A method and apparatus used in a node for wireless communication

Abstract

This application discloses a method and apparatus for use in a node for wireless communication. The first node receives a first dataset; and sends a first feedback and a second feedback. The first dataset includes multiple CSI groups, each of which includes a target CSI, a first CSI feedback, and a second CSI feedback. The first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback. The data structure of the second CSI feedback includes more bits than that of the first CSI feedback. The generation of the second feedback depends on a first encoder; the training of the first encoder depends on the first dataset. This method ensures consistency between AI / ML training and inference, improving the performance of AI / ML-based CSI reporting.

Description

Technical Field

[0001] This application relates to transmission methods and apparatus in wireless communication systems, and more particularly to methods and apparatus for CSI in wireless communication systems. Background Technology

[0002] In traditional wireless communication, the UE (User Equipment) obtains various auxiliary information by measuring downlink signals and / or channels, such as CSI (Channel State Information), beam management-related auxiliary information, and positioning-related auxiliary information. CSI includes, but is not limited to, one or more of CRI (Channel State Information - RS Resource Indicator), RI (Rank Indicator), PMI (Precoding Matrix Indicator), CQI (Channel Quality Indicator), or L1-RSRP (Layer 1 Reference Signal Received Power). The UE can use this information to select appropriate transmission parameters or report this information. Network devices select appropriate transmission parameters for the UE based on its reports, such as the cell to be camped, MCS (Modulation and Coding Scheme), TPMI (Transmitted Precoding Matrix Indicator), and TCI (Transmission Configuration Indication). Furthermore, UE reports can be used to optimize network parameters, such as improving cell coverage and switching base stations on / off based on the UE's location.

[0003] In NRR (release) 18, research on AI (Artificial Intelligence) / ML (Machine Learning) technologies was initiated to explore their impact on system performance and design. AI / ML-based CSI reporting enhancement is a key topic of discussion in wireless communication systems. For application scenarios requiring AI / ML operations at both the transmitting and receiving ends, ensuring model matching between the two sides is a crucial issue. At the 118bis meeting of 3GPP (3rd Generation Partner Project) RAN (Radio Access Network) WG (Working Group) 1, a method was adopted to exchange datasets, including target CSI and CSI feedback, between the network and UE sides to assist in model matching between the network and UE sides. AI / ML technologies may also play a significant role in future 6G communications.

[0004] Compared to traditional processing methods, AI / ML is characterized by its training-based and deployment-required nature. According to the 3GPP standard TS38.300, AI / ML models and algorithms extend beyond the scope of 3GPP (3rd Generation Partnership Project). Summary of the Invention

[0005] The applicant's research revealed that existing measurement, computation, and reporting mechanisms may be inadequate to meet the demands of AI / ML when AI / ML functionality is introduced. For example, AI / ML models are training-based, and their operation mimics the behavior of the human brain, differing significantly from traditional CSI calculations. Existing measurement, computation, and reporting mechanisms do not consider these issues. To address these problems, this application discloses a solution. It should be noted that while this application is motivated by AI / ML applications and many embodiments are specifically designed for AI / ML, it is also applicable to other solutions, such as traditional measurement, computation, and reporting schemes. Although the specification of this application includes descriptions of some AI / ML models and algorithms, those skilled in the art will understand that these descriptions are not essential or irreplaceable for solutions related to wireless cellular communication. Furthermore, adopting a unified solution across different scenarios (including but not limited to AI / ML-based solutions and traditional measurement, computation, and reporting schemes) helps reduce signaling overhead / complexity, hardware complexity, and cost. Where there is no conflict, the embodiments and features in the first node of this application can be applied to the second node, and vice versa. Where there is no conflict, the embodiments and features in the embodiments of this application can be combined with each other arbitrarily.

[0006] When necessary, the interpretation of terms in this application shall be based on the definitions of the 3GPP specification protocol TS38 series, or the definitions of the 3GPP specification protocol TS28 series.

[0007] This application discloses a method used in a first node for wireless communication, characterized by comprising:

[0008] Receive a first dataset, the first dataset comprising multiple CSI groups, each of the multiple CSI groups comprising a target CSI, a first CSI feedback and a second CSI feedback;

[0009] Send the first and second feedback;

[0010] Wherein, the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback; the generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset.

[0011] The problem this application aims to solve includes how to further optimize CSI reporting; in the above method, the first dataset includes a first CSI feedback and a second CSI feedback, the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback, thus solving this problem.

[0012] The benefits of the above methods include optimizing AI / ML-based CSI reporting, further improving the reliability of CSI reporting and / or reducing overhead, thereby optimizing the overall system performance and efficiency.

[0013] The benefits of the above methods include ensuring consistency between AI / ML model training and inference, and ensuring the performance of CSI reporting.

[0014] According to one aspect of this application, the data structure of the input of the first encoder includes the data structure of the target CSI, and the data structure of the output of the first encoder includes the data structure of the second CSI feedback.

[0015] The benefits of the above methods include ensuring consistency between training and inference of AI / ML models and improving the performance of AI / ML-based CSI reporting.

[0016] According to one aspect of this application, the first encoder satisfies the following consistency requirement: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group satisfies the performance requirement.

[0017] The benefits of the above methods include ensuring consistency between training and inference of AI / ML models and improving the performance of AI / ML-based CSI reporting.

[0018] Generally, the target receiver of the second feedback does not need to know the model of the first encoder, but the target receiver of the second feedback and the first node need to calibrate the model used. The consistency requirement provides a way for the calibration of the models of both parties, and provides conditions for accurate CSI recovery or reconstruction.

[0019] According to one aspect of this application, the first dataset comprises a plurality of data subsets, any one of the plurality of data subsets includes a portion of the plurality of CSI groups, and all CSI groups in any one of the plurality of data subsets include the same first CSI feedback.

[0020] The essence of the above method includes that different data subsets include channel information obtained under different scenarios (including but not limited to different channel environments, different moving speeds, different frequency domain ranges, different weather conditions, and different three-dimensional regions, etc.); the above method provides AI / ML model training with data for different scenarios, improves the applicability of the model, and thus improves the robustness and performance of AI / ML-based CSI reporting.

[0021] The benefits of the above methods include improving the lifecycle and efficiency of AI / ML models.

[0022] According to one aspect of this application, the second feedback is conditional upon the first feedback, the first feedback being associated with a first data subset, the first data subset being a subset of the first dataset.

[0023] The advantages of the above method include that the first feedback indicates the correlation between the inference data and the training dataset, enabling the AI / ML model to use this correlation to optimize the generation of CSI, thereby further improving the efficiency and performance of CSI reporting.

[0024] According to one aspect of this application, the first data subset includes a portion of the plurality of CSI groups, and the first feedback indicates the first CSI feedback included in the CSI group of the first data subset.

[0025] The advantages of the above method include that the first feedback indicates the scenario in which the channel state carried by the second feedback is located (the scenario includes, but is not limited to, channel environment, moving speed, time period, frequency domain range, weather conditions, and three-dimensional region), and with the assistance of the first feedback, the load size of the second feedback can be further reduced without sacrificing performance.

[0026] According to one aspect of this application, it is characterized by comprising:

[0027] Receive the first information block;

[0028] The data structure of the first encoder output includes the data structure of the second CSI feedback and the data structure of the first CSI feedback. The first information block activates or deactivates at least one of the data structure of the second CSI feedback and the data structure of the first CSI feedback output by the first encoder.

[0029] The essence of the above method includes supporting performance monitoring of the data structure of the second CSI feedback and the data structure of the first CSI feedback respectively, ensuring the performance of AI / ML-based CSI reporting and improving the utilization rate of AI / ML models.

[0030] According to one aspect of this application, when the first information block deactivates the data structure of the first CSI feedback output by the first encoder, the first information block must simultaneously deactivate the data structure of the second CSI feedback output by the first encoder.

[0031] The advantages of the above method include ensuring the reliability of the data structure of the second CSI feedback output from the first encoder.

[0032] The benefits of the above method include optimizing AI / ML-based CSI reporting, thereby improving the overall system performance and efficiency.

[0033] This application discloses a method used in a second node for wireless communication, characterized by comprising:

[0034] Receive the first and second feedback;

[0035] The generation of the second feedback depends on the first encoder, and the training of the first encoder depends on the first dataset. The first dataset includes multiple CSI groups, and each of the multiple CSI groups includes a target CSI, a first CSI feedback, and a second CSI feedback. The first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback. The data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback.

[0036] According to one aspect of this application, the data structure of the input of the first encoder includes the data structure of the target CSI, and the data structure of the output of the first encoder includes the data structure of the second CSI feedback.

[0037] According to one aspect of this application, the first encoder satisfies the following consistency requirement: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group satisfies the performance requirement.

[0038] According to one aspect of this application, the first dataset comprises a plurality of data subsets, any one of the plurality of data subsets includes a portion of the plurality of CSI groups, and all CSI groups in any one of the plurality of data subsets include the same first CSI feedback.

[0039] According to one aspect of this application, the second feedback is conditional upon the first feedback, the first feedback being associated with a first data subset, the first data subset being a subset of the first dataset.

[0040] According to one aspect of this application, the first data subset includes a portion of the plurality of CSI groups, and the first feedback indicates the first CSI feedback included in the CSI group of the first data subset.

[0041] According to one aspect of this application, it is characterized by comprising:

[0042] Send the first information block;

[0043] The data structure of the first encoder output includes the data structure of the second CSI feedback and the data structure of the first CSI feedback. The first information block activates or deactivates at least one of the data structure of the second CSI feedback and the data structure of the first CSI feedback output by the first encoder.

[0044] According to one aspect of this application, when the first information block deactivates the data structure of the first CSI feedback output by the first encoder, the first information block must simultaneously deactivate the data structure of the second CSI feedback output by the first encoder.

[0045] This application discloses a first node used for wireless communication, characterized in that it includes:

[0046] A first receiver receives a first dataset, which includes multiple CSI groups. Each CSI group includes a target CSI, a first CSI feedback, and a second CSI feedback.

[0047] The first transmitter sends the first and second feedback signals.

[0048] Wherein, the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback; the generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset.

[0049] This application discloses a second node used for wireless communication, characterized by comprising:

[0050] The first processor receives the first and second feedback;

[0051] The generation of the second feedback depends on the first encoder, and the training of the first encoder depends on the first dataset. The first dataset includes multiple CSI groups, and each of the multiple CSI groups includes a target CSI, a first CSI feedback, and a second CSI feedback. The first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback. The data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback.

[0052] As an example, compared with conventional solutions, this application has the following advantages:

[0053] This further improved the performance of CSI reporting and / or reduced overhead;

[0054] This ensures consistency between AI / ML training and inference, and improves the performance of AI / ML-based CSI reporting.

[0055] Optimize AI / ML models to adapt to different scenarios, thereby improving the robustness and performance of AI / ML models;

[0056] This helps ensure consistency between the models of the sender and receiver, and ensures the performance of CSI reporting based on AI / ML. Attached Figure Description

[0057] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0058] Figure 1 A flowchart illustrating a first dataset, a first feedback, and a second feedback according to an embodiment of this application is shown;

[0059] Figure 2 A schematic diagram of a network architecture according to an embodiment of this application is shown;

[0060] Figure 3 A schematic diagram of an embodiment of a wireless protocol architecture for the user plane and control plane according to an embodiment of this application is shown;

[0061] Figure 4 A schematic diagram of a first communication device and a second communication device according to an embodiment of this application is shown;

[0062] Figure 5 The transmission between a first node and a second node according to one embodiment of this application is illustrated;

[0063] Figure 6 A schematic diagram illustrating the deployment of a first encoder according to an embodiment of this application is shown;

[0064] Figure 7 A schematic diagram of the inputs and outputs of a first encoder according to an embodiment of this application is shown;

[0065] Figure 8 A schematic diagram illustrating a first encoder satisfying a conformance requirement according to an embodiment of this application is shown;

[0066] Figure 9 A schematic diagram of a first dataset according to an embodiment of this application is shown;

[0067] Figure 10 A schematic diagram illustrating a first dataset comprising multiple data subsets according to an embodiment of this application is shown;

[0068] Figure 11 A schematic diagram illustrating a first dataset according to an embodiment of this application includes multiple data subsets and multiple first CSI feedbacks;

[0069] Figure 12A schematic diagram showing a second feedback conditional on a first feedback according to an embodiment of this application is illustrated;

[0070] Figure 13 A schematic diagram is shown illustrating a first feedback indicating a first CSI feedback included in a CSI group within a first subset of data, according to an embodiment of this application;

[0071] Figure 14 A schematic diagram of a first information block according to an embodiment of this application is shown;

[0072] Figure 15 A schematic diagram of a first information block according to an embodiment of this application is shown;

[0073] Figure 16 A schematic diagram of a first feedback, a second feedback, and first channel information according to an embodiment of this application is shown;

[0074] Figure 17 A schematic diagram of an artificial intelligence or machine learning-based processing system according to an embodiment of this application is shown;

[0075] Figure 18 A schematic diagram based on artificial intelligence or machine learning according to an embodiment of this application is shown;

[0076] Figure 19 A schematic diagram illustrating the deployment of AI functionality according to an embodiment of this application is shown;

[0077] Figure 20 A schematic diagram illustrating the deployment of AI functionality according to an embodiment of this application is shown;

[0078] Figure 21 A schematic diagram illustrating the deployment of AI functionality according to an embodiment of this application is shown;

[0079] Figure 22 A schematic diagram illustrating the deployment of AI functionality according to an embodiment of this application is shown;

[0080] Figure 23 A structural block diagram of a processing apparatus for a first node according to an embodiment of this application is shown;

[0081] Figure 24 A structural block diagram of a processing apparatus for a second node according to an embodiment of this application is shown. Detailed Implementation

[0082] The technical solutions of this application will be further described in detail below with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other. Considering performance, flexibility, complexity, overhead, and compatibility, those skilled in the art are motivated to flexibly combine the embodiments in different drawings without conflict, such as, but not limited to, those in the accompanying drawings. Figure 1 Examples and appendices Figure 5 -Appendix Figure 24 The embodiments in the appendix Figure 5 Examples and appendices Figure 6 -Appendix Figure 24 Examples, etc.

