Ai / ML model or function operation processing method, apparatus, and communication system

Abstract

The embodiments of the present application provide an AI / ML model or function operation processing method, an apparatus and a communication system. The method comprises receiving configuration information from a network device; and sending first information to the network device on the basis of the configuration information, wherein the first information is obtained by a terminal device performing an AI / ML model or function operation according to precoding layers and / or rank values.

Description

Operation processing method, apparatus and communication system of AI / ML model or function TECHNICAL FIELD

[0001] Embodiments of the present application relate to the field of communication technology. BACKGROUND

[0002] In 3GPP Release 18, the application of Artificial Intelligence (AI) and Machine Learning (ML) i.e. AI / ML model or function to air interface is studied, including the application of AI / ML model or function to Channel State Information (CSI) feedback compression. AI / ML based CSI feedback compression adopts a two-sided model, i.e. AI / ML model or function is located at terminal device (UE) side and network (NW) side (i.e. gNB side). In Rel-18, AI / ML based feedback compression compresses CSI in spatial frequency domain (SF-AI / ML CSI compression).

[0003] In Release 19 phase, the time spatial frequency domain AI / ML based CSI compression feedback enhancement sub use case (TSF-AI / ML CSI compression) is further studied. This sub use case utilizes historical CSI information by AI / ML method (e.g. RNN / GRU / LSTM model) to help the current time CSI compression, aiming to obtain lower compression rate or higher feedback accuracy.

[0004] It should be noted that the above introduction to the technical background is only for the convenience of clearly and completely describing the technical solutions of the present application, and for the convenience of understanding by those skilled in the art. The above technical solutions cannot be considered as known to those skilled in the art only because they are described in the background section of the present application.

[0005] SUMMARY

[0006] TSF-AI / ML CSI compression needs to utilize historical CSI information, however whether the historical CSI information is available or can be utilized is affected by factors such as rank change, CSI dropping (e.g. CSI dropping based on layer priority), uplink control information (UCI) loss, etc. When the historical CSI information cannot be acquired or is not available, the performance of TSF-AI / ML model or function will deteriorate, and even cannot work.

[0007] On the other hand, the CSI time domain correlation determines whether the historical CSI assistance information is helpful for the current CSI compression. Too low correlation will lead to invalid input of the AI / ML model or function, which not only cannot improve the CSI compression performance, but even will be regarded as noise by the AI / ML model or function, and deteriorate the CSI compression performance. Therefore, the TSF-AI / ML is more suitable for the scenario with good CSI time domain correlation, for example, the scenario corresponds to a low UE moving speed, a low rank value, and a large eigenvalue layer.

[0008] The inventors find that in the prior art, the life cycle management (LCM) operation of the AI / ML model or function (for example, including selection, activation, deactivation, switching, fallback, update, etc. of the AI / ML model or function) is the same for all layers and / or rank values, that is, without distinguishing the layers and / or rank values.

[0009] However, in some use cases, for example, for AI / ML-based CSI feedback compression, when TSF-AI / ML compression feedback is adopted, if the historical CSI information cannot be obtained or is unavailable, it can be considered to switch some layers to SF-AI / ML compression feedback or non-AI / ML compression feedback. However, the existing mechanism cannot meet the above requirements.

[0010] In order to solve one or more of the above problems, the embodiments of the present application provide an AI / ML model or function operation processing method, device and communication system.

[0011] According to an aspect of the embodiments of the present application, an AI / ML model or function operation processing method is provided, which is applied to a multiple antenna (MIMO) communication system, and the method is applied to a terminal side. The method comprises: receiving configuration information from a network device; and sending first information to the network device according to the configuration information, wherein the first information is obtained after the terminal device performs AI / ML model or function operation according to a precoding layer and / or rank value.

[0012] According to another aspect of the embodiments of the present application, an AI / ML model or function operation processing method is provided, which is applied to a multiple antenna (MIMO) communication system, and the method is applied to a network side. The method comprises: sending configuration information to a terminal device; and receiving first information, wherein the first information is obtained after the terminal device performs AI / ML model or function operation according to a precoding layer and / or rank value.

[0013] According to another aspect of the embodiments of the present application, there is provided an apparatus for processing operation of AI / ML model or function, applied to a multiple antenna (MIMO) communication system, the apparatus being applied to a terminal side, the apparatus comprising: a first receiving unit configured to receive configuration information from a network device; and a first sending unit configured to send first information to the network device according to the configuration information, the first information being obtained after the terminal device performs operation of AI / ML model or function according to a layer and / or rank value of precoding.

[0014] According to another aspect of the embodiments of the present application, there is provided an apparatus for processing operation of AI / ML model or function, applied to a multiple antenna (MIMO) communication system, the apparatus being applied to a network side, the apparatus comprising: a second sending unit configured to send configuration information to a terminal device; and a second receiving unit configured to receive first information, the first information being obtained after the terminal device performs operation of AI / ML model or function according to a layer and / or rank value of precoding.

[0015] According to another aspect of the embodiments of the present application, there is provided 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 above-mentioned method for processing operation of AI / ML model or function at a terminal device side.

[0016] According to another aspect of the embodiments of the present application, there is provided 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 above-mentioned method for processing operation of AI / ML model or function at a network device side.

[0017] According to another aspect of the embodiments of the present application, there is provided a communication system, wherein the system comprises: a terminal device configured to receive configuration information from a network device; and send first information to the network device according to the configuration information, the first information being obtained after the terminal device performs operation of AI / ML model or function according to a layer and / or rank value of precoding; and / or, a network device configured to send configuration information to a terminal device; and receive first information, the first information being obtained after the terminal device performs operation of AI / ML model or function according to a layer and / or rank value of precoding.

[0018] One of the beneficial effects of the embodiments of the present application is that operation of AI / ML model or function according to a layer and / or rank value of precoding can flexibly utilize AI / ML model or function according to the layer and / or rank value, fully utilize the advantages of AI / ML model or function, and guarantee the performance of AI / ML model or function and the performance of the communication system.

[0019] For example, when historical time instance CSI information of some or all layers at UE side and / or network side cannot be obtained / available, based on the mechanism of AI / ML model or function operation according to the precoded layers and / or rank values, these layers can switch to SF-AI / ML or non-AI / ML method for CSI feedback, which can avoid performance degradation of AI / ML model or function and ensure the performance of communication system.

[0020] In addition, in addition to the CSI compression feedback use case, those skilled in the art can apply the method to other various application AI / ML model or function use cases and / or scenarios, and obtain corresponding beneficial effects.

[0021] Specific embodiments of the application are disclosed herein, and will be described in more detail by reference to the following written specification, drawings and attached claims. It is to be understood that the application is not limited in scope to the specific embodiments disclosed, and that such embodiments are disclosed as examples only. It is to be understood that changes in form of detail of the application as described herein can be made without departing from the spirit of the application and that the scope of the application is limited only by the appended claims.

[0022] Features described and / or illustrated with respect to one implementation can be used in one or more other implementations in the same or similar manner, in combination with or in place of features in other implementations, or in place of or in addition to features described or illustrated with respect to other implementations.

[0023] It should be emphasized that the term "comprises / comprising" when used in this specification is taken to mean the presence of stated features, integers, steps or components but does not preclude the presence or addition of one or more other features, integers, steps, components or groups thereof. BRIEF DESCRIPTION OF DRAWINGS

[0024] Elements and features of one or more embodiments described in one figure or implementation of the application can be combined with elements and features illustrated in one or more other figures or implementations of the application. Additionally, in the drawings, like reference numerals can be used to denote like elements throughout the several views.

[0025] FIG. 1 is a schematic diagram of a communication system according to an embodiment of the application;

[0026] FIG. 2 is a schematic diagram of a time-space-frequency domain AI / ML based CSI compression (TSF-AI / ML CSI compression) feedback enhancement sub-use case according to an embodiment of the application;

[0027] FIG. 3 is a schematic diagram of an AI / ML model or function operation processing method according to an embodiment of the application;

[0028] FIG. 4 is another schematic diagram of an AI / ML model or function operation processing method according to an embodiment of the application;

[0029] FIG. 5 is a schematic diagram of one example of operation of AI / ML models or functions per layer in accordance with embodiments of the present application;

[0030] FIG. 6 is a schematic diagram of another example of operation of AI / ML models or functions per layer in accordance with embodiments of the present application;

[0031] FIG. 7 is a schematic diagram of yet another example of operation of AI / ML models or functions per layer in accordance with embodiments of the present application;

[0032] FIG. 8 is a schematic diagram of one example of acquisition or recovery of historical CSI information using output of a TSF-AI / ML model in accordance with embodiments of the present application;

[0033] FIG. 9 is a schematic diagram of an operation processing apparatus of AI / ML models or functions in accordance with embodiments of the present application;

[0034] FIG. 10 is another schematic diagram of an operation processing apparatus of AI / ML models or functions in accordance with embodiments of the present application;

[0035] FIG. 11 is a schematic block diagram of system configuration of a terminal device in accordance with embodiments of the present application;

[0036] FIG. 12 is a schematic block diagram of system configuration of a network device in accordance with embodiments of the present application. DETAILED DESCRIPTION

[0037] The foregoing and other features of the present application will become apparent to those skilled in the art from the following description with reference to the accompanying drawings. In the description and drawings, particular embodiments of the present application are disclosed in detail. It should be understood that the present application is not limited to the embodiments described but includes all modifications, variations, and equivalents that fall within the scope of the appended claims. Various embodiments of the present application are described herein below with reference to the accompanying drawings. These embodiments are merely exemplary and do not limit the present application.

