corresponds to / is associated with the same AI / ML model, and the storage capacity occupied by this model is... Then take the sum of the storage capacity occupied by these P AI / ML functions as S. P ,Right now

[0280] Alternatively, it can be expressed as follows: for the first occurrence of element p0 in the submultiset C, take... Take other elements in set C (p k ∈C, and p k ≠p0); if If there are multiple submultisets, and the AI / ML functions in each submultiset correspond to / are associated with the same AI / ML model, then the above operation is performed on the storage capacity occupied by the AI / ML functions in each submultiset.

[0281] 7. A terminal device comprising a memory and a processor, the memory storing a computer program and the processor being configured to execute the computer program to implement the CSI processing method as described in Appendix 1.

[0282] 8. A network device comprising a memory and a processor, the memory storing a computer program and the processor being configured to execute the computer program to implement the CSI processing method as described in any one of Appendices 2 to 6.

[0283] 9. A computer program product comprising at least a computer program that, when executed by a processor, causes a terminal device to perform the CSI processing method as described in Appendix 1.

[0284] 10. A computer program product comprising at least a computer program that, when executed by a processor, causes a network device to perform the CSI processing method as described in any one of Appendices 2 to 6.

Claims

1. A CSI processing apparatus, comprising: A transmitter that sends first information to a network device, the first information being used at least to report processing complexity and / or storage capacity related to AI / ML functions / models; A receiver that receives second information from a network device, the second information being used to configure and / or trigger a CSI report related to the AI / ML function / model; as well as The processor calculates / determines, at least based on the first information, a processing complexity metric and / or storage capacity metric related to the CSI report.

2. The apparatus according to claim 1, wherein, The processor determines the update information of the CSI report based at least on a computed / determined processing complexity metric and / or storage capacity metric.

3. The apparatus according to claim 1, wherein, The first information includes at least one of the following related to the AI / ML function / model: the maximum AI / ML processing complexity metric supported by the terminal device, the maximum AI / ML storage capacity metric supported by the terminal device, and the list of AI / ML functions supported by the terminal device.

4. The apparatus according to claim 3, wherein, The maximum AI / ML processing complexity is measured by the number of floating-point operations per second, or by the number of floating-point operations per second after quantization / normalization, or by the number of CSI processing units based on AI / ML. The maximum AI / ML storage capacity metric is a floating-point number, or a quantized / normalized value.

5. The apparatus according to claim 3, wherein, The AI / ML feature list includes at least one AI / ML feature.

6. The apparatus according to claim 5, wherein, The AI / ML function includes the following information: The use case information corresponding to the AI / ML function, and / or The processing complexity metric corresponding to the AI / ML function, and / or The storage capacity metric corresponding to the AI / ML function, and / or The inference latency of the AI / ML function, and / or The model loading time of the AI / ML function, and / or The network configuration information corresponding to the AI / ML function.

7. The apparatus according to claim 6, wherein, The use case information includes at least one of the following: beam management use case, positioning use case, CSI prediction use case, and CSI compression use case.

8. The apparatus according to claim 6, wherein, The processing complexity metric is the number of floating-point operations per second, or the number of floating-point operations per second after quantization / normalization, or the number of AI / ML-based CSI processing units; The storage capacity metric is a floating-point number, or a quantized / normalized value.

9. The apparatus according to claim 6, wherein, The network configuration information includes at least one of the following: number of antenna ports or CSI-RS ports, number of sub-bands, and feedback overhead.

10. The apparatus according to claim 1, wherein, The processor determines the update information of the CSI report based at least on a computed / determined processing complexity metric and / or storage capacity metric, as well as third information.

11. The apparatus according to claim 10, wherein, The third piece of information includes at least one of the following: CSI report priority associated with AI / ML, number of CSI-RS resources, and number of ports.