[0083] Example 1

[0084] Example 1 illustrates a flowchart of a first dataset, a first feedback, and a second feedback according to an embodiment of this application, as shown in the attached diagram. Figure 1 As shown. In the appendix Figure 1 In the 100 shown, each box represents a step. In particular, the order of the steps in the boxes does not represent a specific temporal relationship between the steps.

[0085] In Example 1, the first node receives a first dataset in step 101 and sends a first feedback and a second feedback in step 102. The first dataset includes multiple CSI groups, each of which includes a target CSI, a first CSI feedback, and a second CSI feedback. The first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback. The data structure of the second CSI feedback includes more bits than that of the first CSI feedback. The generation of the second feedback depends on a first encoder, and the training of the first encoder depends on the first dataset.

[0086] As an example, the first dataset includes a training dataset.

[0087] As an example, the first dataset is used for training an AI model or an ML model.

[0088] As an example, the first dataset relies on measurements on at least one RS (Reference Signal) resource.

[0089] As an example, the first dataset relies on channel measurements on at least one RS resource.

[0090] As an example, the first dataset includes channel information that depends on measurements on at least one RS resource.

[0091] As an example, the first dataset includes channel information that depends on channel measurements on at least one RS resource.

[0092] As an example, the at least one RS resource includes a downlink RS resource.

[0093] As an example, the at least one RS resource includes at least one of CSI-RS (Channel State Information Reference Signal) resources and SS / PBCH (Synchronization Signal / Physical Broadcast Channel) block resources.

[0094] As an example, the at least one RS resource includes one or more of DMRS (Demodulation Reference Signal), PRS (Positioning Reference Signal) resources and PTRS (Phase-Tracking Reference Signal).

[0095] As one example, the channel information includes a precoding matrix.

[0096] As one example, the channel information includes a channel matrix.

[0097] As one example, the channel information includes the original channel matrix.

[0098] As one example, the channel information includes the channel impulse response.

[0099] As one example, the channel information includes a feature vector.

[0100] As one example, the channel information includes feature vectors and feature values.

[0101] As one embodiment, the channel information includes the channel matrix or the feature vector of the original channel matrix.

[0102] As one embodiment, the channel information includes the eigenvectors and eigenvalues ​​of the channel matrix or the original channel matrix.

[0103] As one example, the matrix includes vectors.

[0104] As an example, the target CSI includes a precoding matrix.

[0105] As an example, the target CSI is a precoding matrix.

[0106] As an example, the target CSI includes a channel matrix.

[0107] As an example, the target CSI includes the original channel matrix.

[0108] As an example, the target CSI includes a feature vector.

[0109] As an example, the target CSI includes a feature vector and feature values.

[0110] As an example, the target CSI depends on measurements on at least one RS resource.

[0111] As an example, the target CSI is obtained based on measurements on at least one RS resource.

[0112] As one embodiment, the precoding matrix includes precoding vectors.

[0113] As one embodiment, the precoding matrix is ​​in the spatial frequency domain or the angle-delay domain.

[0114] As an example, the precoding matrix includes at least one of the eigenvectors in the spatial frequency domain and the eigenvectors in the angle-delay domain.

[0115] As an example, the precoding matrix is ​​represented in the form of a Type II codebook.

[0116] As an example, the definition of the Type II codebook is found in 3GPP TS38.214.

[0117] As one embodiment, the Type II codebook includes one or more of the following: a Type II port selection codebook, an enhanced Type II codebook, an enhanced Type II port selection codebook, a further enhanced Type II port selection codebook, an enhanced Type II codebook for CJT, a further enhanced Type II port selection codebook for CJT, an enhanced Type II codebook for predicting PMI, and a further enhanced Type II port selection codebook for predicting PMI.

[0118] As an example, the second CSI feedback is based on inference.

[0119] As an example, the second CSI feedback depends on the output of an inference.

[0120] As an example, the second CSI feedback includes an inference output.

[0121] As one example, the second CSI feedback includes compressed CSI.

[0122] As one example, the second CSI feedback is compressed CSI.

[0123] As an example, the second CSI feedback is a compressed CSI of the target CSI included in the same CSI group.

[0124] As an example, the second CSI feedback is the output obtained by taking the target CSI included in the same CSI group as the input of a reasoning process.

[0125] As an example, the first CSI feedback includes at least one string.

[0126] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of Umi, Uma, Sma, Rma, UAV, LOS and NLOS.

[0127] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of Urban, Suburban, Rural, Micro, Macro, Hotspot, Indoor, LOS, and NLOS.

[0128] As a sub-implementation of the above embodiments, the candidates for the at least one string include FR1 and FR2.

[0129] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of speed, sparse, dense, ray, path, road and metropolitan.

[0130] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of sunny, cloudy, rainy, windy, shower, snowy, and misty.

[0131] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of morning, noon, afternoon, dust, evening, night, morningpeak, and evening peak.

[0132] As an example, the first CSI feedback includes at least one number.

[0133] As an example, the first CSI feedback is used to indicate the channel environment.

[0134] As an example, the first CSI feedback is used to indicate the type or specificity of the channel environment.

[0135] As an example, the first CSI feedback is used to indicate the channel environment targeted by the target CSI in the same CSI group.

[0136] As one example, the channel environment includes, but is not limited to, Urban, Suburban, Rural, Micro, Macro, Hotspot, Indoor, LOS, and NLOS.

[0137] As an example, the first CSI feedback is used to indicate a time period.

[0138] As an example, the first CSI feedback is used to indicate the time period targeted by the target CSI in the same CSI group.

[0139] As one example, the time periods include, but are not limited to, morning, noon, afternoon, dust, evening, night, morning peak, and evening peak.

[0140] As an example, the first CSI feedback is used to indicate the movement speed or the range of movement speed.

[0141] As an example, the first CSI feedback is used to indicate the speed of movement targeted by the target CSI in the same CSI group.

[0142] As an example, the first CSI feedback is used to indicate the frequency domain range.

[0143] As an example, the first CSI feedback is used to indicate the frequency domain range targeted by the target CSI in the same CSI group.

[0144] As an example, the first CSI feedback is used to indicate weather conditions.

[0145] As an example, the first CSI feedback is used to indicate the weather conditions targeted by the target CSI in the same CSI group.

[0146] As an example, the weather conditions include, but are not limited to, sunny, cloudy, rainy, windy, shower, snowy, and misty.

[0147] As an example, the first CSI feedback is used to indicate the three-dimensional region.

[0148] As an example, the first CSI feedback is used to indicate the three-dimensional region targeted by the target CSI in the same CSI group.

[0149] As an example, the three-dimensional region includes, but is not limited to, the distribution, density, height and material of buildings, the distribution, density, type and height of vegetation, the distribution and area of ​​water bodies, the distribution and density of pedestrian traffic, indoor and outdoor areas, etc.

[0150] As an example, the first CSI feedback indicates the scenario targeted by the target CSI in the same CSI group (the scenario includes, but is not limited to, channel environment, moving speed, time period, frequency domain range, weather conditions, and three-dimensional region); the above method provides more effective training data for AI / ML model training that is more in line with the working principle of AI / ML, thus optimizing AI / ML model training.

[0151] As an example, a target CSI is obtained under the channel environment, mobile speed, frequency domain range, time period, weather conditions and / or three-dimensional area to which the target CSI is targeted.

[0152] As an example, the first CSI feedback depends on the large-scale characteristics of the target CSI in the same CSI group.

[0153] As an example, the first CSI feedback is used to indicate the large-scale characteristics of the target CSI within the same CSI group.

[0154] As an example, at least two of the plurality of CSI groups include different first CSI feedbacks.

[0155] As an example, at least two of the plurality of CSI groups include the same first CSI feedback.

[0156] As an example, the first feedback and the second feedback are for the same CSI reporting configuration.

[0157] As an example, the same CSI reporting configuration is used to configure both the first feedback and the second feedback.

[0158] As an example, the same CSI reporting configuration is used to configure one or more of the following for the first feedback and the second feedback: RS resources, reporting amount, target frequency domain resources, time domain behavior, period, and time slot offset.

[0159] As one example, the time-domain behavior includes periodic, semi-persistent, and aperiodic.

[0160] As an example, the same CSI reporting configuration is carried by an RRC IE.

[0161] As an example, the same CSI reporting configuration is carried by an IE whose name includes CSI-ReportConfig.

[0162] As an example, the same CSI reporting configuration is carried by a single CSI-ReportConfig IE.

[0163] As an example, the first feedback is based on artificial intelligence or machine learning.

[0164] As an example, the second feedback is based on artificial intelligence or machine learning.

[0165] As an example, both the first feedback and the second feedback are based on artificial intelligence or machine learning.

[0166] As one example, the second feedback includes UCI (Uplink Control Information).

[0167] As one example, the second feedback includes CSI (Channel State Information).

[0168] As one example, the second feedback carries channel information.

[0169] As one example, the second feedback includes a PMI (Precoding Matrix Indicator).

[0170] As an example, the second feedback is used to determine at least one channel matrix.

[0171] As an example, the second feedback is used to determine at least one original channel matrix.

[0172] In a preferred embodiment, the second feedback is used to determine at least one precoding matrix.

[0173] In a preferred embodiment, the first feedback and the second feedback are used together to determine at least one precoding matrix.

[0174] In a preferred embodiment, the second feedback includes compressed CSI.

[0175] As an example, the compressed CSI is based on a non-codebook.

[0176] As an example, the compressed CSI is not a CSI defined in 3GPP Rel-18 or earlier versions.

[0177] As an example, the channel information recovered or reconstructed by the target receiver of the compressed CSI is unknown to the first node.

[0178] As an example, the compressed CSI is based on artificial intelligence or machine learning.

[0179] As an example, the compressed CSI is based on neural networks or CNNs (Conventional Neural Networks).

[0180] As an example, the second feedback includes one or more of CQI (Channel Quality Indicator), CRI (CSI-RS Resource Indicator), RI (Rank Indicator), SSBRI (SS / PBCH Block Resource Indicator), RSRP (Reference Signal received power), and SINR (Signal-to-Interference and Noise Ratio).

[0181] As one embodiment, the second feedback includes compressed CSI, and also includes one or more of CQI, CRI, RI, SSBRI, RSRP, and SINR.

[0182] As one example, the second feedback includes a channel matrix.

[0183] As one example, the second feedback includes a precoding matrix.

[0184] As one example, the second feedback includes precoded information.

[0185] As an example, the first feedback includes UCI.

[0186] As one example, the first feedback includes CSI.

[0187] As an example, the first feedback includes at least one string.

[0188] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of Umi, Uma, Sma, Rma, UAV, LOS and NLOS.

[0189] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of Urban, Suburban, Rural, Micro, Macro, Hotspot, Indoor, LOS, and NLOS.

[0190] As a sub-implementation of the above embodiments, the candidates for the at least one string include FR1 and FR2.

[0191] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of speed, sparse, dense, ray, path, road and metropolitan.

[0192] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of sunny, cloudy, rainy, windy, shower, snowy, and misty.

[0193] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of morning, noon, afternoon, dust, evening, night, morningpeak, and evening peak.

[0194] As an example, the first feedback includes at least one number.

[0195] As an example, the first feedback is used to indicate the channel environment.

[0196] As one example, the first feedback is used to indicate the type or specificity of the channel environment.

[0197] As one embodiment, the first feedback is used to indicate the channel environment to which the second feedback is targeted.

[0198] As one example, the channel environment includes, but is not limited to, Urban, Suburban, Rural, Micro, Macro, Hotspot, Indoor, LOS, and NLOS.

[0199] As an example, the first feedback is used to indicate a time period.

[0200] As one example, the first feedback is used to indicate the time period to which the second feedback is directed.

[0201] As one example, the time periods include, but are not limited to, morning, noon, afternoon, dust, evening, night, morning peak, and evening peak.

[0202] As one embodiment, the first feedback is used to indicate the movement speed or the range of movement speed.

[0203] As one embodiment, the first feedback is used to indicate the movement speed or range of movement speed targeted by the second feedback.

[0204] As an example, the first feedback is used to indicate the frequency domain range.

[0205] As one embodiment, the first feedback is used to indicate the frequency domain range targeted by the second feedback.

[0206] As an example, the first feedback is used to indicate weather conditions.

[0207] As an example, the first feedback is used to indicate the weather conditions to which the second feedback is directed.

[0208] As an example, the weather conditions include, but are not limited to, sunny, cloudy, rainy, windy, shower, snowy, and misty.

[0209] As one example, the first feedback is used to indicate a three-dimensional region.

[0210] As one embodiment, the first feedback is used to indicate the three-dimensional region targeted by the second feedback.

[0211] As an example, the three-dimensional region includes, but is not limited to, the distribution, density, height and material of buildings, the distribution, density, type and height of vegetation, the distribution and area of ​​water bodies, the distribution and density of pedestrian traffic, indoor and outdoor areas, etc.

[0212] As an example, the first feedback indicates the scenario targeted by the second feedback (the scenario includes, but is not limited to, channel environment, mobile speed, time period, frequency domain range, weather conditions, and three-dimensional area). With the assistance of the first feedback, the load size of the second feedback can be further reduced without sacrificing performance.

[0213] As one embodiment, the channel information carried by the second feedback is obtained under the channel environment, moving speed, frequency domain range, time period, weather conditions and / or three-dimensional area targeted by the second feedback.

[0214] As one embodiment, the second feedback relies on measurements of at least one RS resource obtained under the channel environment, mobile speed, frequency domain range, time period, weather conditions, and / or three-dimensional region to which the second feedback is targeted.

[0215] As an example, the first feedback depends on the large-scale characteristics of the channel information carried by the second feedback.

[0216] As an example, the large-scale characteristics include one or more of delay spread, Doppler spread, Doppler shift, average delay, or spatial reception parameters.

[0217] As one embodiment, the large-scale characteristics include a spatial domain transmit filter and a spatial domain receive filter.

[0218] As one embodiment, the large-scale characteristic includes a transmit beam or a set of transmit beams.

[0219] As an example, the first feedback is used to indicate the large-scale characteristics of the channel information carried by the second feedback.

[0220] As an example, the first encoder is based on inference.

[0221] As one embodiment, the first encoder includes inference.

[0222] As an example, the reasoning refers to AI reasoning or ML reasoning.

[0223] As an example, the first encoder is based on artificial intelligence or machine learning.