[0038] In embodiments of the present application, the terms "first", "second", and the like are used to distinguish different elements from one another, but do not indicate spatial arrangement or temporal order, and the elements should not be limited by these terms. The term "and / or" includes any one and all combinations of the associated listed terms. The terms "comprise", "include", "have", and the like mean the presence of the stated feature, element, component, or assembly, but do not exclude the presence or addition of one or more other features, elements, components, or assemblies.

[0039] In the embodiments of the present application, the singular form "a", "an", and "the" include the plural form, and should be broadly understood as "one" or "a kind of" rather than the meaning of "one"; in addition, the term "said" should be understood to include both singular and plural forms, unless the context clearly indicates otherwise. In addition, the term "according to" should be understood as "at least partially according to", and the term "based on" should be understood as "at least partially based on", unless the context clearly indicates otherwise.

[0040] In the embodiments of the present application, the term "communication network" or "wireless communication network" can refer to a network that conforms to any communication standard, 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.

[0041] In addition, the communication between devices in the communication system can be carried out according to any stage communication protocol, which can include but is not limited to the following communication protocols: 1G (generation), 2G, 2.5G, 2.75G, 3G, 4G, 4.5G, and 5G, New Radio (NR), 6G and future communication, etc., and / or other currently known or future to be developed communication protocols.

[0042] In the embodiments of the present application, the term "network device" refers to, for example, a device that accesses a terminal device to a communication network and provides services for the terminal device in a communication system. The network device can include but is 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.

[0043] The base station can include, but is not limited to, a Node B (NodeB or NB), an evolved Node B (eNodeB or eNB), a 5G base station (gNB), a 6G base station, and a future base station, etc., and can further include a remote radio head (RRH), a remote radio unit (RRU), a relay, or a low-power node (such as a femto, a pico, etc.). The term "base station" can include some or all functions of them, and each base station can provide communication coverage for 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.

[0044] In the embodiments of the present application, the term "user equipment" (UE) or "terminal equipment" (TE) refers to a device that accesses a communication network through a network device and receives network services, for example. The user equipment can be fixed or mobile, and can also be referred to as a mobile station (MS), a terminal, a user, a subscriber station (SS), an access terminal (AT), a station, a mobile terminal (MT), etc.

[0045] The terminal equipment can include, but is not limited to, the following devices: a cellular phone, a personal digital assistant (PDA), a wireless modem, a wireless communication device, a handheld device, a machine type communication device, a laptop computer, a cordless phone, a smartphone, a smartwatch, a digital camera, etc.

[0046] For another example, in an Internet of Things (IoT) scenario or the like, the user equipment can also be a machine or device that performs monitoring or measurement, which can include, but is not limited to, the following devices: a machine type communication (MTC) terminal, a vehicle-mounted communication terminal, a device-to-device (D2D) terminal, a machine-to-machine (M2M) terminal, a terminal supporting sidelink communication, etc.

[0047] In addition, the term "network side" or "network device side" refers to the side of the network, which can be a certain base station, or can include one or more network devices as above. The term "user side" or "terminal side" or "terminal device side" refers to the side of the user or terminal, which can be a certain UE, or can include one or more terminal devices as above. In this article, "device" can refer to a network device or a terminal device without special indication.

[0048] The terms "uplink control signal" and "uplink control information (UCI, Uplink Control Information)" or "physical uplink control channel (PUCCH, Physical Uplink Control Channel)" can be interchangeable without causing confusion, and the terms "uplink data signal" and "uplink data information" or "physical uplink shared channel (PUSCH, Physical Uplink Shared Channel)" can be interchangeable;

[0049] The terms "downlink control signal" and "downlink control information (DCI, Downlink Control Information)" or "physical downlink control channel (PDCCH, Physical Downlink Control Channel)" can be interchangeable, and the terms "downlink data signal" and "downlink data information" or "physical downlink shared channel (PDSCH, Physical Downlink Shared Channel)" can be interchangeable.

[0050] In addition, the uplink signal can include uplink data signals and / or uplink control signals and / or PRACH and / or SRS (sounding reference signal) and the like, and can also be referred to as uplink transmission (UL transmission) or uplink information or uplink channel. Transmitting / receiving the uplink transmission on the uplink resource can be understood as transmitting / receiving the uplink transmission using the uplink resource. The downlink signal can include downlink data signals and / or downlink control signals and / or synchronization signals (SS, such as PSS / SSS) and / or broadcast channels (PBCH) and / or SSB (SS / PBCH block, including PSS, SSS and PBCH and its DMRS) and / or CSI-RS and the like, and can also be referred to as downlink transmission (DL transmission) or downlink information or downlink channel. Transmitting / receiving the downlink transmission on the downlink resource can be understood as transmitting / receiving the downlink transmission using the downlink resource.

[0051] In embodiments of the present application, the high layer signaling may, for example, be radio resource control (RRC) signaling; the RRC signaling may, for example, include an RRC message, such as a broadcast / common RRC message / signaling (e.g., a master information block (MIB), system information (SI), a dedicated RRC message / signaling; or an RRC information element (RRC IE); or an information field included in the RRC message or RRC information element (or an information field included in the information field). The high layer signaling may, for example, also be medium access control (MAC) signaling; or a MAC control element (MAC CE). However, the present application is not limited thereto.

[0052] In embodiments of the present application, "at least one" and "one or more than one" are interchangeable, "a plurality of" and "more than one" are interchangeable, and "a plurality of" means at least two, or two or more.

[0053] In embodiments of the present application, pre-defined means defined by a protocol or determined according to a rule defined by a protocol, without additional configuration. Configuration / indication means direct or indirect configuration / indication by a network device through high layer signaling and / or physical layer signaling. The configuration / indication may, for example, be achieved by introducing a high layer parameter in the high layer signaling, where the high layer parameter means a field and / or an information element / unit / element (IE) in the high layer signaling. The physical layer signaling may, for example, be control information (DCI) carried by a physical downlink control channel or control information carried by a sequence, but the present application is not limited thereto.

[0054] For ease of description, a base station is taken as an example of an access network device in the following description.

[0055] In the following description, "if", "in the case of", and "when" may be used interchangeably without causing confusion.

[0056] The scenarios of embodiments of the present application are described below by way of example, but the present application is not limited thereto.

[0057] FIG. 1 is a schematic diagram of a communication system according to an embodiment of the present application, which schematically illustrates a scenario taking a terminal device and a network device as an example. As shown in FIG. 1, the communication system 100 may, for example, include a network device 101, a terminal device 102, and a terminal device 103. For simplicity, FIG. 1 illustrates only two terminal devices and one network device as an example, but embodiments of the present application are not limited thereto.

[0058] In the embodiments of the present application, the network device 101, the terminal device 102 and the terminal device 103 can perform existing services or future implementable service transmission. For example, these services can include but are not limited to: enhanced mobile broadband (eMBB, enhanced Mobile Broadband), massive machine type communication (mMTC, massive Machine Type Communication), high reliability and low latency communication (URLLC, Ultra-Reliable and Low-Latency Communication) and related communication of reduced capability terminal devices, etc.

[0059] Wherein, the terminal device 102, 103 can be in RRC_IDLE state, or RRC_INACTIVE state or RRC_CONNECTED state, and the terminal device 102, 103 can also communicate with the network device 101, for example, taking the terminal device 102 as an example, the terminal device 102 can send data to the network device 101, or can perform data retransmission. The network device 101 can send a paging message to the terminal device 102, and can also send data to the terminal device 102, and the terminal device 102 receives the data sent by the network device 101. In addition, different terminal devices can also communicate with each other, for example, the terminal device 102 and the terminal device 103 can exchange data.

[0060] It is worth noting that FIG. 1 shows that the terminal device 102 and the terminal device 103 are both within the coverage range of the network device 101, but the present application is not limited thereto. The terminal device 102 and the terminal device 103 can both be outside the coverage range of the network device 101, or one of the terminal device 102 and the terminal device 103 is within the coverage range of the network device 101 and the other is outside the coverage range of the network device 101.

[0061] In the embodiments of the present application, one or more AI / ML models can be configured and run in the network device and / or the terminal device. The AI / ML model can be used for various signal processing functions of wireless communication, such as CSI prediction, CSI compression, beam prediction, positioning management, etc.; the present application is not limited thereto.

[0062] FIG. 2 is a schematic diagram of an AI / ML-based CSI compression (TSF-AI / ML CSI compression) feedback enhancement sub-use case in a space-time-frequency domain. As shown in FIG. 2, at the UE side, after the CSI measurement result is subjected to matrix decomposition (for example, SVD decomposition (singular value decomposition) or EVD decomposition (eigenvalue decomposition)), a feature vector is obtained, which is input to an encoder to generate CSI feedback information. At the network side, the CSI feedback information sent by the UE is decoded by a decoder to generate reconstructed CSI.

[0063] In some embodiments, the encoder and the decoder at the UE side can employ an AI / ML method (for example, an RNN / LSTM / GRU cascaded Transformer / CNN model, etc.). The AI / ML method can utilize time domain information (which can also be referred to as historical CSI information, side information, or similar names) to assist current CSI compression and / or decompression, aiming to obtain higher compression rate or higher feedback accuracy.

[0064] When utilizing time domain information for CSI compression and / or decompression, whether the time domain information is available and / or can be utilized is easily affected by factors such as rank change, CSI dropping (for example, CSI dropping based on layer priority), uplink control information (UCI) loss, etc. When the time domain information cannot be obtained or is not available, the performance of the AI / ML model will deteriorate or even fail to work.