12. The apparatus according to claim 1, wherein, The processor calculates / determines processing complexity metrics and / or storage capacity metrics related to the CSI report, including: The processor calculates the sum of the processing complexity metric of the first AI / ML function to be activated and the processing complexity metric of one or more second AI / ML functions that have been activated, and calculates the sum of the storage capacity metric of the first AI / ML function and the storage capacity metric of the one or more second AI / ML functions.

13. The apparatus according to claim 12, wherein, If the sum of the processing complexity metrics does not exceed the maximum AI / ML processing complexity metric, and the sum of the storage capacity metrics does not exceed the maximum AI / ML storage capacity metric, the processor determines to activate the first AI / ML function / model and updates the occupied processing complexity metric and storage capacity metric.

14. The apparatus according to claim 1, wherein, The processor calculates / determines processing complexity metrics and / or storage capacity metrics related to the CSI report, including: The processor calculates the sum of the processing complexity metric of the first AI / ML function to be activated and the processing complexity metric of one or more second AI / ML functions that have already been activated.

15. The apparatus according to claim 14, wherein, If the sum of the processing complexity metrics does not exceed the maximum AI / ML processing complexity metric, the processor determines to activate the first AI / ML function / model and updates the occupied processing complexity metric.

16. The apparatus according to claim 1, wherein, The processor calculates / determines processing complexity metrics and / or storage capacity metrics related to the CSI report, including: When L is present in a certain OFDM symbol period AI If an APU is used for CSI calculations / processing related to CSI reports, then it is said to have L AI If one APU is occupied, then N APUs are available within that OFDM symbol period. CPU,AI -L AI One APU is not occupied; when L is available in a certain OFDM symbol period AI,S If a storage capacity unit is used for AI / ML related storage, then within this OFDM symbol period, there are still S... AI,total -L AI,S N storage capacity units are unused; if N remain CPU,AI -L AI One APU is not occupied, S AI,total -L AI,S Within an OFDM symbol period where no storage capacity unit is used, there are N AI If an AI / ML-based CSI report begins to occupy its corresponding APU and its corresponding storage capacity unit, then there is no need to update N. AI -M AI The lowest priority CSI reports, where each CSI report has n = 0, ..., N. AI -1, corresponding to the amount occupied One APU, and occupy One storage capacity unit, of which and The value of M is determined by the parameters in the first information, where M AI To satisfy the inequality and The largest integer value, 0≤M AI ≤N AI .

17. The apparatus of claim 16, using a second storage capacity metric. replace in, about The value of , N has a certain OFDM symbol period AI One CSI report related to AI / ML functionality is pending processing, in the N AI The multiset consisting of AI / ML functions corresponding to / associated with each CSI report In the context, if there is a set of P AI / ML functions p0, ... p... P-1 The resulting sub-multiset C corresponds to / is associated with the same AI / ML model, and the storage capacity occupied by this model is... Then take the sum of the storage capacity occupied by these P AI / ML functions as S. P ,Right now Where P≥1; or, for the first occurrence of element p0 in the submultiset C, take... Take other elements in set C p k ∈C, and p k ≠p0.

18. The apparatus according to claim 1, wherein, AI / ML functions with the same use case are assigned to / associated with the same model by default, or the first parameter indicates the AI / ML function that is assigned to / associated with the same model, or the second parameter indicates the AI / ML function that is assigned to / associated with a different model.

19. A CSI processing apparatus, comprising: A receiver receives first information sent by a terminal device, the first information being used at least to report the processing complexity and / or storage capacity related to AI / ML functions / models; A transmitter sends second information to the terminal device, the second information being used to configure and / or trigger a CSI report related to the AI / ML function / model; wherein the first information is used to calculate / determine a processing complexity metric and / or storage capacity metric related to the CSI report.

20. A communication system, comprising: A network device receives first information sent by a terminal device, the first information being used at least to report processing complexity and / or storage capacity related to AI / ML functions / models; and sends second information to the terminal device, the second information being used to configure and / or trigger CSI reports related to the AI / ML functions / models. as well as The terminal device calculates / determines at least based on the first information a processing complexity metric and / or storage capacity metric related to the CSI report.

Images