[0224] As an example, the first encoder is based on a neural network or CNN (Conventional Neural Networks).

[0225] As an example, the first encoder is based on a non-codebook.

[0226] As an example, the output of the first encoder does not belong to the CSI defined in 3GPP Rel-18 or earlier versions.

[0227] As an example, the channel information recovered or reconstructed by the target receiver based on the output of the first encoder is unknown to the first node.

[0228] As an example, the first encoder needs to be trained.

[0229] As an example, the model of the first encoder needs to be trained.

[0230] As an example, the training of the first encoder refers to the training of the model of the first encoder.

[0231] In a preferred embodiment, the training dataset of the first encoder includes the first dataset.

[0232] The benefits of the above methods include ensuring that AI / ML model training and inference are matched, thereby improving the performance of AI / ML.

[0233] The advantages of the above method include that the first feedback indicates the correlation between inference data and training data, enabling AI / ML models to use this correlation to optimize CSI generation, thereby further improving the efficiency and performance of CSI reporting.

[0234] As an example, the first dataset was used for the training of the first encoder.

[0235] Generally speaking, the target receiver of the second feedback does not need to know the model of the first encoder, but the target receiver of the second feedback and the first node need to have some consensus on the first encoder; through the first dataset, the target receiver of the second feedback and the first node reach a consensus on the training dataset of the first encoder, which is beneficial to the model calibration of both parties.

[0236] As an example, the training of the first encoder is performed by the first node.

[0237] As an example, the training of the first encoder is performed by the sender of the first dataset.

[0238] As an example, the training of the first encoder is performed by the target receiver of the second feedback.

[0239] As an example, the training of the first encoder is performed by a core network device.

[0240] As an example, the training of the first encoder is performed by an OTT (Over-The-Top server).

[0241] As an example, the training of the first encoder is performed by OAM (Operation Administration and Maintenance).

[0242] As an example, the training of the first encoder is performed by a NAS (Network Access Server) device.

[0243] As an example, the training of the first encoder is performed by the MnS (Management Service) producer.

[0244] As one example, the second feedback depends on the output of the first encoder.

[0245] As an example, the output of the first encoder includes the second feedback.

[0246] As an example, the output of the first encoder is used to generate the second feedback.

[0247] As one embodiment, all or part of the output of the first encoder is used to generate the second feedback.

[0248] As one embodiment, all or part of the output of the first encoder is post-processed and used to generate the second feedback.

[0249] As one embodiment, the second feedback includes all or part of the output of the first encoder.

[0250] As one embodiment, the second feedback includes all or part of the post-processed output of the first encoder.

[0251] As an example, the post-processing includes one or more of quantization, shortening, puncture, padding, matrix factorization, domain transformation, and DFT (Discrete Fourier Transform).

[0252] As an example, the domain transformation includes one or more of the following: angular domain to spatial domain transformation, spatial domain to angular domain transformation, time domain to frequency domain transformation, frequency domain to time domain transformation, delay domain to frequency domain transformation, frequency domain to delay domain transformation, Doppler domain to time domain transformation, and time domain to Doppler domain transformation.

[0253] As one embodiment, the output obtained by the first encoder with the first channel information as input includes the second feedback.

[0254] As one embodiment, the output obtained by the first encoder with the first channel information as input is used to generate the second feedback.

[0255] As one embodiment, all or part of the output obtained by the first encoder with the first channel information as input is used to generate the second feedback.

[0256] As one embodiment, all or part of the output obtained by the first encoder with the first channel information as input is post-processed and used to generate the second feedback.

[0257] As one embodiment, the second feedback includes all or part of the output obtained by the first encoder with the first channel information as input.

[0258] As one embodiment, the second feedback includes all or part of the post-processed output obtained by the first encoder with the first channel information as input.

[0259] As an example, the first channel information depends on measurements on at least one RS resource.

[0260] As an example, the first channel information depends on channel measurements on at least one RS resource.

[0261] As one embodiment, the first channel information includes a precoding matrix.

[0262] As one embodiment, the first channel information includes a channel matrix or the original channel matrix.

[0263] As one embodiment, the first channel information includes the channel impulse response.

[0264] As one embodiment, the first channel information includes a feature vector.

[0265] As one embodiment, the first channel information includes a feature vector and feature values.

[0266] As one embodiment, the first channel information includes the channel matrix or the feature vector of the original channel matrix.

[0267] As one embodiment, the first channel information includes the eigenvectors and eigenvalues ​​of the channel matrix or the original channel matrix.

[0268] As one example, the matrix includes vectors.

[0269] As one embodiment, the second feedback carries channel information, and the channel information carried by the second feedback is the first channel information.

[0270] As an example, the generation of the first feedback depends on the first encoder.

[0271] As an example, the output of the first encoder includes the first feedback.

[0272] As an example, the output of the first encoder includes the second feedback and the first feedback.

[0273] In a preferred embodiment, the first encoder takes the first channel information as input and obtains an output including the second feedback and the first feedback.

[0274] As an example, the first encoder takes the first channel information as input and the output is used to generate the second feedback and the first feedback.

[0275] As one embodiment, the first encoder takes the first channel information as input, and the output obtained includes the first feedback and is used to generate the second feedback.

[0276] As an example, the generation of the first feedback does not depend on the first encoder.

[0277] As an example, the output of the first encoder does not include the first feedback.

[0278] As one embodiment, the input to the first encoder includes the first feedback.

[0279] As one embodiment, the first encoder takes the first channel information and the first feedback as input, and the output obtained is used to generate the second feedback.

[0280] As one embodiment, the first encoder takes the first channel information and the first feedback as input, and the obtained output includes the second feedback.

[0281] As an example, the generation of the first feedback depends on another encoder that is different from the first encoder.

[0282] As one embodiment, the other encoder takes the first channel information as input and obtains an output including the first feedback; the first encoder takes the first channel information and the first feedback as input and obtains an output including the second feedback or is used to generate the second feedback.

[0283] As an example, the other encoder is based on inference.

[0284] As an example, the other encoder includes inference.

[0285] As an example, the other encoder is based on artificial intelligence or machine learning.

[0286] As an example, the other encoder is based on a neural network or CNN.

[0287] As an example, the training of the other encoder and the training of the first encoder are performed on the same node.

[0288] As an example, the other encoder and the first encoder are jointly trained.

[0289] As an example, the training dataset of the other encoder includes the first dataset.

[0290] As one embodiment, the output obtained by the first encoder with the second channel information as input includes the first feedback; the output obtained by the first encoder with the first channel information and the first feedback as input includes the second feedback or is used to generate the second feedback.

[0291] As one embodiment, the second channel information is obtained earlier than the first channel information.

[0292] As one embodiment, the second channel information depends on a measurement of a first set of transmission timings for one or more RS resources, the first channel information depending on a measurement of a second set of transmission timings for the one or more RS resources, the first set of transmission timings being earlier than the second set of transmission timings.

[0293] As one embodiment, the second channel information includes a channel matrix or the original channel matrix.

[0294] As one embodiment, the second channel information includes a feature vector.

[0295] As one embodiment, the second channel information includes a precoding matrix.

[0296] As one embodiment, the first feedback and the second feedback are transmitted on the same physical layer channel.

[0297] As one embodiment, the first feedback and the second feedback are transmitted on different physical layer channels.

[0298] As an example, the load size of the first feedback is smaller than the load size of the second feedback.

[0299] As one example, the second feedback includes a greater number of bits than the first feedback.

[0300] As an example, the period of the first feedback is longer than the period of the second feedback.

[0301] As a sub-implementation of the above embodiments, both the first feedback and the second feedback are periodic or quasi-static.

[0302] The advantages of the above method include further reducing reporting overhead by leveraging the slow-changing nature of the scene.

[0303] As an example, the first feedback and the second feedback are for the same CSI reporting configuration, and the period of the first feedback is longer than the period of the second feedback.

[0304] As a sub-implementation of the above embodiments, the same CSI reporting configuration indicates the period of the first feedback and the period of the second feedback, respectively.

[0305] In a preferred embodiment, the first feedback and the second feedback depend on measurements on the same one or more RS resources.

[0306] In a preferred embodiment, the first feedback and the second feedback are directed to the same time-frequency resource.

[0307] As one example, the first feedback and the second feedback rely on measurements on the same one or more RS resources and are for the same time-frequency resource.

[0308] As an example, the same one or more RS resources include downlink RS resources.

[0309] As an example, the same one or more RS resources include at least one of CSI-RS resources and SS / PBCHblock resources.

[0310] As one embodiment, the first feedback and the second feedback refer to the same time-frequency resource, including: both the first feedback and the second feedback involve the same time-frequency resource.

[0311] As one embodiment, the first feedback and the second feedback being for the same time-frequency resource includes: both the first feedback and the second feedback being reported for the same time-frequency resource.

[0312] As one embodiment, the first feedback and the second feedback refer to the same time-frequency resource, including: the CSI reference resource of the first feedback and the CSI reference resource of the second feedback are both the same time-frequency resource.

[0313] As an example, the definition of the CSI reference resource is based on 3GPP TS38.214.

[0314] As one embodiment, the first feedback and the second feedback refer to the same time-frequency resource, including: both the first feedback and the second feedback reflect the channel state information within the same time-frequency resource.

[0315] As one embodiment, the first feedback and the second feedback refer to the same time-frequency resource, meaning that the effective range of both the first feedback and the second feedback is limited to the same time-frequency resource.

[0316] As an example, the first feedback depends on the first CSI feedback.

[0317] As an example, the first feedback depends on the first CSI feedback included in the plurality of CSI groups.

[0318] As an example, the first feedback indicates the first CSI feedback included in one of the plurality of CSI groups.

[0319] As an example, the first feedback is the first CSI feedback included in one of the plurality of CSI groups.

[0320] As one example, the data structure is the same, including the same number of bits.

[0321] As one example, the data structures are the same, including the candidate set.

[0322] As one example, the data structures are the same, including the same value range.

[0323] As one example, the data structures are identical, and their physical meanings are the same.

[0324] As an example, the data structure is the same, and the terminology is the same.

[0325] As one example, the data structures are the same, and the mathematical descriptions are the same.

[0326] As one example, the data structures are identical, and the definition methods are identical.

[0327] As an example, the data structure of the first CSI feedback includes at least one string.

[0328] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of Umi, Uma, Sma, Rma, UAV, Urban, Suburban, Rural, Micro, Macro, Hotspot, Indoor, LOS, and NLOS.

[0329] As a sub-example of the above embodiments, the candidates for the at least one string include one or more of FR1, FR2, speed, sparse, dense, ray, path, road, and metropolitan.

[0330] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of the following: sunny, cloudy, rainy, windy, shower, snowy, misty, morning, noon, afternoon, dust, evening, night, morningpeak, and evening peak.

[0331] As an example, the data structure of the first CSI feedback includes at least one number.

[0332] As an example, the data structure of the first feedback is the same as that of the first CSI feedback, including that the number of bits included in the first feedback is equal to the number of bits included in the first CSI feedback.

[0333] As one embodiment, the data structure of the first feedback and the first CSI feedback is the same, including that the first feedback and the first CSI feedback are represented by the same number of bits.

[0334] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that the candidate set of the first feedback and the candidate set of the first CSI feedback are the same.

[0335] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that the first feedback includes at least one string and the first CSI feedback includes at least one string.

[0336] As a sub-implementation of the above embodiments, any string included in the first feedback is a string in the first candidate string set, and any string included in the first CSI feedback is a string in the first candidate string set.

[0337] As a sub-implementation of the above embodiments, the number of strings included in the first feedback is equal to the number of strings included in the first CSI feedback.

[0338] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that the first feedback includes at least one number and the first CSI feedback includes at least one number.

[0339] As a sub-implementation of the above embodiments, any number included in the first feedback is a number in the first candidate number set, and any number included in the first CSI feedback is a number in the first candidate number set.

[0340] As a sub-implementation of the above embodiments, the value range of any number included in the first feedback is the same as the value range of any number included in the first CSI feedback.

[0341] As a sub-implementation of the above embodiments, the number of numbers included in the first feedback is equal to the number of numbers included in the first CSI feedback.

[0342] As an example, the first feedback has the same data structure as the first CSI feedback, including that the first feedback is the first CSI feedback included in one of the plurality of CSI groups.

[0343] As an example, the data structure of the first feedback is the same as that of the first CSI feedback, including the physical meaning of the first feedback being the same as that of the first CSI feedback.

[0344] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that both the first feedback and the first CSI feedback indicate the channel environment.

[0345] As one embodiment, the data structure of the first feedback and the first CSI feedback is the same, including that both the first feedback and the first CSI feedback indicate the movement speed.

[0346] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that both the first feedback and the first CSI feedback indicate a time period.

[0347] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that both the first feedback and the first CSI feedback indicate the frequency domain range.

[0348] As one embodiment, the data structure of the first feedback and the first CSI feedback is the same, including that both the first feedback and the first CSI feedback indicate weather conditions.

[0349] As one embodiment, the data structure of the first feedback and the first CSI feedback is the same, including that both the first feedback and the first CSI feedback indicate a three-dimensional region.

[0350] As an example, the data structure of the first feedback and the first CSI feedback is the same, including that the first feedback and the first CSI feedback both indicate one or more of the following: channel environment, mobile speed, time period, frequency domain range, weather conditions, and three-dimensional region.

[0351] As an example, the data structure of the first feedback is the same as that of the first CSI feedback, including that the first feedback and the first CSI feedback have the same terminology.

[0352] As an example, the data structure of the first feedback is the same as that of the first CSI feedback, including that the first feedback and the first CSI feedback have the same mathematical description.

[0353] As an example, the data structure of the second CSI feedback includes at least one number or at least one complex number.

[0354] As an example, the data structure of the second CSI feedback includes at least one vector.

[0355] As one embodiment, the data structure of the second CSI feedback includes compressed CSI.

[0356] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the number of bits included in the second feedback is equal to the number of bits included in the second CSI feedback.

[0357] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the second feedback and the second CSI feedback are represented by the same number of bits.

[0358] As one embodiment, the data structure of the second feedback is the same as that of the second CSI feedback, including that the candidate set of the second feedback and the candidate set of the second CSI feedback are the same.

[0359] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the second feedback includes at least one number, and the second CSI feedback includes at least one number or a complex number.