[0065] In addition, the CSI time domain correlation determines whether the historical CSI information is helpful for current CSI compression and / or decompression. Too low correlation will result in invalid input of the AI / ML model, which not only cannot improve the CSI compression performance, but even will be regarded as noise by the AI / ML model, leading to deterioration of the CSI compression performance. Therefore, AI / ML is more suitable for scenarios with good CSI time domain correlation, which usually corresponds to a case of low UE moving speed, or a layer with low rank value and / or large eigenvalue.

[0066] In the embodiments of the present application, the CSI compression can also be referred to as “CSI encoding”, “CSI generation”, or similar names, and the result of the CSI compression or CSI encoding or CSI generation operation is referred to as CSI feedback information, CSI reporting information, or similar names.

[0067] In the embodiments of the present application, the CSI decompression can also be referred to as “CSI decoding”, “CSI reconstruction”, “CSI recovery”, “CSI reconstruction”, or similar names.

[0068] In the embodiments of the present application, the AI / ML model can also be referred to as an AI / ML method, an AI / ML unit, an AI / ML function, or a similar name of an AI / ML element. On the UE side, the AI / ML model can also be referred to as an encoder, a CSI generation part, or a similar name, and on the network device side, the AI / ML model can also be referred to as a decoder, a CSI reconstruction part, or a similar name.

[0069] It should be noted that the operation processing method of the AI / ML model or function in the embodiments of the present application is applicable to use cases including but not limited to CSI compression feedback, but those skilled in the art can understand that the embodiments of the present application are also applicable to other various use cases and / or scenarios in which AI / ML models or functions are applied.

[0070] Embodiments of the first aspect

[0071] The embodiments of the present application provide an operation processing method of an AI / ML model or function, which is applied to a multiple antenna (MIMO) communication system and described from the terminal device side and / or the network side.

[0072] FIG. 3 is a schematic diagram of the operation processing method of the AI / ML model or function according to the embodiments of the present application. As shown in FIG. 3, from the terminal (UE) side, the method includes:

[0073] 301: receiving configuration information from a network device; and

[0074] 302: sending first information to the network device according to the configuration information, the first information being information obtained after the terminal device performs operation of the AI / ML model or function according to a precoding layer and / or a rank value.

[0075] FIG. 4 is another schematic diagram of the operation processing method of the AI / ML model or function according to the embodiments of the present application. As shown in FIG. 4, from the network (NW) side, the method includes:

[0076] 401: sending configuration information to a terminal device;

[0077] 402: receiving first information, the first information being information obtained after the terminal device performs operation of the AI / ML model or function according to a precoding layer and / or a rank value.

[0078] According to the above embodiments, the operation of the AI / ML model or function according to the precoded layer and / or rank value can flexibly utilize the AI / ML model or function according to the layer and / or rank value, fully play the advantages of the AI / ML model or function, and guarantee the performance of the AI / ML model or function and the performance of the communication system.

[0079] In some embodiments, the AI / ML model on the UE side and the AI / ML model on the network side can be corresponding or matched. The present application is not limited thereto, and the AI / ML model can be used only on the UE side or only on the network side. In the subsequent content, the AI / ML model is the model on the UE side and / or the model on the network side unless otherwise specified.

[0080] In some embodiments, the operation of the AI / ML model or function according to the precoded layer and / or rank value can be referred to as the operation of the AI / ML model or function according to the layer and / or rank value.

[0081] In some embodiments, the rank value represents the number of precoded layers. For example, the rank value of 1 represents that the number of precoded layers is 1, i.e., the first layer; the rank value of 2 represents that the number of precoded layers is 2, i.e., the first layer and the second layer; and so on.

[0082] In some embodiments, the configuration information is used at least for sending the first information, for example, the configuration information includes but is not limited to resource configuration information and / or reporting configuration information for sending the first information.

[0083] In some embodiments, the configuration information can include one or more predefined rules for the operation of the AI / ML model or function according to the layer and / or rank value; for example, the rules include but are not limited to: which layers to perform the operation of one or more AI / ML models or functions, and / or which layers to perform the operation of one or more AI / ML models or functions when the rank value changes.

[0084] Alternatively, the rules can not be included in the configuration information, but be sent through another configuration information or other information.

[0085] In some embodiments, the configuration information can include an indication that the terminal device can perform the operation of one or more AI / ML models or functions on a certain layer or certain layers.

[0086] Alternatively, the indication can not be included in the configuration information, but be sent through another configuration information or other information.

[0087] In some embodiments, one or more of the above configuration information can be sent to the terminal device through RRC signaling. The present application is not limited thereto, and one or more of the above configuration information can also be sent through other manners.

[0088] In some embodiments, the operation of the AI / ML model or function is also referred to as the life cycle management (LCM) operation of the AI / ML model or function.

[0089] In some embodiments, the operation of the AI / ML model or function includes but is not limited to one or more of the following operations:

[0090] selection of the AI / ML model or function;

[0091] activation of the AI / ML model or function;

[0092] deactivation of the AI / ML model or function;

[0093] switching of the AI / ML model or function;

[0094] updating of the AI / ML model or function;

[0095] fallback of the AI / ML model or function.

[0096] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on a predefined rule, which includes but is not limited to: on which layer to perform the operation of a certain AI / ML model or function, and / or on which layer to perform the operation of a certain AI / ML model or function when the rank value changes.

[0097] In some embodiments, the rule is related to the change of rank indication (RI): when the feedback RI value is different from the last feedback RI value, the operation of the AI / ML model or function is performed according to the layer and / or rank value;

[0098] The terminal device implicitly indicates the network device whether it performs the operation of the AI / ML model or function through the RI value: when the RI changes, the terminal device takes the operation; when the RI does not change, the terminal device does not operate.

[0099] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on: the network side configures one or more predefined rules, which includes but is not limited to: on which layer to perform the operation of a certain AI / ML model or function, and / or on which layer to perform the operation of a certain AI / ML model or function when the rank value changes;

[0100] In addition, as described above, the rule configured by the network side can be included in the configuration information or not included in the configuration information.

[0101] The terminal device selects one of the rules, performs the operation, and notifies the network side of the corresponding operation, for example, the notification includes explicit or implicit notification, for example, by reporting RI, resource allocation, CSI discard, etc.

[0102] In some embodiments, the notification includes at least one of the following two notifications:

[0103] The network device is notified whether the terminal device has performed the operation of the AI / ML model or function;

[0104] The network device is notified which rule the terminal device is based on to perform the operation of the AI / ML model or function.

[0105] In some embodiments, the rule is related to RI change: when the feedback RI value is different from the last feedback RI value, the terminal device performs the operation of the AI / ML model or function based on a certain rule according to the layer and / or rank value.

[0106] The terminal device implicitly indicates the network device whether it has performed the operation of the AI / ML model or function through the RI value feedback: when the RI changes, the terminal device takes the operation; when the RI does not change, the terminal device has no operation.

[0107] The terminal device explicitly indicates the network device which rule it is based on to perform the operation of the AI / ML model or function through certain indication information. When the network device only configures one rule, then the indication information can be default or the overhead is 0.

[0108] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on the network side indicating the terminal device that it can perform a certain AI / ML model or function operation on a certain layer or certain layers;

[0109] In addition, as described above, the indication of the network side can be included in the configuration information or not included in the configuration information.

[0110] In some embodiments, the configuration information is determined by the network device based on the terminal device requesting the operation of the AI / ML model or function.

[0111] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on that the terminal device decides to perform the operation of the AI / ML model or function on a certain layer or layers and informs the network device of the operation. For example, the terminal device itself judges that the operation of the AI / ML model or function on a certain layer or layers needs to be performed and informs the network device of the specific operation.

[0112] In some embodiments, as shown in FIG. 3, the method further includes:

[0113] 303: performing the operation of the AI / ML model or function according to the layer and / or rank value;

[0114] For example, the terminal device performs the operation of the AI / ML model or function according to the layer and / or rank value to obtain or generate the first information.

[0115] In some embodiments, similarly, as shown in FIG. 4, the method further includes:

[0116] 403: performing the operation of the AI / ML model or function according to the layer and / or rank value;

[0117] For example, the network device performs the operation of the AI / ML model or function according to the layer and / or rank value on the received first information; and / or, the network device performs the operation of the AI / ML model or function according to the layer and / or rank value on other information or other conditions. That is, the operation of the AI / ML model or function according to the layer and / or rank value by the network device can be performed on the first information, or on other information or other conditions.

[0118] In operation 303 and / or operation 403, for example, the terminal device and / or the network device performs the operation of the AI / ML model or function on part or all of the layers and / or rank values, and / or performs the same or different operation of the AI / ML model or function on different layers and / or rank values.

[0119] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is applied to a two-sided AI / ML model or function, or to a one-sided AI / ML model or function.

[0120] For the case where the operation of the AI / ML model or function according to the layer and / or rank value is applied to a two-sided AI / ML model or function, the operation of the AI / ML model or function according to the layer and / or rank value on the terminal side and the network side is corresponding.

[0121] In some embodiments, the operation of AI / ML models or functions according to layer and / or rank values is applied at least to one or more of:

[0122] rank specific AI / ML models or functions;

[0123] rank common AI / ML models or functions;

[0124] layer specific AI / ML models or functions;

[0125] layer common AI / ML models or functions,

[0126] wherein the layer specific AI / ML models or functions include: layer specific and rank common AI / ML models or functions, and / or, layer specific and rank specific AI / ML models or functions;

[0127] the layer common AI / ML models or functions include: layer common and rank common AI / ML models or functions, and / or, layer common and rank specific AI / ML models or functions.