[0360] As a sub-implementation of the above embodiments, the number of numbers or complex numbers included in the second feedback is equal to the number of numbers or complex numbers included in the second feedback CSI.

[0361] As a sub-implementation of the above embodiments, the value range of any number or complex number included in the second feedback is the same as the value range of any number or complex number included in the second CSI feedback.

[0362] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the second feedback is used to determine at least one matrix, and the second CSI feedback is used to determine at least one matrix.

[0363] As a sub-implementation of the above embodiment, the number of matrices used to determine by the second feedback is equal to the number of matrices used to determine by the second CSI feedback.

[0364] As a sub-implementation of the above embodiments, any matrix determined by the second feedback and any matrix determined by the second CSI feedback have the same number of rows and columns.

[0365] As a sub-implementation of the above embodiments, any matrix determined by the second feedback and any matrix determined by the second CSI feedback have the same physical meaning.

[0366] As a sub-implementation of the above embodiments, both the matrix determined by the second feedback and the matrix determined by the second CSI feedback are precoding matrices.

[0367] As a sub-implementation of the above embodiments, any matrix determined by the second feedback and any matrix determined by the second CSI feedback are both channel matrices or original channel matrices.

[0368] As a sub-implementation of the above embodiments, both the matrix determined by the second feedback and the matrix determined by the second CSI feedback are composed of eigenvectors.

[0369] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the physical meaning of the second feedback is the same as the physical meaning of the second CSI feedback.

[0370] As an example, the second feedback and the second CSI feedback have the same data structure, including that the second feedback and the second CSI feedback both follow the definition of the same codebook.

[0371] As a sub-example of the above embodiments, the same type of codebook includes codebooks defined by 3GPP Rel-18 or earlier versions.

[0372] As a sub-example of the above embodiments, the same type of codebook includes codebooks for PMI defined in 3GPP Rel-18 or earlier versions.

[0373] As a sub-implementation of the above embodiments, the same type of codebook includes the Type II codebook.

[0374] As an example, the definition of the Type II codebook is found in 3GPP TS38.214.

[0375] As one embodiment, the Type II codebook includes one or more of the following: a Type II port selection codebook, an enhanced Type II codebook, an enhanced Type II port selection codebook, a further enhanced Type II port selection codebook, an enhanced Type II codebook for CJT, a further enhanced Type II port selection codebook for CJT, an enhanced Type II codebook for predicting PMI, and a further enhanced Type II port selection codebook for predicting PMI.

[0376] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the second feedback and the second CSI feedback have the same terminology.

[0377] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the second feedback and the second CSI feedback have the same mathematical description.

[0378] As one embodiment, the second feedback and the second CSI feedback have the same data structure, including that the second feedback and the second CSI feedback are used as inputs to the same decoder, which is the inverse operation of the first encoder.

[0379] As a sub-example of the above embodiments, the same decoder is used to recover or reconstruct channel information.

[0380] As a sub-implementation of the above embodiments, the same decoder is used to recover or reconstruct the channel information carried by the second feedback.

[0381] As a sub-example of the above embodiments, the same decoder is based on inference.

[0382] As one embodiment, the second feedback has the same data structure as the second CSI feedback, including that the generation of the second feedback depends on the first encoder, and the training of the first encoder depends on the second CSI feedback.

[0383] As one embodiment, the second feedback having the same data structure as the second CSI feedback includes the second feedback being the output obtained by the first encoder when it takes a data structure that is the same as the data structure of the target CSI as input.

[0384] As a sub-example of the above embodiment, the error between the output obtained by the first encoder when the target CSI in a CSI group is used as input and the second CSI feedback in the CSI group meets the performance requirements.

[0385] As an example, the data structure that is the same as the target CSI's data structure includes at least one precoding matrix.

[0386] As an example, the data structure, which is the same as the target CSI's data structure, includes at least one precoding matrix represented in the form of a Type II codebook.

[0387] As an example, the data structure that is the same as the target CSI's data structure includes PMI.

[0388] As an example, the data structure that is the same as the target CSI's data structure includes a PMI based on the Type II codebook.

[0389] As an example, the data structure that is the same as the target CSI data structure includes at least one channel matrix.

[0390] As an example, the data structure that is the same as the target CSI data structure includes at least one original channel matrix.

[0391] As an example, the data structure identical to the target CSI's data structure includes the channel impulse response.

[0392] As an example, the data structure, which is the same as the data structure of the target CSI, includes at least one feature vector.

[0393] As an example, the data structure, which is the same as the target CSI's data structure, includes at least one feature vector and at least one feature value.

[0394] As an example, the data structure identical to the target CSI data structure includes at least one channel matrix or one or more eigenvectors of the original channel matrix.

[0395] As an example, the data structure identical to the target CSI data structure includes at least one channel matrix or one or more eigenvectors and one or more eigenvalues ​​of the original channel matrix.

[0396] Example 2

[0397] Example 2 illustrates a schematic diagram of a network architecture according to an embodiment of this application, as shown in the attached diagram. Figure 2 As shown.

[0398] Appendix Figure 2The network architecture 200 is described. The network architecture 200 is a 5G NR (New Radio) / LTE (Long-Term Evolution) / LTE-A (Long-Term Evolution Advanced) system, or a 5G+ network architecture, or a 6G network architecture, or a network architecture adopted in future evolutions by 3GPP; the network architecture 200 may be referred to as 5GS (5G System) / EPS (Evolved Packet System), or 6GS (6G System); the network architecture 200 includes at least one of UE (User Equipment) 201, RAN (Radio Access Network) 202, core network 210, HSS (Home Subscriber Server) / UDM (Unified Data Management) 220, and Internet service 230. The network architecture 200 can interconnect with other access networks, but these entities / interfaces are not shown for simplicity. As shown, the network architecture 200 provides packet-switched services; however, those skilled in the art will readily understand that the various concepts presented throughout this application can be extended to networks providing circuit-switched services or other cellular networks. The RAN includes node 203. The RAN may also include other nodes 204. Node 203 provides user and control plane protocol termination toward UE 201. Node 203 may be connected to other nodes 204 via an Xn interface (e.g., backhaul) / X2 interface. Node 203 may also be referred to as a base station, base transceiver station, radio base station, radio transceiver, transceiver function, basic service set (BSS), extended service set (ESS), TRP (transmitter-receiver node), or some other suitable term. The core network 210 is a 5GC (5G Core Network) / EPC (Evolved Packet Core), or the core network 210 is a 6GC; node 203 provides UE 201 with an access point to the core network 210.Examples of UE201 include cellular phones, smartphones, Session Initiation Protocol (SIP) phones, laptops, personal digital assistants (PDAs), satellite radios, non-terrestrial base station communications, satellite mobile communications, global positioning systems, multimedia devices, video devices, digital audio players (e.g., MP3 players), cameras, game consoles, drones, aircraft, narrowband IoT devices, machine-type communication devices, land vehicles, automobiles, wearable devices, or any other similar functional devices. Those skilled in the art may also refer to UE201 as a mobile station, subscriber station, mobile unit, subscriber unit, radio unit, remote unit, mobile device, radio device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handheld device, user agent, mobile client, client, or any other suitable term. Node 203 is connected to the core network 210 via an S1 / NG interface. The core network 210 includes an MME (Mobility Management Entity) / AMF (Authentication Management Field) / SMF (Session Management Function) 211, other MMEs / AMFs / SMFs 214, an S-GW (Service Gateway) / UPF (User Plane Function) 212, and a P-GW (Packet Data Network Gateway) / UPF 213. The MME / AMF / SMF 211 is the control node that handles signaling between the UE 201 and the core network 210. Generally, the MME / AMF / SMF 211 provides bearer and connection management. All user IP (Internet Protocol) packets are transmitted through the S-GW / UPF 212, which is itself connected to the P-GW / UPF 213. The P-GW provides UE IP address allocation and other functions. The P-GW / UPF 213 is connected to the Internet service 230. Internet services 230 include operator-compliant Internet protocol services, which may specifically include Internet, intranet, IMS (IP Multimedia Subsystem), and packet switching services.

[0399] As an example, the first node includes the UE201.

[0400] As one embodiment, the second node includes the node 203.

[0401] As an example, the wireless link between the UE201 and the node203 includes a cellular link.

[0402] As an example, the sender of the first dataset includes the node 203.

[0403] As an example, the recipient of the first dataset includes the UE201.

[0404] As an example, the sender of the first feedback and the second feedback includes the UE201.

[0405] As an example, the recipients of the first feedback and the second feedback include the node 203.

[0406] As an example, the sender of the first information block includes the node 203.

[0407] As an example, the recipient of the first information block includes the UE201.

[0408] As an example, the UE201 supports AI- or ML-based operations.

[0409] As an example, node 203 supports AI- or ML-based operations.

[0410] Example 3

[0411] Example 3 illustrates a schematic diagram of an embodiment of a wireless protocol architecture for the user plane and control plane according to an embodiment of this application, as shown in the attached diagram. Figure 3 As shown.

[0412] Example 3 illustrates a schematic diagram of an embodiment of a wireless protocol architecture for a user plane and a control plane according to this application, as shown in the attached diagram. Figure 3 As shown. Figure 3 This is a schematic diagram illustrating an embodiment of a radio protocol architecture for the user plane 350 and the control plane 300. Figure 3The radio protocol architecture for the control plane 300 between the first communication node device (UE, gNB, or RSU in V2X) and the second communication node device (gNB, UE, or RSU in V2X), or between two UEs, is illustrated using three layers: Layer 1, Layer 2, and Layer 3. Layer 1 (L1 layer) is the lowest layer and implements various PHY (Physical Layer) signal processing functions. Layer 1 will be referred to as PHY 301 in this document. Layer 2 (L2 layer) 305, above PHY 301, is responsible for the link between the first and second communication node devices, or between two UEs. Layer 2 305 includes the MAC (Medium Access Control) sublayer 302, the RLC (Radio Link Control) sublayer 303, and the PDCP (Packet Data Convergence Protocol) sublayer 304, which terminate at the second communication node device. The PDCP sublayer 304 provides multiplexing between different radio bearers and logical channels. PDCP sublayer 304 also provides security through encrypted data packets and supports cross-cell mobility between second communication node devices and the first communication node device. RLC sublayer 303 provides upper layer data packet segmentation and reassembly, retransmission of lost data packets, and data packet reordering to compensate for out-of-order reception due to HARQ. MAC sublayer 302 provides multiplexing between logical and transport channels. MAC sublayer 302 is also responsible for allocating various radio resources (e.g., resource blocks) within a cell between the first communication node devices. MAC sublayer 302 is also responsible for HARQ operations. The RRC (Radio Resource Control) sublayer 306 in Layer 3 (L3) of the control plane 300 is responsible for acquiring radio resources (i.e., radio bearers) and configuring the lower layer using RRC signaling between the second and first communication node devices. The radio protocol architecture of user plane 350 includes layer 1 (L1 layer) and layer 2 (L2 layer). The radio protocol architecture for the first and second communication node devices in user plane 350 is largely the same as the corresponding layers and sublayers in control plane 300 for physical layer 351, PDCP sublayer 354 in L2 layer 355, RLC sublayer 353 in L2 layer 355 and MAC sublayer 352 in L2 layer 355. However, PDCP sublayer 354 also provides header compression for upper layer data packets to reduce radio transmission overhead.The L2 layer 355 in the user plane 350 also includes an SDAP (Service Data Adaptation Protocol) sublayer 356, which is responsible for mapping between QoS streams and data radio bearers (DRBs) to support service diversity. Although not illustrated, the first communication node device may have several upper layers above the L2 layer 355, including a network layer (e.g., IP layer) terminating at the P-GW on the network side and an application layer terminating at the other end of the connection (e.g., a remote UE, server, etc.).

[0413] As an example, Appendix Figure 3 The wireless protocol architecture described above is applicable to the first node.

[0414] As an example, Appendix Figure 3 The wireless protocol architecture described above is applicable to the second node.

[0415] As an example, the higher layer mentioned in this application refers to the layer above the physical layer.

[0416] As an example, the first dataset is generated in the RRC sublayer 306.

[0417] As an example, the first dataset is generated in a layer above the RRC sublayer 306.

[0418] As an example, the first feedback is generated in the PHY301 or the PHY351.

[0419] As an example, the second feedback is generated in the PHY301 or the PHY351.

[0420] As an example, the first information block is generated in the RRC sublayer 306.

[0421] As an example, the first information block is generated in the MAC sublayer 302 or the MAC sublayer 352.

[0422] Example 4

[0423] Example 4 illustrates a schematic diagram of a first communication device and a second communication device according to an embodiment of this application, as shown in the attached diagram. Figure 4 As shown. (Attached) Figure 4 This is a block diagram of a first communication device 410 and a second communication device 450 communicating with each other in an access network.

[0424] The first communication device 410 includes a controller / processor 475, a memory 476, a receiver processor 470, a transmitter processor 416, a multi-antenna receiver processor 472, a multi-antenna transmitter processor 471, a transmitter / receiver 418, and an antenna 420.

[0425] The second communication device 450 includes a controller / processor 459, a memory 460, a data source 467, a transmitting processor 468, a receiving processor 456, a multi-antenna transmitting processor 457, a multi-antenna receiving processor 458, a transmitter / receiver 454, and an antenna 452.

[0426] In the transmission from the first communication device 410 to the second communication device 450, at the first communication device 410, upper-layer data packets from the core network are provided to the controller / processor 475. The controller / processor 475 implements L2 layer functionality. In DL (Downlink), the controller / processor 475 provides header compression, encryption, packet segmentation and reordering, multiplexing between logical and transport channels, and radio resource allocation to the second communication device 450 based on various priority metrics. The controller / processor 475 is also responsible for HARQ operation, retransmission of lost packets, and signaling to the second communication device 450. The transmit processor 416 and the multi-antenna transmit processor 471 implement various signal processing functions for L1 layer (i.e., physical layer). Transmit processor 416 performs encoding and interleaving to facilitate forward error correction (FEC) at the second communication device 450, and constellation mapping based on various modulation schemes (e.g., binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), M-phase shift keying (M-PSK), and M-quadrature amplitude modulation (M-QAM). Multi-antenna transmit processor 471 performs digital spatial precoding on the encoded and modulated symbols, including codebook-based precoding and non-codebook-based precoding, and beamforming processing, generating one or more parallel... The transmit processor 416 then maps each parallel stream to a subcarrier, multiplexes the modulated symbols with a reference signal (e.g., a pilot) in the time and / or frequency domains, and then uses an inverse fast Fourier transform (IFFT) to generate a physical channel carrying the time-domain multicarrier symbol stream. The multi-antenna transmit processor 471 then performs transmit analog precoding / beamforming operations on the time-domain multicarrier symbol stream. Each transmitter 418 converts the baseband multicarrier symbol stream provided by the multi-antenna transmit processor 471 into an RF stream, which is then provided to a different antenna 420.

[0427] In the transmission from the first communication device 410 to the second communication device 450, at the second communication device 450, each receiver 454 receives a signal through its corresponding antenna 452. Each receiver 454 recovers the information modulated onto the radio frequency carrier and converts the radio frequency stream into a baseband multicarrier symbol stream, which is then provided to the receiver processor 456. The receiver processor 456 and the multi-antenna receiver processor 458 implement various signal processing functions of the L1 layer. The multi-antenna receiver processor 458 performs receive analog precoding / beamforming operations on the baseband multicarrier symbol stream from the receiver 454. The receiver processor 456 uses a Fast Fourier Transform (FFT) to convert the baseband multicarrier symbol stream after the receive analog precoding / beamforming operations from the time domain to the frequency domain. In the frequency domain, the physical layer data signal and the reference signal are demultiplexed by the receiver processor 456, where the reference signal is used for channel estimation, and the data signal is recovered in the multi-antenna receiver processor 458 after multi-antenna detection to recover any parallel stream destined for the second communication device 450. Symbols on each parallel stream are demodulated and recovered in the receive processor 456, generating soft decisions. The receive processor 456 then decodes and deinterleaves the soft decisions to recover the upper-layer data and control signals transmitted over the physical channel by the first communication device 410. The upper-layer data and control signals are then provided to the controller / processor 459. The controller / processor 459 implements the functions of Layer 2 (L2). The controller / processor 459 may be associated with a memory 460 storing program code and data. The memory 460 may be referred to as computer-readable media. In the DL (Layered Logic), the controller / processor 459 provides multiplexing, packet reassembly, decryption, header decompression, and control signal processing between the transmission and logical channels to recover upper-layer packets from the core network. The upper-layer packets are then provided to all protocol layers above Layer 2. Various control signals may also be provided to Layer 3 (L3) for L3 processing. The controller / processor 459 is also responsible for error detection using ACK and / or NACK protocols to support HARQ operation.

[0428] In the transmission from the second communication device 450 to the first communication device 410, at the second communication device 450, a data source 467 is used to provide upper-layer data packets to the controller / processor 459. The data source 467 represents all protocol layers above the L2 layer. Similar to the transmission functions at the first communication device 410 described in the DL, the controller / processor 459 implements header compression, encryption, packet segmentation and reordering, and multiplexing between logical and transport channels based on the radio resource allocation of the first communication device 410, implementing L2 layer functions for the user plane and control plane. The controller / processor 459 is also responsible for HARQ operations, retransmission of lost packets, and signaling to the first communication device 410. Transmit processor 468 performs modulation mapping and channel coding processing, while multi-antenna transmit processor 457 performs digital multi-antenna spatial precoding, including codebook-based and non-codebook-based precoding, and beamforming processing. Subsequently, transmit processor 468 modulates the generated parallel stream into a multi-carrier / single-carrier symbol stream. After analog precoding / beamforming operations in multi-antenna transmit processor 457, the stream is provided to different antennas 452 via transmitter 454. Each transmitter 454 first converts the baseband symbol stream provided by multi-antenna transmit processor 457 into a radio frequency symbol stream before providing it to antenna 452.

[0429] In the transmission from the second communication device 450 to the first communication device 410, the function at the first communication device 410 is similar to the receiving function at the second communication device 450 described in the transmission from the first communication device 410 to the second communication device 450. Each receiver 418 receives radio frequency signals through its corresponding antenna 420, converts the received radio frequency signals into baseband signals, and provides the baseband signals to the multi-antenna receiving processor 472 and the receiving processor 470. The receiving processor 470 and the multi-antenna receiving processor 472 jointly implement the L1 layer functions. The controller / processor 475 implements the L2 layer functions. The controller / processor 475 may be associated with a memory 476 that stores program code and data. The memory 476 may be referred to as computer-readable media. The controller / processor 475 provides multiplexing, packet reassembly, decryption, header decompression, and control signal processing between the transmission and logical channels to recover upper-layer data packets from the second communication device 450. The upper-layer data packets from the controller / processor 475 may be provided to the core network. The controller / processor 475 is also responsible for error detection using ACK and / or NACK protocols to support HARQ operation.

[0430] As one embodiment, the second communication device 450 includes: at least one processor and at least one memory, the at least one memory including computer program code; the at least one memory and the computer program code are configured to be used with the at least one processor. The second communication device 450 means at least: receiving the first dataset; and transmitting the first feedback and the second feedback. The first dataset includes a plurality of CSI groups, each of the plurality of CSI groups including a target CSI, a first CSI feedback, and a second CSI feedback; the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes a greater number of bits than the data structure of the first CSI feedback; the generation of the second feedback depends on a first encoder; the training of the first encoder depends on the first dataset.

[0431] As one embodiment, the second communication device 450 includes: a memory storing a computer-readable instruction program that produces actions when executed by at least one processor, the actions including: receiving the first dataset; and sending the first feedback and the second feedback.

[0432] As one embodiment, the first communication device 410 includes: at least one processor and at least one memory, the at least one memory including computer program code; the at least one memory and the computer program code are configured to be used with the at least one processor. The first communication device 410 means at least: receiving first feedback and second feedback. The generation of the second feedback depends on a first encoder, the training of the first encoder depends on a first dataset, the first dataset including multiple CSI groups, each of the multiple CSI groups including a target CSI, a first CSI feedback and a second CSI feedback; the first feedback has the same data structure as the first CSI feedback, the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback.

[0433] As one embodiment, the first communication device 410 includes: a memory storing a computer-readable instruction program that produces an action when executed by at least one processor, the action including: receiving the first feedback and the second feedback.

[0434] As an example, the first node in this application includes the second communication device 450.

[0435] As an example, the second node in this application includes the first communication device 410.

[0436] As an example, at least one of {the antenna 452, the receiver 454, the receiving processor 456, the multi-antenna receiving processor 458, the controller / processor 459, the memory 460, and the data source 467} is used to receive the first dataset.

[0437] As an example, at least one of the following is used to transmit the first dataset: the antenna 420, the transmitter 418, the transmission processor 416, the multi-antenna transmission processor 471, the controller / processor 475, and the memory 476.

[0438] As an example, at least one of {the antenna 420, the receiver 418, the receiving processor 470, the multi-antenna receiving processor 472, the controller / processor 475, and the memory 476} is used to receive the first feedback and the second feedback; at least one of {the antenna 452, the transmitter 454, the transmitting processor 468, the multi-antenna transmitting processor 457, the controller / processor 459, the memory 460, and the data source 467} is used to transmit the first feedback and the second feedback.

[0439] As an example, at least one of {the antenna 452, the receiver 454, the receiving processor 456, the multi-antenna receiving processor 458, the controller / processor 459, the memory 460, and the data source 467} is used to receive the first information block; at least one of {the antenna 420, the transmitter 418, the transmitting processor 416, the multi-antenna transmitting processor 471, the controller / processor 475, and the memory 476} is used to transmit the first information block.

[0440] Example 5

[0441] Example 5 illustrates a flowchart of a transmission according to an embodiment of this application; as attached Figure 5 As shown. In the appendix Figure 5 In this context, the second node U1 and the first node U2 are communication nodes that transmit data via an air interface. (Appendix) Figure 5 In the middle, the steps in boxes F51 to F58 are selectable respectively.

[0442] For the second node U1, in step S5101, a first decoder is deployed; in step S5102, a first dataset is sent; in step S5103, the first dataset is received; in step S511, first feedback and second feedback are received; in step S5104, the first channel information is recovered or reconstructed based on the first feedback and second feedback; and in step S5105, a first information block is sent.

[0443] For the first node U2, the first encoder is deployed in step S5201; the first dataset is received in step S521; the first encoder is executed in step S5202; the first feedback and the second feedback are sent in step S522; and the first information block is received in step S5203.

[0444] In Example 5, the first dataset includes multiple CSI groups, each of which includes a target CSI, a first CSI feedback, and a second CSI feedback. The first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback. The data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback. The generation of the second feedback depends on the first encoder. The training of the first encoder depends on the first dataset.

[0445] As an example, the first node U2 is the first node in this application.

[0446] As an example, the second node U1 is the second node in this application.

[0447] As one embodiment, the air interface between the second node U1 and the first node U2 includes a wireless interface between the base station equipment and the user equipment.

[0448] As one embodiment, the air interface between the second node U1 and the first node U2 includes a wireless interface between the relay node device and the user equipment.

[0449] As one embodiment, the air interface between the second node U1 and the first node U2 includes the interface between the core network equipment and the user equipment.

[0450] As one embodiment, the air interface between the second node U1 and the first node U2 includes the interface between the OTT server (Over-The-Top server) and the user equipment.

[0451] As one embodiment, the air interface between the second node U1 and the first node U2 includes the interface between the NAS (Network Access Server) device and the user equipment.

[0452] As one embodiment, the air interface between the second node U1 and the first node U2 includes a wireless interface between user equipment and user equipment.

[0453] As one embodiment, the first node U2 includes a terminal.

[0454] As one embodiment, the first node U2 includes a user equipment.

[0455] As one embodiment, the second node U1 includes the serving cell sustaining base station of the first node U2.

[0456] As one embodiment, the second node U1 includes an OTT server (Over-The-Top server).

[0457] As an example, the second node U1 includes OAM (Operation Administration and Maintenance).

[0458] As one embodiment, the second node U1 includes a NAS device.

[0459] As one embodiment, the second node U1 includes core network equipment.

[0460] As an example, the first dataset is transmitted on PDSCH (Physical Downlink SharedChannel).

[0461] As an example, both the first feedback and the second feedback are transmitted on the PUCCH (Physical Uplink Control Channel).

[0462] As an example, both the first feedback and the second feedback are transmitted on PUSCH (Physical Uplink Shared Channel).

[0463] As an example, Appendix Figure 5 The steps in box F53 are present, and the method described above for the second node used in wireless communication includes: transmitting the first dataset.

[0464] As an example, Appendix Figure 5The steps in box F54 are present, and the method described above for the second node used in wireless communication includes: receiving the first dataset.

[0465] As a sub-example of the above embodiments, the sender of the first dataset is a network device other than the first node and the second node.

[0466] As a sub-example of the above embodiments, the sender of the first dataset is a core network device.

[0467] Appendix Figure 5 The steps in boxes F53 and F54 cannot exist simultaneously.

[0468] As an example, Appendix Figure 5 The steps in block F55 are present, and the method described above for the first node used in wireless communication includes: executing the first encoder.

[0469] As an example, Appendix Figure 5 The steps in block F51 are present, and the method described above for the first node used in wireless communication includes: deploying the first encoder.

[0470] As an example, Appendix Figure 5 The step in box F51 is not present, and the first encoder does not need to be deployed.

[0471] As a sub-implementation of the above embodiments, the training of the first encoder is performed by the first node.

[0472] As an example, Appendix Figure 5 The steps in block F56 are present, and the method used in the second node for wireless communication includes: recovering or reconstructing the first channel information based on the first feedback and the second feedback.

[0473] In a preferred embodiment, the recovery or reconstruction of the first channel information is based on inference.

[0474] As an example, the recovery or reconstruction of the first channel information depends on a first decoder.

[0475] As one embodiment, the first decoder takes the received first feedback and second feedback as input to recover or reconstruct the first channel information.

[0476] As an example, the first decoder is based on inference.

[0477] As an example, the first decoder is the inverse operation of the first encoder.

[0478] As an example, the first decoder is based on artificial intelligence or machine learning.

[0479] As an example, the first decoder is based on a neural network or CNN.

[0480] As an example, the first decoder needs to be trained.

[0481] As an example, the model of the first decoder needs to be trained.

[0482] As an example, the training of the first decoder depends on the first dataset.

[0483] As an example, the training of the first decoder refers to the training of the model of the first decoder.

[0484] As an example, the training dataset for the first decoder includes the first dataset.

[0485] As an example, the training of the first decoder and the training of the first encoder are performed on the same node.

[0486] In a preferred embodiment, the training of the first decoder and the training of the first encoder are performed on different nodes.

[0487] As an example, the data structure of the output of the first decoder includes the data structure of the target CSI.

[0488] As an example, the input data structure of the first decoder includes the data structure of the first CSI feedback and the data structure of the second CSI feedback.

[0489] As an example, Appendix Figure 5 The steps in block F52 are present, and the method described above for the second node used in wireless communication includes: deploying the first decoder.

[0490] As an example, Appendix Figure 5 The step in box F52 is missing, and the first decoder does not need to be deployed.

[0491] As a sub-implementation of the above embodiment, the training of the first decoder is performed by the second node.

[0492] As an example, the first decoder satisfies the following consistency requirements: the input includes the first CSI feedback and the second CSI feedback in a CSI group, and the error between the output and the target CSI in the CSI group meets the performance requirements.

[0493] As an example, the input data structure of the first encoder includes the data structure of the target CSI, and the output data structure of the first encoder includes the data structure fed back by the second CSI.

[0494] As an example, the first encoder satisfies the following consistency requirement: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group meets the performance requirements.

[0495] As an example, the first dataset includes multiple data subsets, any one of the multiple data subsets includes a portion of the multiple CSI groups, and all CSI groups in any one of the multiple data subsets include the same first CSI feedback.

[0496] As an example, the second feedback is conditional on the first feedback, which is associated with a first data subset, which is a subset of the first dataset.

[0497] As a preferred embodiment of the above embodiments, all CSI groups in the first data subset include the same first CSI feedback.

[0498] As one embodiment, the first data subset includes a portion of the plurality of CSI groups, and the first feedback indicates the first CSI feedback included in the CSI group of the first data subset.