[0128] In some embodiments, as shown in FIG. 3, the method further comprises:

[0129] 304: for the operation of AI / ML models or functions according to layer and / or rank values, monitoring of the AI / ML models or functions according to layer and / or rank values.

[0130] In some embodiments, similarly, as shown in FIG. 4, the method further comprises:

[0131] 404: for the operation of AI / ML models or functions according to layer and / or rank values, monitoring of the AI / ML models or functions according to layer and / or rank values.

[0132] Operation 304 and operation 404 are optional operations.

[0133] In some embodiments, the monitoring of the AI / ML model or function includes: monitoring the performance of the model or function of at least one layer and / or rank; and / or, monitoring whether historical CSI information of at least one layer and / or rank is available.

[0134] It is worth noting that the above Figs. 3 and 4 only schematically illustrate the embodiments of the present application, but the present application is not limited thereto. For example, the execution order between various operations can be appropriately adjusted, and in addition, some operations can be added or some operations can be reduced. Those skilled in the art can appropriately modify the above content, and the present application is not limited to the above Figs. 3 and 4.

[0135] The operation of the AI / ML model or function according to the layer and / or rank value is described below through several examples.

[0136] Fig. 5 is a schematic diagram of one example of the operation of the AI / ML model or function according to the layer in the embodiments of the present application, in which the rank value is unchanged.

[0137] As shown in Fig. 5, at T0 and T1, both the first layer and the second layer use AI / ML model A for inference. At T2, the AI / ML model A of the second layer is switched to AI / ML model B, and the AI / ML model of the first layer remains unchanged. For example, the AI / ML model A is TSF-AI / ML, and the UE and / or the network monitor that the CSI time domain correlation of the second layer or the TSF-AI / ML feedback accuracy is too low, and the second layer is switched to SF-AI / ML with lower complexity, i.e., AI / ML model B.

[0138] Fig. 6 is a schematic diagram of another example of the operation of the AI / ML model or function according to the layer in the embodiments of the present application, in which the rank value changes.

[0139] As shown in Fig. 6, at T1, the rank value changes from 2 to 1, at which time the AI / ML model A only works in the first layer. At T2, the rank value changes from 1 to 2, at which time the AI / ML model A of the second layer is switched to model B. For example, for TSF-AI / ML, since the rank = 1 at T1, the AI / ML model cannot obtain the CSI information of the second layer at T2. Considering that the TSF-AI / ML inference result at the first time (T2) after the loss of historical CSI information can have poor performance or have no advantage compared to SF-AI / ML, the second layer is switched to SF-AI / ML with lower complexity, i.e., AI / ML model B.

[0140] Fig. 7 is a schematic diagram of still another example of the operation of the AI / ML model or function according to the layer in the embodiments of the present application, in which the rank value is unchanged.

[0141] As shown in FIG. 7, Model A2 is used for Rank = 2. At T2, AI / ML Model A2 for Rank = 2 is switched to AI / ML Model B2. For example, AI / ML Model A2 is TSF-AI / ML, whose input is 2-layer CSI. The UE and / or the network monitor that the CSI time-domain correlation or the TSF-AI / ML feedback accuracy is too low, and switch the AI / ML model at this time to SF-AI / ML, that is, AI / ML Model B2.

[0142] The embodiments of the present application are described below by taking the application to CSI compression feedback as an example, but the embodiments of the present application are not limited thereto, and can also be applied to other various application AI / ML model or function use cases and / or scenarios.

[0143] In some embodiments, for CSI compression feedback based on a bilateral AI / ML model, the CSI compression feedback scheme is determined according to the layer.

[0144] For example, for CSI compression feedback based on a bilateral AI / ML model, the same CSI compression feedback scheme is used on all layers at the same time, or different CSI compression feedback schemes are used on different layers.

[0145] At different times, the same CSI compression feedback scheme is used on the same layer, or different CSI compression feedback schemes can be used on the same layer.

[0146] In some embodiments, the CSI compression feedback scheme includes but is not limited to one or more of the following:

[0147] AI / ML-based space-time-frequency domain (TSF-AI / ML) CSI compression feedback;

[0148] AI / ML-based joint CSI prediction and CSI compression feedback;

[0149] AI / ML-based space-frequency domain (SF-AI / ML) CSI compression feedback;

[0150] CSI compression feedback based on a non-AI / ML method.

[0151] In some embodiments, for TSF-AI / ML CSI compression feedback, the AI / ML model or function needs CSI information at one or more historical times,

[0152] When the historical time instance CSI information of one or more layers at the terminal side and / or the network side cannot be obtained or is unavailable, the layers switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, so that the TSF-AI / ML model or function obtains or restores or resets the historical CSI information; after the historical CSI information is obtained or restored or reset, the TSF-AI / ML CSI compression feedback is activated or used at the layers.

[0153] In some embodiments, the switching is applied to the initial running or reactivation of the TSF-AI / ML model or function, or is applied to the case where the historical CSI information is unavailable due to a change in the rank value or loss of UCI or CSI discard.

[0154] In some embodiments, one or more layers need to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method within a time window.

[0155] For example, the time window can also be referred to as a warming up window or an initialization window or a re-initialization window, and the embodiments of the present application do not limit the name of the time window.

[0156] In some embodiments, the time window is defined according to the number of time instances required for the TSF-AI / ML model or function to obtain or restore or reset the historical CSI information, and the definition is explicit or implicit.

[0157] In some embodiments, the time window is applied to the initial running or reactivation of the TSF-AI / ML model or function, or is applied to the case where the historical CSI information is unavailable due to a change in the rank value or loss of UCI or CSI discard.

[0158] In some embodiments, the time window is predefined in a standard, or is updated and / or indicated based on RRC configuration and / or through MAC-CE or DCI.

[0159] In some embodiments, the time window is used at the terminal side and / or the network side, and when the time window is used at the terminal side and the network side, the time windows at the two sides can be the same or different.

[0160] In some embodiments, the size of the time window is the same or different for different layers and / or rank values.

[0161] In some embodiments, the size of the time window depends on the terminal capability of the terminal device.

[0162] In some embodiments, whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method within the time window is configured by predefinition, or based on RRC signaling configuration and / or based on MAC CE or DCI update and / or indication.

[0163] In some embodiments, whether to support switching to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method depends on the terminal capability of the terminal device.

[0164] In some embodiments, whether to support running TSF-AI / ML CSI compression feedback and non-TSF-AI / ML CSI compression feedback simultaneously depends on the terminal capability of the terminal device.

[0165] In some embodiments, whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method within the time window is determined by the terminal device.

[0166] In some embodiments, whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method at one or more layers is reported by the terminal device to the network device.

[0167] In some embodiments, TSF-AI / ML CSI compression feedback is only applied to the first layer, and SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method is applied to other layers, or,

[0168] SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method is applied to the first layer and other layers.

[0169] Next, for the case where the historical CSI information of a certain layer or certain layers at the terminal device side and / or network side cannot be obtained or is unavailable in TSF-AI / ML CSI compression feedback, the following is explained.

[0170] In some embodiments, for TSF-AI / ML CSI compression feedback, when the historical CSI information of one or more layers at the terminal device side and / or network side cannot be obtained or is unavailable, one or more of the following methods can be used to obtain or recover the historical CSI information:

[0171] Manner 1: The layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the output of the TSF-AI / ML encoder about the layer is still sent to the network side through the air interface for updating the historical CSI information of the network side; wherein the network side inputs the output of the feedback TSF-AI / ML encoder into the TSF-AI / ML decoder to complete the acquisition or recovery of the historical CSI information.

[0172] Manner 2: The network side has a proxy or approximate TSF encoder, in which case the network side acquires or recovers the historical CSI information at one or more time instants through SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method; wherein the network side inputs the reconstructed eigenvectors and / or channel matrix of the layer through SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method into the proxy or approximate TSF encoder, and then inputs the output of the proxy or approximate TSF encoder into the TSF decoder to obtain or recover the historical CSI information of the network side.

[0173] Manner 3: When the historical CSI information of the network side is reconstructed CSI, the layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the reconstructed CSI based on SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method is used for acquisition or recovery of the historical CSI information.

[0174] For example, for the above-mentioned manner 1, SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method is used for CSI feedback, TSF-AI / ML is involved in feedback, but does not involve CSI feedback and PDSCH scheduling, and is only used for acquisition / recovery of historical CSI information.

[0175] For example, for the above-mentioned manner 2 and manner 3, for layers for which historical CSI information cannot be acquired or is unavailable, the LCM switches the TSF-AI / ML model to a SF-AI / ML model, or a non-AI / ML method (for example, fallback to an existing codebook scheme). The SF-AI / ML or non-AI / ML method of these layers is not only used for CSI feedback, but also used for acquisition / recovery of historical CSI information.

[0176] In some embodiments, the way of obtaining or recovering the historical CSI information is configured by the network side, that is, the network side can configure which way or combination of multiple ways to obtain or recover the historical CSI information;

[0177] Alternatively, the terminal device can also report to the network side the recommended way (which way or combination of multiple ways) for use, and the network side finally determines the way of obtaining or recovering the historical CSI information.

[0178] FIG. 8 is a schematic diagram of one example of the use of the output of the TSF-AI / ML model for obtaining or recovering the historical CSI information according to an embodiment of the present application.