[0499] As an example, Appendix Figure 5 The steps in box F57 are present, wherein the data structure of the output of the first encoder includes the data structure of the second CSI feedback and the data structure of the first CSI feedback, and the first information block activates or deactivates at least one of the data structure of the second CSI feedback and the data structure of the first CSI feedback output of the first encoder.

[0500] As an example, when the first information block deactivates the data structure of the first CSI feedback output by the first encoder, the first information block must simultaneously deactivate the data structure of the second CSI feedback output by the first encoder.

[0501] As an example, the first information block is transmitted on the PDSCH.

[0502] As an example, the first information block is transmitted on the PDCCH (Physical Downlink Control Channel).

[0503] Example 6

[0504] Example 6 illustrates a schematic diagram of deploying a first encoder according to an embodiment of this application; as shown in the appendix. Figure 6 As shown.

[0505] As one embodiment, the deployment includes obtaining the first encoder.

[0506] As one example, the deployment includes obtaining the model of the first encoder.

[0507] As one example, the deployment includes obtaining an AI entity or function that performs the first encoder.

[0508] As one embodiment, the deployment includes loading the first encoder.

[0509] As one example, the deployment includes loading the model of the first encoder.

[0510] As one embodiment, the deployment includes making a request to load the first encoder.

[0511] As an example, Appendix Figure 6 The request in the request is a request from the first node to load the first encoder.

[0512] As an example, Appendix Figure 6 The response in the code is a response to the request made by the first node to load the first encoder.

[0513] As an example, the first node is attached Figure 6 The response obtained in the first encoder or the model of the first encoder is obtained.

[0514] As an example, the first producer via attached Figure 6 The response in the first node provides the first encoder or a model of the first encoder.

[0515] As an example, the deployment is accomplished by an AI function.

[0516] As an example, the deployment is accomplished by an AI deployment function.

[0517] As an example, the deployment is performed by an AI entity.

[0518] As an example, the deployment is performed by an AI entity with a deployment function.

[0519] As an example, the first producer generates and provides the first encoder.

[0520] As an example, the training of the first encoder is performed by the first producer.

[0521] As an example, the first producer is the producer of the training of the first encoder.

[0522] As one example, the first producer includes an AI entity producer.

[0523] As one example, the first producer includes an AI function producer.

[0524] As one example, the first producer includes an AI deployment producer.

[0525] As one example, the first producer includes an AI training producer.

[0526] As an example, the first producer includes the producer of the AI ​​model training.

[0527] As one example, the first producer includes an MnS (Management Service) producer.

[0528] As an example, the first producer is the serving cell of the first node.

[0529] As an example, the first producer is the maintenance base station of the serving cell of the first node.

[0530] As an example, the first producer is a core network device.

[0531] As an example, the first producer is a NAS device.

[0532] As an example, the first producer is an OTT server.

[0533] Example 7

[0534] Example 7 illustrates a schematic diagram of the input and output of a first encoder according to an embodiment of this application; as shown in the appendix. Figure 7 As shown. In Embodiment 7, the data structure of the input of the first encoder includes the data structure of the target CSI, and the data structure of the output of the first encoder includes the data structure of the second CSI feedback.

[0535] In the appendix Figure 7 In (a), the data structure of the output of the first encoder also includes the data structure of the first CSI feedback.

[0536] In the appendix Figure 7 In (b), the data structure of the input of the first encoder also includes the data structure of the first CSI feedback.

[0537] As an example, the input data structure of the first encoder includes the data structure of the target CSI, and the output data structure of the first encoder includes the data structure of the first CSI feedback and the data structure of the second CSI feedback, as shown in the appendix. Figure 7 As shown in (a).

[0538] As an example, the input data structure of the first encoder includes the data structure of the target CSI and the data structure of the first CSI feedback, and the output data structure of the first encoder includes the data structure of the second CSI feedback, as shown in the appendix. Figure 7 As shown in (b).

[0539] As an example, in the appendix Figure 7 In (b), the generation of the data structure fed back by the first CSI depends on the second encoder, which is based on inference.

[0540] As one example, the second encoder is based on artificial intelligence or machine learning.

[0541] As an example, the second encoder is based on a neural network or CNN.

[0542] As an example, the second encoder needs to be trained.

[0543] As an example, the model of the second encoder needs to be trained.

[0544] As an example, the training of the second encoder depends on the first dataset.

[0545] As an example, the training of the second encoder refers to the training of the model of the second encoder.

[0546] As an example, the training dataset of the second encoder includes the first dataset.

[0547] In a preferred embodiment, the training of the second encoder and the training of the first encoder are performed on the same node.

[0548] As one example, the second encoder and the first encoder are jointly trained.

[0549] As an example, the training of the second encoder and the training of the first encoder are performed on different nodes.

[0550] As an example, the data structure of the output of the second encoder includes the data structure of the first CSI feedback.

[0551] As an example, the data structure of the output of the second encoder is the same as the data structure fed back by the first CSI.

[0552] As an example, the data structure of the input to the second encoder includes the data structure of the target CSI.

[0553] As an example, the input data structure of the second encoder includes the data structure of the target CSI, and the output data structure of the second encoder includes the data structure of the feedback from the first CSI.

[0554] As an example, the generation of the data structure fed back by the first CSI depends on the previous execution of the first encoder.

[0555] As an example, the data structure of the output of the first encoder includes the data structure of the first CSI feedback, and the data structure of the input of the first encoder in one execution includes the data structure of the first CSI feedback output of the previous execution of the first encoder.

[0556] As a sub-implementation of the above embodiment, the data structure of the first CSI feedback of the first encoder's first execution output is used as the input for the first encoder's next execution.

[0557] As an example, the data structure of the target CSI includes at least one precoded matrix.

[0558] As an example, the data structure of the target CSI includes at least one precoded matrix represented in the form of a Type II codebook.

[0559] As an example, the data structure of the target CSI includes PMI.

[0560] As an example, the data structure of the target CSI includes a PMI based on the Type II codebook.

[0561] As an example, the data structure of the target CSI includes at least one channel matrix or a raw channel matrix.

[0562] As an example, the data structure of the target CSI includes a channel impulse response.

[0563] As an example, the data structure of the target CSI includes at least one feature vector.

[0564] As an example, the data structure of the target CSI includes at least one feature vector and at least one feature value.

[0565] As an example, the data structure of the target CSI includes at least one channel matrix or one or more eigenvectors of the original channel matrix.

[0566] As an example, the data structure of the target CSI includes at least one channel matrix or one or more eigenvectors and one or more eigenvalues ​​of the original channel matrix.

[0567] As an example, the data structure of the second CSI feedback includes at least one number.

[0568] As an example, the data structure of the second CSI feedback includes at least one complex number.

[0569] As an example, the data structure of the second CSI feedback includes at least one vector.

[0570] As one embodiment, the data structure of the second CSI feedback includes compressed CSI.

[0571] As one example, the data structure of the second CSI feedback includes PMI.

[0572] As an example, the data structure of the second CSI feedback includes pre-coded information.

[0573] As an example, the data structure of the second CSI feedback is used to determine at least one precoding matrix.

[0574] As an example, the data structure of the second CSI feedback is used to determine at least one channel matrix or the original channel matrix.

[0575] As an example, the data structure of the second CSI feedback is used to determine at least one feature vector.

[0576] As an example, the data structure of the first CSI feedback includes at least one string.

[0577] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of Umi, Uma, Sma, Rma, UAV, Urban, Suburban, Rural, Micro, Macro, Hotspot, Indoor, LOS, and NLOS.

[0578] As a sub-example of the above embodiments, the candidates for the at least one string include one or more of FR1, FR2, speed, sparse, dense, ray, path, road, and metropolitan.

[0579] As a sub-implementation of the above embodiments, the candidates for the at least one string include one or more of the following: sunny, cloudy, rainy, windy, shower, snowy, misty, morning, noon, afternoon, dust, evening, night, morningpeak, and evening peak.

[0580] As an example, the data structure of the first CSI feedback indicates one or more of the following: channel environment, mobile speed, time period, frequency domain range, weather conditions, and three-dimensional region.

[0581] As an example, the data structure of the first CSI feedback includes at least one number.

[0582] Example 8

[0583] Example 8 illustrates a schematic diagram of a first encoder satisfying a conformity requirement according to an embodiment of this application; as shown in the attached diagram. Figure 8 As shown. In Embodiment 8, the first encoder satisfies the following consistency requirement: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group meets the performance requirement.

[0584] As an example, the CSI group is any one of the plurality of CSI groups.

[0585] As an example, the consistency requirement includes that the input includes the target CSI in a CSI group, and the error between the data structure of the output second CSI feedback and the second CSI feedback in the CSI group meets the performance requirements.

[0586] As an example, the consistency requirement includes that the input includes the target CSI in a CSI group, the data structure of the output first CSI feedback is the same as the first CSI feedback in the CSI group, and the error between the data structure of the output second CSI feedback and the second CSI feedback in the CSI group meets the performance requirements.

[0587] As one embodiment, the consistency requirement includes that the input includes the target CSI in a CSI group and the first CSI feedback, and the error between the output and the second CSI feedback in the CSI group meets the performance requirements.

[0588] As an example, the performance requirements are configurable.

[0589] As an example, the performance requirements do not need to be configured.

[0590] As an example, the performance requirements are predefined.

[0591] As an example, the performance requirements are configured for the first node.

[0592] As an example, the performance requirements are configured by higher-layer signaling.

[0593] As an example, the performance requirements are configured by RRC signaling.

[0594] As an example, the performance requirements are configured for MAC CE.

[0595] As one example, the performance requirements depend on the UE capabilities of the first node.

[0596] As an example, the performance requirement is reported by the first node.

[0597] As an example, the performance requirement for the error between the output and the second CSI feedback in the CSI group to meet the performance requirement includes that the NMSE (Normalized Mean Square Error), Cosine Similarity, or SGCS (Squared Generalized Cosine Similarity) between the output and the second CSI feedback in the CSI group is less than or not greater than a threshold.

[0598] As an example, the performance requirement for the error between the output and the second CSI feedback in the CSI group to meet the performance requirement includes that the Manhattan distance, Canberra distance, Euclidean distance, Standardized Euclidean distance, Squared Euclidean distance, or Chebyshev distance between the output and the second CSI feedback in the CSI group is less than or not greater than a threshold.

[0599] As an example, the threshold is configurable.

[0600] As an example, the threshold does not need to be configured.

[0601] As an example, the threshold is predefined.

[0602] As an example, the threshold is configured for the first node.

[0603] As one example, the threshold is configured by a higher-layer signaling.

[0604] As an example, the threshold is configured by RRC signaling.

[0605] As an example, the threshold is configured by MAC CE.

[0606] As one example, the threshold depends on the UE capabilities of the first node.

[0607] As an example, the threshold is reported by the first node.

[0608] As an example, the executor of the training of the first encoder trains the first encoder such that the consistency requirement is met.

[0609] Generally, how the first encoder is trained is determined by the hardware manufacturer. Below are some non-limiting implementation methods:

[0610] As an example, the first encoder is trained such that, when the target CSI included in one of the plurality of CSI groups is taken as input, the error between the output of the first encoder and the feedback of the second CSI included in the same CSI group meets the performance requirements.

[0611] As an example, the first encoder is trained such that, when the target CSI included in one of the plurality of CSI groups is taken as input, the output of the first encoder is as close as possible to the second CSI feedback included in the same CSI group.

[0612] As an example, the trainer of the first encoder takes the target CSI included in one of the plurality of CSI groups as input and the second CSI feedback included in the same CSI group as the desired output when training the first encoder.

[0613] As an example, the trainer of the first encoder takes the target CSI included in one of the plurality of CSI groups as input when training the first encoder, and adjusts the model parameters of the first encoder to reduce the error between the output of the first encoder and the feedback of the second CSI included in the same CSI group.

[0614] As an example, the trainer of the first encoder takes the target CSI included in one of the plurality of CSI groups as input when training the first encoder, and takes the first CSI feedback and the second CSI feedback included in the same CSI group as the desired output.

[0615] As an example, the first encoder is trained such that, when the target CSI included in one of the plurality of CSI groups is taken as input, the data structure of the first CSI feedback output by the first encoder is the same as the first CSI feedback included in the same CSI group, and the error between the data structure of the second CSI feedback output by the first encoder and the second CSI feedback included in the same CSI group meets the performance requirements.

[0616] As an example, the trainer of the first encoder takes the target CSI and the first CSI feedback included in one of the plurality of CSI groups as input and the second CSI feedback included in the same CSI group as the desired output when training the first encoder.

[0617] As an example, the first encoder is trained such that, when the target CSI included in one of the plurality of CSI groups and the first CSI feedback are taken as input, the error between the data structure of the second CSI feedback output by the first encoder and the second CSI feedback included in the same CSI group meets the performance requirements.

[0618] As an example, the trainer of the first encoder takes the target CSI and the first CSI feedback included in one of the plurality of CSI groups as input when training the first encoder, and adjusts the model parameters of the first encoder to reduce the error between the output of the first encoder and the second CSI feedback included in the same CSI group.

[0619] Example 9

[0620] Example 9 illustrates a schematic diagram of a first dataset according to an embodiment of this application; as shown in the appendix. Figure 9 As shown. In Embodiment 9, the first dataset includes multiple data subsets, each corresponding to a multiple first CSI feedback. (See Appendix) Figure 9 In the appendix, the plurality of data subsets includes M data subsets, the plurality of data subsets are M data subsets, the plurality of first CSI feedbacks are M first CSI feedbacks, and the M data subsets and the M first CSI feedbacks correspond one-to-one. Figure 9 In this context, the M data subsets are respectively represented as data subset #0, ..., data subset #(M-1); the M first CSI feedbacks are respectively represented as first CSI feedback #0, ..., first CSI feedback #(M-1).

[0621] As an example, any one of the plurality of first CSI feedbacks is the first CSI feedback included in any CSI group of the corresponding data subset.

[0622] As an example, any one of the plurality of data subsets corresponds to a first CSI feedback, and the first CSI feedback corresponding to any one of the plurality of data subsets is the first CSI feedback included in any CSI group of this data subset.

[0623] As an example, none of the multiple CSI groups simultaneously belong to two data subsets within the multiple data subsets.

[0624] As an example, any one of the plurality of CSI groups belongs to a subset of data in the plurality of data subsets.

[0625] As an example, the first dataset consists of the plurality of data subsets.