[0179] As shown in FIG. 8, at the second time (T1 time), the rank value changes from 1 to 2, so the TSF-AI / ML model cannot obtain or cannot use the historical CSI information of the second layer, and the LCM switches the second layer to the SF-AI / ML model for CSI feedback. In addition, at the second time and the third time (T1 and T2 times), the output of the TSF-AI / ML model about the second layer is also fed back to the network side for obtaining or updating the historical CSI information, at this time the size of the time window is 2 (2 times). At the fourth time (T3 time), the historical CSI information of the second layer has been obtained or recovered, so the first layer and the second layer both use the TSF-AI / ML model for CSI feedback.

[0180] The above embodiments are only exemplary, but the present application is not limited thereto, and appropriate modifications can be made on the basis of the above embodiments. For example, each of the above embodiments can be used alone, or one or more of the above embodiments can be combined.

[0181] According to the above embodiments, the AI / ML model or function operates according to the precoding layer and / or rank value, which can flexibly use the AI / ML model or function according to the layer and / or rank value, fully utilize the advantages of the AI / ML model or function, and ensure the performance of the AI / ML model or function and the performance of the communication system;

[0182] For example, when the historical time CSI information of some or all layers at the UE side and / or the network side cannot be obtained / used, based on the mechanism of the AI / ML model or function operating according to the precoding layer and / or rank value, these layers can be switched to the SF-AI / ML or non-AI / ML method for CSI feedback, which can avoid the performance degradation of the AI / ML model or function and ensure the performance of the communication system.

[0183] Embodiments of the second aspect

[0184] The embodiments of the present application provide an AI / ML model or function operation processing apparatus. The apparatus can be a terminal device, or a component or assembly configured to the terminal device, which corresponds to the method applied to the terminal device side in the embodiments of the first aspect. The same content as the embodiments of the first aspect will not be described again.

[0185] FIG. 9 is a schematic diagram of an AI / ML model or function operation processing apparatus according to an embodiment of the present application. As shown in FIG. 9, the AI / ML model or function operation processing apparatus 900 includes:

[0186] a first receiving unit 901 configured to receive configuration information from a network device;

[0187] a first sending unit 902 configured to send first information to the network device according to the configuration information, the first information being information obtained after the terminal device performs AI / ML model or function operation according to pre-coded layers and / or rank values.

[0188] In some embodiments, the AI / ML model or function operation includes one or more of the following operations:

[0189] selection of an AI / ML model or function;

[0190] activation of an AI / ML model or function;

[0191] deactivation of an AI / ML model or function;

[0192] switching of an AI / ML model or function;

[0193] updating of an AI / ML model or function;

[0194] fallback of an AI / ML model or function.

[0195] In some embodiments, the AI / ML model or function operation according to the layers and / or rank values is based on a predefined rule, which at least includes: which layers to perform one or more AI / ML model or function operations, and / or which layers to perform one or more AI / ML model or function operations when the rank value changes.

[0196] In some embodiments, the rule is related to rank indication (RI) change: when the feedback RI value is different from the last feedback RI value, perform AI / ML model or function operation according to the layers and / or rank values;

[0197] The terminal device implicitly indicates to the network device whether it has performed the operation of the AI / ML model or function through the feedback of the RI value: when the RI changes, the terminal device takes the operation; when the RI does not change, the terminal device does not take the operation.

[0198] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on that the network side configures one or more predefined rules, which at least include: on which layers to perform the operation of a certain AI / ML model or function, and / or on which layers to perform the operation of a certain AI / ML model or function when the rank value changes; the rules configured by the network side are included or not included in the configuration information,

[0199] The terminal device selects one of the rules, performs the operation, and notifies the network side of the corresponding operation, which includes explicit or implicit notification.

[0200] In some embodiments, the notification includes at least one of the following two notifications:

[0201] The network device is notified whether the terminal device has performed the operation of the AI / ML model or function.

[0202] The network device is notified which rule the terminal device has performed the operation of the AI / ML model or function based on.

[0203] In some embodiments, the rules are related to the change of the RI: when the feedback RI value is different from the last feedback RI value, the terminal device performs the operation of the AI / ML model or function according to a certain rule according to the layer and / or rank value.

[0204] The terminal device implicitly indicates to the network device whether it has performed the operation of the AI / ML model or function through the feedback of the RI value: when the RI changes, the terminal device takes the operation; when the RI does not change, the terminal device does not take the operation.

[0205] The terminal device explicitly indicates to the network device which rule it has performed the operation of the AI / ML model or function based on through certain indication information. When the network device only configures one rule, the indication information can be default or the overhead is 0.

[0206] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on that the network side indicates to the terminal device that it can perform the operation of a certain AI / ML model or function on a certain layer or layers; the indication of the network side is included or not included in the configuration information;

[0207] In some embodiments, the configuration information is determined by the network device based on a request of the terminal device for AI / ML model or function operation.

[0208] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is based on that the terminal device decides to perform a certain AI / ML model or function operation on a certain layer or layers and informs the network device of the operation. For example, the terminal device itself judges that a certain AI / ML model or function operation on a certain layer or layers needs to be performed and informs the network device of the specific operation.

[0209] In some embodiments, as shown in FIG. 9, the apparatus 900 further comprises:

[0210] a processing unit 903 configured to perform AI / ML model or function operation according to the layer and / or rank value, wherein the processing unit 903 performs AI / ML model or function operation on part or all of the layers and / or rank values, and / or the processing unit 903 performs the same or different AI / ML model or function operation on different layers and / or rank values.

[0211] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is applied to a two-sided AI / ML model or function, or to a one-sided AI / ML model or function.

[0212] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is applied to at least one or more of the following:

[0213] a rank specific AI / ML model or function;

[0214] a rank common AI / ML model or function;

[0215] a layer specific AI / ML model or function;

[0216] a layer common AI / ML model or function,

[0217] The layer specific AI / ML model or function includes a layer specific and rank common AI / ML model or function, and / or a layer specific and layer specific AI / ML model or function.

[0218] The layer common AI / ML model or function includes a layer common and rank common AI / ML model or function, and / or a layer common and rank specific AI / ML model or function.

[0219] In some embodiments, as shown in FIG. 9, the apparatus 900 further includes:

[0220] The monitoring unit 904 monitors the AI / ML model or function according to the layer and / or rank value for the operation of the AI / ML model or function according to the layer and / or rank value,

[0221] The monitoring of the AI / ML model or function includes monitoring the performance of the model or function of at least one layer and / or rank; and / or monitoring whether historical CSI information of at least one layer and / or rank is available.

[0222] In some embodiments, the processing unit determines a CSI compression feedback scheme according to the layer for the CSI compression feedback based on the bilateral AI / ML model.

[0223] In some embodiments, for the CSI compression feedback based on the bilateral AI / ML model,

[0224] At the same time, the same CSI compression feedback scheme is used on all layers, or different CSI compression feedback schemes are used on different layers.

[0225] At different times, the same CSI compression feedback scheme is used on the same layer, or different CSI compression feedback schemes can be used on the same layer.

[0226] In some embodiments, the CSI compression feedback scheme includes one or more of the following:

[0227] AI / ML based space-time-frequency domain (TSF-AI / ML) CSI compression feedback;

[0228] AI / ML based joint CSI prediction and CSI compression feedback;

[0229] AI / ML based spatial-frequency domain (SF-AI / ML) CSI compression feedback;

[0230] CSI compression feedback based on non-AI / ML method.

[0231] In some embodiments, for TSF-AI / ML CSI compression feedback, the AI / ML model or function needs CSI information of one or more historical time instants,

[0232] When the historical time instant CSI information of one or more layers at the terminal side and / or network side cannot be obtained or is unavailable, the layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, so that the TSF-AI / ML model or function obtains or restores or resets the historical CSI information; after the historical CSI information is obtained or restored or reset, the TSF-AI / ML CSI compression feedback is activated or used at the layer.

[0233] In some embodiments, within a time window, one or more layers need to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method.

[0234] In some embodiments, the time window is predefined in the standard, or is configured based on RRC and / or updated and / or indicated by MAC-CE or DCI.

[0235] In some embodiments, whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method within the time window is configured by predefinition, or is configured based on RRC signaling and / or updated and / or indicated based on MAC CE or DCI.

[0236] In some embodiments, TSF-AI / ML CSI compression feedback is only applied to the first layer, SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method is applied to other layers, or SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method is applied to the first layer and other layers.

[0237] In some embodiments, for TSF-AI / ML CSI compression feedback, when the historical CSI information of one or more layers at the terminal device side and / or network side cannot be obtained or is unavailable, one or more of the following methods are used to obtain or restore the historical CSI information:

[0238] The layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the output of the TSF-AI / ML encoder with respect to the layer is still sent to the network side through the air interface for updating of historical CSI information of the network side;

[0239] The network side has a proxy or approximate TSF encoder, in which case the network side obtains or recovers historical CSI information at one or more time points through SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method;

[0240] When the historical CSI information of the network side is reconstructed CSI, the layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the reconstructed CSI based on the SF-AI / ML CSI compression feedback or the CSI compression feedback based on a non-AI / ML method is used for obtaining or recovering of the historical CSI information.

[0241] The above embodiments are only exemplary and the present application is not limited thereto. For example, one or more of the above embodiments can be combined.

[0242] It should be noted that the above only describes the components or modules related to the present application, but the present application is not limited thereto. The operation processing apparatus 900 of the AI / ML model or function can also include other components or modules, and the specific content of these components or modules can be referred to related technologies.

[0243] In addition, for the sake of simplicity, only the connection relationship or signal path between the components or modules is exemplarily shown in FIG. 9, but it should be clear to those skilled in the art that various related technologies such as bus connection can be used. The above components or modules can be implemented by hardware facilities such as processors, memories, transmitters, receivers, etc.; the present application is not limited thereto.

[0244] According to the above embodiments, the AI / ML model or function is operated according to the precoding layer and / or rank value, which can flexibly utilize the AI / ML model or function according to the layer and / or rank value, fully utilize the advantages of the AI / ML model or function, and guarantee the performance of the AI / ML model or function and the performance of the communication system;

[0245] For example, when historical time instance CSI information of some layer or some layers on the UE side and / or the network side cannot be obtained / are unavailable, based on the mechanism of AI / ML model or function operation according to the precoded layer and / or rank value, the layers can switch to SF-AI / ML or non-AI / ML method for CSI feedback, which can avoid performance degradation of the AI / ML model or function and ensure the performance of the communication system.

[0246] Embodiments of the third aspect

[0247] Embodiments of the present application provide an AI / ML model or function operation processing apparatus. The apparatus may, for example, be a network device, or a certain component or component group configured in the network device, which corresponds to the method applied to the network side in the embodiments of the first aspect, and the same content as the embodiments of the first aspect will not be repeated.

[0248] FIG. 10 is another schematic diagram of the AI / ML model or function operation processing apparatus according to an embodiment of the present application. As shown in FIG. 10, the AI / ML model or function operation processing apparatus 1000 includes:

[0249] A second sending unit 1001 configured to send configuration information to a terminal device;

[0250] A second receiving unit 1002 configured to receive first information, wherein the first information is obtained after the terminal device performs AI / ML model or function operation according to a precoded layer and / or rank value.

[0251] In some embodiments, the AI / ML model or function operation includes one or more of the following operations:

[0252] Selection of the AI / ML model or function;

[0253] Activation of the AI / ML model or function;

[0254] Deactivation of the AI / ML model or function;

[0255] Switching of the AI / ML model or function;

[0256] Updating of the AI / ML model or function;

[0257] Fallback of the AI / ML model or function.

[0258] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on a predefined rule, which at least includes: which layer(s) to perform the operation of which AI / ML model or function, and / or which layer(s) to perform the operation of which AI / ML model or function when the rank value changes.

[0259] In some embodiments, the rule is related to the change of rank indication (RI): when the feedback RI value is different from the last feedback RI value, the operation of the AI / ML model or function according to the layer and / or rank value is based on a certain rule.

[0260] The terminal device implicitly indicates to the network device whether it has performed the operation of the AI / ML model or function through the RI value feedback: when the RI changes, the terminal device takes the operation; when the RI does not change, the terminal device does not operate.

[0261] In some embodiments, the operation of the AI / ML model or function according to the layer and / or rank value is based on: the network side configures one or more predefined rules, which at least include: which layer(s) to perform the operation of which AI / ML model or function, and / or which layer(s) to perform the operation of which AI / ML model or function when the rank value changes; the rule configured by the network side is included or not included in the configuration information,

[0262] The terminal device selects one of the rules, performs the operation and notifies the network side of the corresponding operation, which includes explicit or implicit notification.

[0263] In some embodiments, the notification includes at least one of the following two notifications:

[0264] The network device is notified whether the terminal device has performed the operation of the AI / ML model or function;

[0265] The network device is notified which rule the terminal device is based on to perform the operation of the AI / ML model or function.

[0266] In some embodiments, the rule is related to the change of RI: when the feedback RI value is different from the last feedback RI value, the terminal device performs the operation of the AI / ML model or function according to the layer and / or rank value based on a certain rule.

[0267] The terminal device implicitly indicates to the network device whether it has performed the operation of the AI / ML model or function through the RI value feedback: when the RI changes, the terminal device takes the operation; when the RI does not change, the terminal device does not operate.

[0268] The terminal device displays indication information indicating which rule the network device is based on for AI / ML model or function operation. When the network device only configures one rule, the indication information can be default or 0 overhead.

[0269] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is based on that the network side indicates that the terminal device can perform one or more AI / ML model or function operations on one or more layers; and the indication of the network side is included or not included in the configuration information.

[0270] In some embodiments, the configuration information is determined by the network device based on the terminal device requesting AI / ML model or function operation.

[0271] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is based on that the terminal device decides to perform one or more AI / ML model or function operations on one or more layers and informs the network device of the operation. For example, the terminal device itself judges that it needs to perform one or more AI / ML model or function operations on one or more layers and informs the network device of the specific operation.

[0272] In some embodiments, as shown in FIG. 10, the apparatus 1000 further includes:

[0273] The processing unit 1003 performs AI / ML model or function operation according to the layer and / or rank value, wherein the processing unit 1003 performs AI / ML model or function operation on part or all of the layers and / or rank values, and / or the processing unit 1003 performs the same or different AI / ML model or function operation on different layers and / or rank values.

[0274] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is applied to a two-sided AI / ML model or function, or to a one-sided AI / ML model or function.

[0275] In some embodiments, the AI / ML model or function operation according to the layer and / or rank value is applied to at least one or more of the following:

[0276] Rank specific AI / ML model or function;

[0277] Rank common AI / ML model or function;

[0278] Layer specific AI / ML model or function;

[0279] Layer common AI / ML model or function,

[0280] wherein the Layer specific AI / ML model or function comprises: Layer specific and rank common AI / ML model or function, and / or, Layer specific and rank specific AI / ML model or function;

[0281] the Layer common AI / ML model or function comprises: Layer common and rank common AI / ML model or function, and / or, Layer common and rank specific AI / ML model or function.

[0282] In some embodiments, as shown in FIG. 10, the apparatus 1000 further comprises:

[0283] a monitoring unit 1004 configured to monitor the AI / ML model or function according to the layer and / or rank value for the operation of the AI / ML model or function according to the layer and / or rank value,

[0284] the monitoring of the AI / ML model or function comprises: monitoring the performance of the model or function of at least one layer and / or rank; and / or, monitoring whether the historical CSI information of at least one layer and / or rank is available.

[0285] In some embodiments, the processing unit determines the CSI compression feedback scheme according to the layer for the CSI compression feedback based on the bilateral AI / ML model.

[0286] In some embodiments, for the CSI compression feedback based on the bilateral AI / ML model,

[0287] At the same time, the same CSI compression feedback scheme is used on all layers, or different CSI compression feedback schemes are used on different layers;

[0288] At different times, the same CSI compression feedback scheme is used on the same layer, or different CSI compression feedback schemes can be used on the same layer.

[0289] In some embodiments, the CSI compression feedback scheme comprises one or more of the following:

[0290] AI / ML based spatio-temporal-frequency domain (TSF-AI / ML) CSI compression feedback;

[0291] AI / ML based joint CSI prediction and CSI compression feedback;

[0292] AI / ML based spatio-frequency domain (SF-AI / ML) CSI compression feedback;

[0293] CSI compression feedback based on non-AI / ML method.

[0294] In some embodiments, for TSF-AI / ML CSI compression feedback, the AI / ML model or function needs CSI information of one or more historical time instants,

[0295] When the CSI information of one or more historical time instants of one or more layers at the terminal side and / or network side cannot be obtained or is unavailable, the layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method, so that the TSF-AI / ML model or function obtains or recovers or resets the historical CSI information; when the historical CSI information is obtained or recovered or reset, the TSF-AI / ML CSI compression feedback is activated or used at the layer.

[0296] In some embodiments, one or more layers need to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method within a time window.

[0297] In some embodiments, the time window is predefined in the standard, or is configured based on RRC and / or updated and / or indicated by MAC-CE or DCI.

[0298] In some embodiments, whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method within the time window is configured by predefinition, or is configured based on RRC signaling and / or updated and / or indicated based on MAC CE or DCI.

[0299] In some embodiments, TSF-AI / ML CSI compression feedback is only applied to the first layer, SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method is applied to other layers, or SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method is applied to the first layer and other layers.

[0300] In some embodiments, for TSF-AI / ML CSI compression feedback, when the historical CSI information of one or more layers at the terminal device side and / or the network side cannot be obtained or is unavailable, one or more of the following methods is used to obtain or recover the historical CSI information:

[0301] The layer is switched to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the output of the SF-AI / ML encoder with respect to the layer is still sent to the network side through the air interface for updating the historical CSI information of the network side;

[0302] The network side has a proxy or approximate TSF encoder, in which case the network side obtains or recovers the historical CSI information at one or more time points through SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method;

[0303] When the historical CSI information of the network side is reconstructed CSI, the layer is switched to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the reconstructed CSI based on the SF-AI / ML CSI compression feedback or the CSI feedback based on the non-AI / ML method is used to obtain or recover the historical CSI information.

[0304] The above embodiments are only exemplary, but the present application is not limited thereto, and appropriate modifications can be made on the basis of the above embodiments. For example, each of the above embodiments can be used alone, or one or more of the above embodiments can be combined.

[0305] It should be noted that the above only describes the components or modules related to the present application, but the present application is not limited thereto. The AI / ML model or function operation processing apparatus 1000 can also include other components or modules, and the specific content of these components or modules can be referred to related technologies.

[0306] In addition, for the sake of simplicity, only the connection relationship or signal path between each component or module is exemplarily shown in FIG. 10, but it should be clear to those skilled in the art that various related technologies such as bus connection can be used. The above-mentioned components or modules can be realized by hardware facilities such as processors, memories, transmitters, receivers, etc.; the present application is not limited thereto.