[0626] As an example, the plurality of data subsets are arranged sequentially in the first dataset.

[0627] As one example, the plurality of data subsets are identified by a plurality of indexes.

[0628] As a sub-implementation of the above embodiments, the first dataset includes an index of each of the plurality of data subsets.

[0629] As an example, any one of the plurality of data subsets depends on a measurement on at least one RS resource.

[0630] As an example, any one of the plurality of data subsets depends on channel measurements on at least one RS resource.

[0631] As an example, any subset of the plurality of data subsets includes channel information, which depends on measurements on at least one RS resource.

[0632] As an example, any subset of the plurality of data subsets includes channel information, which depends on channel measurements on at least one RS resource.

[0633] As an example, the number of the plurality of data subsets is equal to the number of the plurality of first CSI feedbacks, and the plurality of data subsets and the plurality of first CSI feedbacks correspond one-to-one.

[0634] As an example, the plurality of first CSI feedbacks are all different from each other.

[0635] As an example, any one of the plurality of first CSI feedbacks includes at least one string, and for any two of the plurality of first CSI feedbacks, a string included in one of the two first CSI feedbacks does not belong to the other of the two first CSI feedbacks.

[0636] As an example, any one of the plurality of first CSI feedbacks is used to indicate one or more of the following: channel environment, mobile speed, time period, frequency domain range, weather conditions, and three-dimensional region to which the corresponding data subset is targeted.

[0637] As an example, the channel information included in any of the plurality of data subsets is obtained under the channel environment, mobile speed, frequency domain range, time period, weather conditions and / or three-dimensional region indicated by the first CSI feedback corresponding to any of the data subsets.

[0638] As an example, the data in the first dataset is divided into the plurality of data subsets.

[0639] As an example, a given node divides the data in the first dataset to obtain the multiple data subsets.

[0640] As a sub-implementation of the above embodiment, the given node determines the first CSI feedback corresponding to each of the plurality of data subsets.

[0641] As a sub-implementation of the above embodiments, the given node is the first node.

[0642] As a sub-example of the above embodiments, the given node is the sender of the first dataset.

[0643] As a sub-example of the above embodiments, the given node is the target receiver of the second feedback.

[0644] Generally, how to divide the multiple data subsets and how to determine the first CSI feedback corresponding to each data subset are implementation-related. Some non-limiting implementation methods are described below:

[0645] As an example, a given node divides the data in the first dataset to obtain the multiple data subsets, wherein the given node is the first node, the sender of the first dataset, or the target receiver of the second feedback.

[0646] As an example, the given node divides the data in the first dataset according to the scenario obtained from the data in the first dataset. The scenario includes, but is not limited to, channel environment, moving speed, frequency range, time period, weather conditions, and three-dimensional region.

[0647] As an example, in the first dataset, data obtained under the same or similar scenarios are divided into the same data subset, and the scenarios include, but are not limited to, channel environment, movement speed, frequency range, time period, weather conditions, and three-dimensional region.

[0648] As an example, the given node calculates the large-scale characteristics of the data in the first dataset, and data with the same or similar large-scale characteristics are divided into the same data subset.

[0649] As an example, the given node inputs the first dataset into an AI / ML model, and the output of the AI / ML model includes the plurality of data subsets.

[0650] As a sub-implementation of the above embodiment, the AI / ML model divides the first dataset to obtain and output the plurality of data subsets.

[0651] As an example, the given node divides the data in the first dataset according to the scenario targeted by the data in the first dataset. Data obtained under the same or similar scenarios are divided into the same data subset. The first CSI feedback corresponding to each data subset is used to indicate the scenario targeted by this data subset, such as one or more of the following: channel environment, moving speed, frequency domain range, time period, weather conditions, and three-dimensional region.

[0652] As an example, after the given node divides the first dataset, it statistically analyzes the large-scale characteristics of each of the multiple data subsets obtained, and the first CSI feedback corresponding to each of the multiple data subsets indicates the large-scale characteristics of this data subset.

[0653] As an example, the partitioning of the first dataset includes the partitioning of the plurality of CSI groups, that is, dividing the plurality of CSI groups into the plurality of data subsets.

[0654] As an example, the given node divides the multiple CSI groups according to the scene obtained by the target CSI in the multiple CSI groups. The scene includes, but is not limited to, channel environment, moving speed, frequency domain range, time period, weather conditions, and three-dimensional region.

[0655] As an example, among the multiple CSI groups, the CSI groups obtained by the target CSI under the same or similar scenarios are divided into the same data subset. The scenarios include, but are not limited to, channel environment, moving speed, frequency domain range, time period, weather conditions, and three-dimensional region.

[0656] As an example, any one of the plurality of CSI groups corresponds to a first CSI feedback, and the CSI groups corresponding to the same first CSI feedback constitute a data subset of the plurality of data subsets.

[0657] Example 10

[0658] Example 10 illustrates a schematic diagram of a first dataset comprising multiple data subsets according to an embodiment of this application; as shown in the appendix. Figure 10 As shown. In the appendix Figure 10 In this context, the plurality of data subsets includes M data subsets, which are M data subsets, each identified by M indexes, and the first dataset includes the M indexes. (See appendix...) Figure 10 In this context, the M data subsets are represented as data subset #0, ..., data subset #(M-1); and the M indices are represented as index #0, ..., index #(M-1).

[0659] As an example, any subset of data in the plurality of data subsets includes a corresponding index, as shown in the appendix. Figure 10 As shown in (a).

[0660] As an example, the first dataset includes each data subset from the plurality of data subsets and its corresponding index, as shown in the appendix. Figure 10 As shown in (b).

[0661] As an example, the plurality of data subsets are arranged sequentially in the first dataset.

[0662] Example 11

[0663] Example 11 illustrates a schematic diagram of a first dataset according to an embodiment of this application, comprising multiple data subsets and multiple first CSI feedbacks; as shown in the attached diagram. Figure 11 As shown. In the appendix Figure 11 In the appendix, any one of the plurality of data subsets includes a portion of the plurality of CSI groups; the plurality of data subsets correspond to a plurality of first CSI feedbacks, the plurality of data subsets includes a number equal to M, the plurality of data subsets are M data subsets, and the plurality of first CSI feedbacks are M first CSI feedbacks. Figure 11 In this context, the M data subsets are respectively represented as data subset #0, ..., data subset #(M-1); the M first CSI feedbacks are respectively represented as first CSI feedback #0, ..., first CSI feedback #(M-1).

[0664] In the appendix Figure 11 In (a), any one of the plurality of data subsets includes the corresponding first CSI feedback.

[0665] In the appendix Figure 11In (b), the first dataset includes each of the plurality of data subsets and the corresponding first CSI feedback.

[0666] As an example, the first CSI feedback included in any of the plurality of CSI groups is the first CSI feedback corresponding to the data subset to which any CSI group belongs, for example, in the appendix Figure 11 (a) or appendix Figure 11 (b) in.

[0667] In the appendix Figure 11 In (c), any data subset #i (i = 0, ..., M-1) in the plurality of data subsets includes the CSI group in which the first CSI feedback is the first CSI feedback #i.

[0668] As an example, all CSI groups in any data subset of the plurality of data subsets include the same first CSI feedback, and the first CSI feedback corresponding to any data subset of the plurality of data subsets is the first CSI feedback included in any CSI group of this data subset.

[0669] Example 12

[0670] Example 12 illustrates a schematic diagram of a second feedback conditional on a first feedback according to an embodiment of this application; as shown in the attached diagram. Figure 12 As shown. In Example 12, the first feedback is associated with a first subset of data.

[0671] In a preferred embodiment, the first data subset is a proper subset of the first dataset.

[0672] In a preferred embodiment, the first data subset is one of the plurality of data subsets.

[0673] As one embodiment, the second feedback is included conditionally with respect to the first feedback, and the second feedback is calculated under the condition of the first feedback.

[0674] As an example, the second feedback is conditional upon the first feedback, the second feedback is based on inference, and the output of the inference corresponding to the second feedback includes the first feedback.

[0675] As one embodiment, the second feedback is conditional upon the first feedback, and the generation of both the second feedback and the first feedback depends on the first encoder.

[0676] As one embodiment, the second feedback is included conditionally with respect to the first feedback, and the output of the first encoder includes the second feedback.

[0677] As one embodiment, the second feedback is conditionally included based on the first feedback, and the output of the first encoder includes both the first feedback and the second feedback.

[0678] As an example, the second feedback is conditionally included with the first feedback, and the data structure of the output of the first encoder includes the data structure of the second CSI feedback and the data structure of the first CSI feedback. When the data structure of the second CSI feedback output by the first encoder is the second feedback, the data structure of the first CSI feedback output by the first encoder is the first feedback.

[0679] As one embodiment, the second feedback is conditional upon the first feedback, and the first encoder takes the first channel information as input and obtains an output that includes the first feedback and the second feedback.

[0680] As one embodiment, the second feedback is conditional upon the first feedback, the second feedback is based on reasoning, and the input of the reasoning corresponding to the second feedback includes the first feedback.

[0681] As one embodiment, the second feedback is conditionally included with respect to the first feedback, the input of the first encoder includes the first feedback, and the output of the first encoder includes the second feedback.

[0682] As an example, the second feedback is conditionally included with the first feedback, wherein the data structure of the input of the first encoder includes the data structure of the first CSI feedback, the data structure of the output of the first encoder includes the data structure of the second CSI feedback, and when the data structure of the first CSI feedback input by the first encoder is the first feedback, the data structure of the second CSI feedback output by the first encoder is the second feedback.

[0683] As one embodiment, the second feedback is conditional upon the first feedback, meaning that when the first encoder takes the first channel information and the first feedback as input, the output obtained includes the second feedback.

[0684] As one embodiment, the second feedback includes, conditionally, the recovery or reconstruction of the channel information carried by the second feedback depends on the first feedback.

[0685] As one embodiment, the second feedback is conditional upon the first feedback, and in order to recover or reconstruct the channel information carried by the second feedback, the second feedback and the first feedback are used together as input to a decoder.

[0686] As one embodiment, the second feedback includes, conditionally, the first feedback, and the target receiver of the second feedback recovers or reconstructs the channel information carried by the second feedback based on the first feedback.

[0687] As one embodiment, the second feedback is conditional upon the first feedback, and the first data subset is used to generate the second feedback.

[0688] As one embodiment, the second feedback is conditionally included with respect to the first feedback, and the first data subset is used to recover or reconstruct the channel information carried by the second feedback.

[0689] Example 13

[0690] Example 13 illustrates a schematic diagram of a first feedback indicating a first CSI feedback included in a first subset of data, according to an embodiment of this application; as attached. Figure 13 As shown.

[0691] In a preferred embodiment, all CSI groups in the first data subset include the same first CSI feedback.

[0692] As an example, the first feedback indicates the first CSI feedback included in any CSI group of the first data subset.

[0693] As an example, the first data subset corresponds to a first CSI feedback, and the first CSI feedback corresponding to the first data subset is the first CSI feedback included in any CSI group in the first data subset.

[0694] As an example, the first feedback indicates the first CSI feedback corresponding to the first data subset.

[0695] As an example, the first feedback is the first CSI feedback corresponding to the first data subset.

[0696] Example 14

[0697] Example 14 illustrates a schematic diagram of a first information block according to an embodiment of this application; as attached Figure 14 As shown. In the appendix Figure 14 In (a), the first information block activates at least one of the data structure of the second CSI feedback output by the first encoder and the data structure of the first CSI feedback; in the appendix Figure 14In (b), the first information block deactivates at least one of the data structure of the second CSI feedback output by the first encoder and the data structure of the first CSI feedback.

[0698] As an example, the first information block is carried by an RRC (Radio Resource Control) message.

[0699] As one embodiment, the first information block is carried by higher layer signaling.

[0700] As an example, the first information block is carried by RRC signaling.

[0701] As an example, the first information block is carried by one or more RRC IEs (Information Elements).

[0702] As one embodiment, the first information block includes some or all of the information of each of one or more RRC IEs.

[0703] As an example, the first information block is carried by a MAC CE (Medium Access Control layer Control Element).

[0704] As an example, the first information block is carried by DCI (Downlink Control Information).

[0705] As an example, the first information block is carried by both RRC signaling and MAC CE.

[0706] As an example, the first information block is carried by both MAC CE and DCI.

[0707] As an example, the first information block is carried by both RRC signaling and DCI.

[0708] As an example, the first information block includes a first field indicating whether to activate or deactivate at least one of the data structure of the second CSI feedback output by the first encoder and the data structure of the first CSI feedback.

[0709] As an example, the first information block includes a second field and a third field, the second field indicating whether the data structure of the second CSI feedback output by the first encoder is activated or deactivated, and the third field indicating whether the data structure of the first CSI feedback output by the first encoder is activated or deactivated.

[0710] As a sub-implementation of the above embodiment, if the second field is equal to the first value, the first information block indicates that the data structure of the second CSI feedback output by the first encoder is activated, where the first value is a non-negative integer.

[0711] As a sub-implementation of the above embodiment, if the second field is equal to the second value, the first information block indicates that the data structure of the second CSI feedback output by the first encoder is deactivated, and the second value is a non-negative integer.

[0712] As a sub-implementation of the above embodiment, if the third field is equal to the third value, the first information block indicates that the data structure of the first CSI feedback output by the first encoder is activated, and the third value is a non-negative integer.

[0713] As a sub-implementation of the above embodiment, if the third field is equal to the fourth value, the first information block indicates that the data structure of the first CSI feedback output by the first encoder is deactivated, and the fourth value is a non-negative integer.

[0714] As an example, the activation of a data structure output by the first encoder includes the data structure.

[0715] As an example, activating a data structure output by the first encoder includes reporting the data structure output by the first encoder.

[0716] As an example, the activation of a data structure output by the first encoder includes the use of the data structure output by the first encoder for performance monitoring of the first encoder.

[0717] As an example, the activation of a data structure output by the first encoder includes the generation of the data structure during the execution of the first encoder.

[0718] As an example, the deactivation of a data structure output by the first encoder includes the fact that the output of the first encoder does not include the data structure.

[0719] As an example, deactivating a data structure output by the first encoder includes not reporting the data structure output by the first encoder.