[0307] According to the above-mentioned embodiments, the operation of the AI / ML model or function according to the precoding layer and / or rank value can flexibly utilize the AI / ML model or function according to the layer and / or rank value, fully exert the advantages of the AI / ML model or function, and guarantee the performance of the AI / ML model or function and the performance of the communication system;

[0308] For example, when the historical time CSI information of some layer or some layers on the UE side and / or the network side cannot be obtained / are unavailable, based on the mechanism of the operation of the AI / ML model or function according to the precoding layer and / or rank value, these layers can switch to the SF-AI / ML or non-AI / ML method for CSI feedback, which can avoid the performance degradation of the AI / ML model or function and guarantee the performance of the communication system.

[0309] Embodiments of the fourth aspect

[0310] The embodiments of the present application provide a terminal device, which can execute the method applied to the terminal side in the embodiments of the first aspect, and in addition, the terminal device can include the apparatus described in the embodiments of the second aspect.

[0311] FIG. 11 is a schematic block diagram of the system structure of the terminal device according to an embodiment of the present application. As shown in FIG. 11, the terminal device 1100 can include a processor 1110 and a memory 1120; the memory 1120 is coupled to the processor 1110. It is worth noting that this figure is exemplary; other types of structures can also be used to supplement or replace this structure to achieve telecommunication functions or other functions.

[0312] In one embodiment, the processor 1110 is configured to: receive configuration information from a network device; and send first information to the network device according to the configuration information, the first information being information obtained after the terminal device operates an AI / ML model or function according to a precoding layer and / or rank value.

[0313] As shown in FIG. 11, the terminal device 1100 can further include a communication module 1130, an input unit 1140, a display 1150, and a power supply 1160. It is worth noting that the terminal device 1100 does not necessarily include all the components shown in FIG. 11; in addition, the terminal device 1100 can also include components not shown in FIG. 11, which can be referred to related technologies.

[0314] As shown in FIG. 11, the processor 1110, also known as a controller or operation control, can include a microprocessor or other processor device and / or a logic device, which receives input and controls the operation of each component of the terminal device 1100.

[0315] The memory 1120, for example, can be one or more of a buffer, a flash memory, a hard drive, a removable media, a volatile memory, a non-volatile memory, or other suitable device. Various data can be stored, and further programs for information processing can be stored. The processor 1110 can execute the programs stored in the memory 1120 to achieve information storage or processing, and the like. The functions of other components are similar to the existing, and will not be described here. The components of the terminal device 1100 can be implemented by dedicated hardware, firmware, software, or a combination thereof, without departing from the scope of the present application.

[0316] According to the embodiments of the present application, the AI / ML model or function is operated according to the precoding layer and / or rank value, which can flexibly utilize the AI / ML model or function according to the layer and / or rank value, fully utilize the advantages of the AI / ML model or function, and ensure the performance of the AI / ML model or function and the performance of the communication system;

[0317] For example, when the historical time CSI information of some layer or some layers on the UE side and / or the network side cannot be obtained / available, based on the mechanism of operating the AI / ML model or function according to the precoding layer and / or rank value, these layers can switch to the SF-AI / ML or non-AI / ML method for CSI feedback, which can avoid the performance degradation of the AI / ML model or function and ensure the performance of the communication system.

[0318] Embodiments of the fifth aspect

[0319] The network device provided in the embodiments of the present application can execute the method applied to the network side in the embodiments of the first aspect, and in addition, the network device can include the apparatus described in the embodiments of the third aspect.

[0320] FIG. 12 is a schematic block diagram of the system structure of the network device according to the embodiments of the present application. As shown in FIG. 12, the network device 1200 can include a processor 1210 and a memory 1220, and the memory 1220 is coupled to the processor 1210. The memory 1220 can store various data, and further store programs for information processing 1230, and execute the programs 1230 under the control of the processor 1210 to receive various information sent by the terminal device and send various information to the terminal device.

[0321] In one embodiment, the processor 1210 can be configured to send configuration information to the terminal device, and receive first information, which is obtained after the terminal device operates the AI / ML model or function according to the precoding layer and / or rank value.

[0322] In addition, as shown in FIG. 12, the network device 1200 can further include a transceiver 1540 and an antenna 1550, etc., wherein the functions of the above components are similar to those of the prior art, and will not be repeated here. It is worth noting that the network device 1200 does not necessarily include all the components shown in FIG. 12; in addition, the network device 1200 can also include components not shown in FIG. 12, which can be referred to the prior art.

[0323] According to the embodiments of the present application, the AI / ML model or function is operated according to the precoding layer and / or rank value, which can flexibly utilize the AI / ML model or function according to the layer and / or rank value, fully utilize the advantages of the AI / ML model or function, and guarantee the performance of the AI / ML model or function and the performance of the communication system;

[0324] For example, when the historical time CSI information of some layer or some layers on the UE side and / or the network side cannot be obtained / are unavailable, based on the mechanism of operating the AI / ML model or function according to the precoding layer and / or rank value, these layers can switch to the SF-AI / ML or non-AI / ML method for CSI feedback, which can avoid the performance degradation of the AI / ML model or function and guarantee the performance of the communication system.

[0325] Embodiments of the sixth aspect

[0326] The embodiments of the present application provide a communication system, which includes the terminal device according to the embodiments of the eighth aspect and / or the network device according to the embodiments of the ninth aspect. The specific content can be referred to the description in the embodiments of the eighth aspect and the ninth aspect.

[0327] For example, the structure of the communication system can refer to FIG. 1, as shown in FIG. 1, the communication system 100 includes a network device 101 and terminal devices 102 and 103, the terminal device 102 and / or the terminal device 103 can be the same as the terminal device described in the embodiments of the fourth aspect, and / or the network device 101 can be the same as the network device described in the embodiments of the fifth aspect, and the repeated content will not be repeated.

[0328] The above apparatus and method of the present application can be realized by hardware, or by a combination of hardware and software. The present application relates to a computer readable program, which, when executed by a logic component, can enable the logic component to realize the above-described apparatus or constituent components, or to realize the various methods or steps described above. The present application also relates to a storage medium for storing the above program, such as a hard disk, a magnetic disk, an optical disk, a DVD, a flash memory, etc.

[0329] The methods / apparatuses described in connection with the embodiments disclosed herein can be embodied directly in hardware, software executed by a processor, or combination thereof. For example, one or more of the functional blocks shown in FIG. 9 or FIG. 10 and / or one or more combinations of the functional blocks can correspond to individual software modules of a computer program flow, or to individual hardware modules. These software modules can correspond to the individual steps shown in FIG. 3 or FIG. 4, respectively. These hardware modules can be implemented, for example, by means of a field-programmable gate array (FPGA) that is programmed to solidify the software modules.

[0330] The software modules can be located in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a mobile disk, a CD-ROM, or any other form of storage medium known in the art. A storage medium can be coupled to the processor, such that the processor can read information from, and write information to, the storage medium; or the storage medium can be integral to the processor. The processor and the storage medium can be located in an ASIC. The software modules can be stored in a memory of the mobile terminal, or in a memory card that can be inserted into the mobile terminal. For example, if the device (e.g., the mobile terminal) uses a MEGA-SIM card or a large capacity flash memory device, the software modules can be stored in the MEGA-SIM card or the large capacity flash memory device.

[0331] One or more of the functional blocks described in connection with FIG. 9 or FIG. 10 and / or one or more combinations of the functional blocks can be implemented as a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any suitable combination thereof, for performing the functions described herein. One or more of the functional blocks described in connection with FIG. 9 or FIG. 10 and / or one or more combinations of the functional blocks can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in communication with a DSP core, or any other such configuration.

[0332] The present application has been described in connection with specific embodiments thereof, but those skilled in the art will understand that these descriptions are illustrative of the application and are not intended to limit the scope of the application. Various modifications of the application in addition to those described herein will become apparent to those skilled in the art from the teachings herein. Such modifications are intended to fall within the scope of the application.

[0333] According to the various embodiments disclosed in connection with the embodiments disclosed herein, the following clauses are also disclosed:

[0334] 1. A method for processing operation of an AI / ML model or function, applied to a multiple antenna (MIMO) communication system, the method being applied to a terminal side, the method comprising:

[0335] receiving configuration information from a network device;

[0336] sending first information to the network device according to the configuration information, the first information being information obtained after operation of the AI / ML model or function by the terminal device according to a precoding layer and / or rank value.

[0337] 2. A method for processing operation of an AI / ML model or function, applied to a multiple antenna (MIMO) communication system, the method being applied to a network side, the method comprising:

[0338] sending configuration information to a terminal device;

[0339] receiving first information, the first information being information obtained after operation of the AI / ML model or function by the terminal device according to a precoding layer and / or rank value.

[0340] 3. The method according to any one of appendices 1 or 2, wherein,

[0341] for TSF-AI / ML CSI compression feedback, the AI / ML model or function needs CSI information of one or more historical time instants,

[0342] when historical time instant CSI information of one or more layers of the terminal side and / or the network side cannot be obtained or is unavailable, the layer is switched to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, so that the TSF-AI / ML model or function obtains or recovers or resets the historical CSI information; when the historical CSI information is obtained or recovered or reset, the TSF-AI / ML CSI compression feedback is activated or used at the layer,

[0343] the switching is applied to initial running or reactivation of the TSF-AI / ML model or function, or is applied to a case where the historical CSI information is unavailable due to a rank value change or UCI loss or CSI discard.

[0344] 4. The method according to any one of appendices 1-3, wherein,

[0345] one or more layers need to be switched to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method within a time window,

[0346] The time window is defined according to the number of time instants required by the TSF-AI / ML model or function to acquire or recover or reset the historical CSI information, the definition being explicit or implicit.

[0347] 5. The method according to any one of appendices 1-4, wherein,

[0348] The time window is applied to the initial running or reactivation of the TSF-AI / ML model or function, or to the case where the historical CSI information is not available due to rank value change or UCI loss or CSI discard.

[0349] 6. The method according to any one of appendices 1-5, wherein,

[0350] The time window is used at the terminal side and / or at the network side,

[0351] When the time window is used at the terminal side and at the network side, the time windows at the two sides are the same or different.

[0352] 7. The method according to any one of appendices 1-6, wherein,

[0353] The size of the time window is the same or different for different layers and / or rank values, and / or,

[0354] The size of the time window depends on the terminal capability of the terminal device.

[0355] 8. The method according to any one of appendices 1-7, wherein,

[0356] Whether to support switching to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method depends on the terminal capability of the terminal device, and / or whether to support running TSF-AI / ML CSI compression feedback and non-TSF-AI / ML CSI compression feedback at the same time depends on the terminal capability of the terminal device; and / or,

[0357] Whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method within the time window is determined by the terminal device, and / or whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method at one or more layers is reported by the terminal device to the network device.

[0358] 9. The method according to any one of appendices 1-8, wherein,

[0359] For TSF-AI / ML CSI compression feedback, when the historical CSI information of one or more layers at the terminal device side and / or the network side cannot be obtained or is unavailable, one or more of the following methods is used to obtain or recover the historical CSI information:

[0360] The layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the output of the TSF-AI / ML encoder with respect to the layer is still sent to the network side through the air interface for updating the historical CSI information at the network side, wherein the network side obtains or recovers the historical CSI information by inputting the output of the TSF-AI / ML encoder into the TSF-AI / ML decoder;

[0361] The network side has a proxy or approximate TSF encoder, in which case the network side obtains or recovers the historical CSI information at one or more time instants through SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, wherein the network side reconstructs the feature vector and / or channel matrix of the layer through SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and inputs the reconstructed feature vector and / or channel matrix into the proxy or approximate TSF encoder, and then inputs the output of the proxy or approximate TSF encoder into the TSF decoder to obtain or recover the historical CSI information at the network side;

[0362] When the historical CSI information at the network side is reconstructed CSI, the layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on a non-AI / ML method, and the reconstructed CSI based on SF-AI / ML CSI compression feedback or CSI feedback based on a non-AI / ML method is used to obtain or recover the historical CSI information.

[0363] 10. The method of any one of appendices 1-9, wherein,

[0364] The method of any one of appendices 1-9, wherein,

[0365] The terminal device reports to the network side the recommended method, and the network side finally determines the method for obtaining or recovering the historical CSI information.

Claims

1. An operation processing device for an AI / ML model or function, applied to a multi-antenna (MIMO) communication system, the device being applied to a terminal side, the device comprising: a first receiving unit, configured to receive configuration information from a network device; A first sending unit sends first information to the network device according to the configuration information, where the first information is information obtained after the terminal device operates the AI / ML model or function according to the precoded layer and / or rank value.

2. An operation processing device for an AI / ML model or function, applied to a multi-antenna (MIMO) communication system, the device being applied to a network side, the device comprising: a second sending unit, configured to send configuration information to a terminal device; A second receiving unit receives first information, where the first information is information obtained after the terminal device operates the AI / ML model or function according to the precoded layer and / or rank value.

3. The device according to claim 1 or 2, wherein: The operation of the AI / ML model or function includes at least one or more of the following operations: Choice of AI / ML models or functions; Activation of AI / ML models or functions; Deactivation of AI / ML models or functions; Switching of AI / ML models or functions; Updates to AI / ML models or features; Rollback of AI / ML models or features.

4. The device according to claim 1 or 2, wherein: The operation of AI / ML models or functions according to layers and / or rank values ​​is based on predefined rules, which include at least: on which layers the operation of one or more AI / ML models or functions is performed, and / or on which layers the operation of one or more AI / ML models or functions is performed when the rank value changes.

5. The device according to claim 1 or 2, wherein: The operation of AI / ML models or functions according to layers and / or rank values ​​is based on: one or more predefined rules configured on the network side, which at least include: on which layers to perform one or more AI / ML model or function operations, and / or on which layers to perform one or more AI / ML models or functions when the rank value changes; The operation of the type or function; the rule configured on the network side is included or not included in the configuration information, The terminal device selects one of the rules, performs the operation and notifies the network side of the corresponding operation, wherein the notification includes explicit or implicit notification.

6. The device according to claim 1 or 2, wherein: The operation of the AI / ML model or function according to the layer and / or rank value is based on: the network side instructs the terminal device that it can perform one or several AI / ML model or function operations on a certain layer or certain layers; the network side's indication is included or not included in the configuration information.

7. The device according to claim 6, wherein The configuration information is determined by the network device based on the operation of the terminal device requesting the AI / ML model or function.

8. The device according to claim 1 or 2, wherein: The operation of AI / ML models or functions according to layers and / or rank values ​​is based on: the terminal device decides to perform one or more AI / ML model or function operations on a certain layer or layers, and notifies the network device of the said operation.

9. The device according to claim 1 or 2, wherein: The device further comprises: A processing unit that operates on an AI / ML model or function according to a layer and / or rank value, wherein: The processing unit performs AI / ML model or function operations on some or all layers and / or rank values, and / or, The processing unit performs the same or different AI / ML models or functions on different layers and / or rank values.

10. The device according to claim 1 or 2, wherein: Operations on AI / ML models or functions according to layer and / or rank values ​​are applied to two-sided AI / ML models or functions, or alternatively, to one-sided AI / ML models or functions.

11. The device according to claim 1 or 2, wherein: The operations on AI / ML models or functions based on layer and / or rank values ​​apply to at least one or more of the following: Rank-specific AI / ML models or functions; Rank common AI / ML models or functions; Layer-specific AI / ML models or functions; Layer common AI / ML models or functions, The layer-specific AI / ML model or function includes: layer-specific and rank-common AI / ML models or functions, and / or layer-specific and rank-specific AI / ML models or functions; The layer-common AI / ML models or functions include: layer-common and rank-common AI / ML models or functions, and / or layer-common and rank-specific AI / ML models or functions.

12. The device according to claim 1 or 2, wherein: The device further comprises: A monitoring unit that monitors the operation of an AI / ML model or function according to the layer and / or rank value, The monitoring of the AI / ML model or function includes: monitoring the performance of the model or function of at least one layer and / or rank; and / or monitoring whether historical CSI information of at least one layer and / or rank is available.

13. The device according to claim 7, wherein The processing unit determines a CSI compression feedback scheme according to layers for CSI compression feedback based on the bilateral AI / ML model, wherein: For CSI compression feedback based on bilateral AI / ML models, At the same time, all layers use the same CSI compression feedback scheme, or different layers use different CSI compression feedback schemes. At different times, the same CSI compression feedback scheme may be used on the same layer, or different CSI compression feedback schemes may be used on the same layer.

14. The device according to claim 13, wherein The CSI compression feedback scheme includes at least one or more of the following: AI / ML-based time-space frequency domain (TSF-AI / ML) CSI compression feedback; AI / ML-based joint CSI prediction and CSI compression feedback; AI / ML-based spatial-frequency domain (SF-AI / ML) CSI compression feedback; CSI compression feedback based on non-AI / ML methods.

15. The device according to claim 14, wherein For TSF-AI / ML CSI compression feedback, the AI / ML model or function requires CSI information at one or more historical moments. When the historical CSI information of one or more layers on the terminal side and / or the network side cannot be obtained or is unavailable, The layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods, so that the TSF-AI / ML model or function obtains, restores, or resets historical CSI information; when the historical CSI information is obtained, restored, or reset, the TSF-AI / ML CSI compression feedback is activated or used at the layer.

16. The device according to claim 15, wherein Within a time window, one or more layers need to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods. The time window is predefined in the standard, or is based on RRC configuration and / or updated and / or indicated through MAC-CE or DCI.

17. The device according to claim 16, wherein Whether to switch to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods within the time window is configured by predefined configuration, or based on RRC signaling configuration and / or based on MAC CE or DCI update and / or indication.

18. The device according to claim 14, wherein TSF-AI / ML CSI compression feedback is applied only to the first layer, and SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods is applied to other layers, or, SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods is applied to the first layer and other layers.

19. The device according to claim 1 or 2, wherein: For TSF-AI / ML CSI compression feedback, when historical CSI information of one or more layers on the terminal device side and / or the network side cannot be obtained or is unavailable, a combination of one or more of the following methods is used to obtain or restore historical CSI information: The layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods, while the output of the TSF-AI / ML encoder for the layer is still sent to the network side over the air interface for updating historical CSI information on the network side; A proxy or approximate TSF encoder exists on the network side. In this case, the network side obtains or recovers historical CSI information at one or more time points through SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML methods. When the historical CSI information on the network side is reconstructed CSI, the layer switches to SF-AI / ML CSI compression feedback or CSI compression feedback based on non-AI / ML method, and the CSI compression feedback based on SF-AI / ML method or non-AI / ML method is switched to SF-AI / ML CSI compression feedback or non-AI / ML method. The reconstructed CSI of the CSI feedback of the AI / ML method is used to obtain or recover historical CSI information.

20. A communication system comprising a terminal device and a network device, The terminal device includes the apparatus according to claim 1, and / or The network device includes the apparatus according to claim 2.

Images