[0720] As an example, deactivating a data structure output by the first encoder includes ensuring that the data structure output by the first encoder is not used for performance monitoring of the first encoder.

[0721] As an example, deactivating a data structure output by the first encoder includes the execution of the first encoder not including the generation of the data structure.

[0722] As one embodiment, the data structure is either the data structure of the second CSI feedback or the data structure of the first CSI feedback.

[0723] As an example, the first encoder outputs only the data structure that is activated between the data structure of the second CSI feedback and the data structure of the first CSI feedback.

[0724] As an example, the first node only reports the active data structure of both the data structure of the second CSI feedback output by the first encoder and the data structure of the first CSI feedback.

[0725] As an example, the first encoder is deactivated when both the data structure of the second CSI feedback output by the first encoder and the data structure of the first CSI feedback are deactivated.

[0726] As an example, the first encoder is activated when at least one of the data structure of the second CSI feedback output by the first encoder and the data structure of the first CSI feedback is activated.

[0727] As one embodiment, deactivating the first encoder includes preventing the first node from using the first encoder.

[0728] As one embodiment, deactivating the first encoder includes stopping or suspending CSI reporting that depends on the first encoder.

[0729] As one embodiment, the activation of the first encoder includes the first node being able to use the first encoder.

[0730] As one embodiment, the activation of the first encoder includes the activation of CSI reporting dependent on the first encoder.

[0731] Example 15

[0732] Example 15 illustrates a schematic diagram of a first information block according to an embodiment of this application; as shown in the appendix. Figure 15 As shown.

[0733] In the appendix Figure 15 In (a), when the first information block deactivates the data structure of the first CSI feedback output by the first encoder, the first information block must simultaneously deactivate the data structure of the second CSI feedback output by the first encoder.

[0734] In the appendix Figure 15 In (b), when the first information block activates the data structure of the second CSI feedback output by the first encoder, the first information block must simultaneously activate the data structure of the first CSI feedback output by the first encoder.

[0735] As an example, if the first information block deactivates only the data structure of the first CSI feedback and the data structure of the second CSI feedback, the first node ignores the first information block.

[0736] As an example, if the first information block deactivates only the data structure of the first CSI feedback and the data structure of the second CSI feedback, the first node considers an error to have occurred.

[0737] As an example, the first node does not expect the first information block to activate only the data structure of the first CSI feedback and the data structure of the second CSI feedback.

[0738] As an example, if the first information block activates only the data structure of the second CSI feedback, whichever of the first CSI feedback data structure is activated, the first node ignores the first information block.

[0739] As an example, if the first information block activates only the data structure of the second CSI feedback, whichever of the first CSI feedback data structure is activated, the first node considers an error to have occurred.

[0740] As an example, the first node does not expect the first information block to activate only the data structure of the second CSI feedback, either the data structure of the first CSI feedback or the data structure of the second CSI feedback.

[0741] Example 16

[0742] Example 16 illustrates a schematic diagram of a first feedback, a second feedback, and first channel information according to an embodiment of this application; as attached. Figure 16 As shown.

[0743] In the appendix Figure 16 In (a), the first node uses the first channel information as the input of the first encoder, and the output of the first encoder includes the first feedback and the second feedback; the target receiver of the second feedback uses the received first feedback and the second feedback as the input of the first decoder to recover or reconstruct the first channel information.

[0744] In the appendix Figure 16 In (b), the first node uses the first channel information and the first feedback as inputs to the first encoder, and the output of the first encoder includes the second feedback; the target receiver of the second feedback uses the received first feedback and the second feedback as inputs to the first decoder to recover or reconstruct the first channel information.

[0745] As an appendix Figure 16 (b) is a sub-implementation where the generation of the first feedback depends on another encoder that is different from the first encoder.

[0746] The embodiment of the other encoder is described in Embodiment 1.

[0747] As an appendix Figure 16 (b) is a sub-implementation where the first feedback is generated when the first encoder takes the second channel information as input.

[0748] The embodiment of the second channel information is shown in Embodiment 1.

[0749] As one embodiment, the first channel information includes a precoding matrix.

[0750] As one embodiment, the first channel information includes a channel matrix.

[0751] As one embodiment, the first channel information includes the original channel matrix.

[0752] As one embodiment, the first channel information includes the channel impulse response.

[0753] As one embodiment, the first channel information includes a feature vector.

[0754] As one embodiment, the first channel information includes the eigenvectors of the channel matrix.

[0755] As one embodiment, the first channel information includes the eigenvectors of the original channel matrix.

[0756] As an example, the recovery or reconstruction of the first channel information depends on the output of the first decoder.

[0757] As an example, when the received first feedback and second feedback are taken as inputs, the output of the first decoder includes the recovered or reconstructed first channel information.

[0758] As an example, when the received first feedback and second feedback are used as inputs, all or part of the output of the first decoder is used to recover or reconstruct the first channel information.

[0759] In a preferred embodiment, the training of the first decoder depends on the first dataset.

[0760] In a preferred embodiment, the training dataset for the first decoder includes the first dataset.

[0761] As an example, the training of the first decoder is performed by the second node.

[0762] As an example, the training of the first decoder is performed by the sender of the first dataset.

[0763] As an example, the training of the first decoder is performed by the core network equipment.

[0764] As an example, the training of the first decoder is performed by an OTT (Over-The-Top) server.

[0765] As an example, the training of the first decoder is performed by OAM.

[0766] As an example, the training of the first decoder is performed by a NAS device.

[0767] As an example, the training of the first decoder is performed by the MnS producer.

[0768] As an example, the training of the first encoder and the training of the first decoder are performed on different nodes.

[0769] Generally, how the first decoder is trained is determined by the hardware manufacturer. Below are some non-limiting implementation methods:

[0770] As an example, the first decoder is trained with the first CSI feedback and the second CSI feedback included in a CSI group as input, and the target CSI included in the same CSI group as the desired output.

[0771] As an example, when the producer training the first decoder takes the first CSI feedback and the second CSI feedback included in a CSI group as input, the model parameters of the first decoder are adjusted to reduce the error between the output of the first decoder and the target CSI included in the same CSI group.

[0772] As an example, the first decoder is trained such that, when the first CSI feedback and the second CSI feedback included in a CSI group are taken as inputs, the error between the output of the first decoder and the target CSI included in the same CSI group meets the performance requirements.

[0773] The above method ensures the matching between the model used to generate the second feedback and the model used to recover / reconstruct the first channel information, thereby improving the performance of CSI reporting.

[0774] Example 17

[0775] Example 17 illustrates a schematic diagram of a processing system based on artificial intelligence or machine learning according to an embodiment of this application; as shown in the appendix. Figure 17 As shown in the figure. In Example 17, the second processor sends a second dataset to the third processor and a third dataset to the fourth processor; the third processor generates a target first-class parameter set based on the second dataset, and sends the generated target first-class parameter set to the fourth processor; the fourth processor processes the third dataset using the target first-class parameter set to obtain a first-class output, and sends the first-class output to the fifth processor. (See Appendix...) Figure 17 In this configuration, the first type of feedback and the second type of feedback are optional; the third processor includes ML training functionality; and the fourth processor includes ML inference functionality.

[0776] As one embodiment, the fifth processor includes ML inference capabilities.

[0777] As one embodiment, the fifth processor includes the inverse operation of the fourth processor.

[0778] As one embodiment, the fifth processor includes ML testing functionality.

[0779] As one example, the fifth processor includes performance monitoring / evaluation of the ML model.

[0780] As an example, the fourth processor sends a first type of feedback to the third processor. The first type of feedback is used to trigger the recalculation or update of the target first type of parameter set, that is, to trigger ML initial training or ML retraining.

[0781] As one embodiment, the fifth processor sends a second type of feedback to the second processor, the second type of feedback being used to generate the second dataset or the third dataset, or the second type of feedback being used to trigger the sending of the second dataset or the third dataset.

[0782] As one embodiment, the second processor generates at least one of the second dataset and the third dataset based on the measurement of the reference signal.

[0783] As one embodiment, the second processor generates at least one of the second dataset and the third dataset based on measurements of the physical layer channel.

[0784] As one embodiment, the second processor generates at least one of the second dataset and the third dataset based on data from the MAC layer or a layer higher than the MAC layer.

[0785] As one embodiment, the fourth processor is located at the first node, and the fifth processor is located at the second node.

[0786] As one embodiment, the fourth processor includes the first encoder.

[0787] As one embodiment, the fifth processor includes the first decoder.

[0788] As an example, the third dataset includes inference data.

[0789] As an example, the third dataset is an inference dataset.

[0790] As an example, the input to an inference belongs to an inference dataset.

[0791] As an example, the second dataset includes training data.

[0792] As an example, the second dataset is the training dataset.

[0793] As an example, the second dataset includes the first dataset.

[0794] As an example, the first dataset includes the second dataset.

[0795] As an example, the second dataset is the first dataset.

[0796] As an example, the third processor is used to train an ML model, and the trained model is described by the target first class of parameter sets.

[0797] As one embodiment, the third processor is located at the first node.

[0798] The above embodiments save on air interface overhead.

[0799] As one embodiment, the third processor is located at the second node.

[0800] The above embodiments support joint training and optimize system performance.

[0801] As one embodiment, the third processor is located in the core network.

[0802] The above embodiments support network-wide joint training, further optimizing system performance.

[0803] As an example, the fourth processor constructs a model based on the target first type of parameter group, and then inputs the third dataset into the constructed model to obtain the first type of output.

[0804] As an example, the first feedback belongs to the first type of output.

[0805] As an example, the second feedback belongs to the first type of output.

[0806] As an example, the fourth processor compares the real data with the first type of output, and the resulting error is used to generate the first type of feedback.

[0807] As an example, the fourth processor generates the first type of feedback through performance monitoring.

[0808] As an example, the first type of feedback is used to reflect the performance of the trained model; when the performance of the trained model fails to meet the requirements, the third processing opportunity recalculates the target first type of parameter set.

[0809] As an example, the fifth processor compares the real data with the first type of output, and the resulting error is used to generate the second type of feedback.

[0810] As an example, the fifth processor generates the second type of feedback through performance monitoring.

[0811] As an example, the second type of feedback is used to reflect the performance of the trained model; when the performance of the trained model fails to meet the requirements, the second processor sends the second dataset to trigger or assist the third processor in recalculating the target first type of parameter set.

[0812] As an example, when the error is too large or the update has not been performed for too long, the performance of the trained model is considered to be unsatisfactory.

[0813] As an example, the target first type of parameter group includes one or more of the following: convolution kernel size, number of convolution layers, convolution stride, pooling kernel size, pooling kernel stride, pooling function, activation function, or number of feature maps.

[0814] As an example, the target first type of parameter group includes one or more of the following: convolution kernel, pooling kernel, pooling function, activation function, parameters of pooling function, or parameters of activation function.

[0815] As one example, the ML includes AI.

[0816] As an example, the ML includes ML and AI.

[0817] Example 18

[0818] Example 18 illustrates a schematic diagram based on artificial intelligence or machine learning according to an embodiment of this application; as attached. Figure 18 As shown. (Attached) Figure 18 This includes a first operation, a second operation, a third operation, a fourth operation, and a fifth operation. In Example 18, the first and second operations belong to a first stage, the third operation belongs to a second stage, the fourth operation belongs to a third stage, and the fifth operation belongs to a fourth stage. (See Appendix...) Figure 18 In the diagram, the lines with arrows indicate the sequence of processes.

[0819] As an example, the first operation includes ML training, the second operation includes ML testing, the third operation includes ML emulation, the fourth operation includes ML entity loading, and the fifth operation includes AI inference.

[0820] As one embodiment, the first stage includes a training phase, the second stage includes an emulation phase, the third stage includes a deployment phase, and the fourth stage includes an emulation phase.

[0821] As an example, the first stage includes ML model training.

[0822] As an example, the first stage includes ML model training and ML testing.

[0823] As an example, the ML model training includes initial training and re-training of one or a group of ML models.

[0824] As an example, the training of the ML model depends on training data.

[0825] As an example, the ML model training includes ML entity validation.

[0826] As an example, the ML entity verification is used to evaluate the performance of the ML entity.

[0827] As an example, the ML entity verification depends on verification data.

[0828] As an example, if the results of ML entity verification do not meet expectations, the ML model will be retrained.

[0829] As an example, the ML testing includes testing the validated ML entities to estimate the performance of the trained ML model.

[0830] As an example, if the ML test results meet expectations, the ML entity proceeds to the next stage; otherwise, the ML model will be retrained.

[0831] As an example, the ML test relies on test data.

[0832] As one embodiment, the second stage includes ML simulation, which performs inference of ML entities in a simulation environment.

[0833] As an example, the ML simulation estimates the performance of ML entity reasoning in a simulation environment before using ML entities.

[0834] As one embodiment, the second stage is optional.

[0835] As an example, the third stage includes ML entity loading, which is to obtain trained ML entities to obtain the desired AI inference capabilities.

[0836] As an example, the third stage is optional.

[0837] As an example, the third stage is no longer needed when the training and inference functions are co-located.

[0838] As an example, the fourth stage includes AI inference.

[0839] As an example, the AI ​​inference relies on inference data.

[0840] As an example, the input to an AI inference belongs to the inference dataset of the AI ​​inference model.

[0841] As one example, the ML includes AI.

[0842] As one example, the AI ​​includes ML.

[0843] Example 19

[0844] Example 19 illustrates a schematic diagram of AI function deployment according to an embodiment of this application; as shown in the appendix. Figure 19 As shown.

[0845] In Example 19, the AI ​​training function of the RAN (RadioAccess Network) domain is located in the 3GPP RAN domain-specific management function, while the AI ​​inference function is located in the UE.

[0846] In Example 19, RAN domain-specific management functions provide AI training function management capabilities and AI inference function management capabilities.

[0847] Example 20

[0848] Example 20 illustrates a schematic diagram of AI function deployment according to an embodiment of this application; as shown in the appendix. Figure 20 As shown.

[0849] In Example 20, the AI ​​training function is a RAN domain-specific management function, while the AI ​​inference function is located locally on the UE.

[0850] In Example 20, the management capability of the AI ​​training function is provided by the RAN domain-specific management function, while the management capability of the AI ​​inference function is provided locally by the UE.

[0851] In the appendix Figure 20 In this context, MnF refers to Management Function.

[0852] Example 21

[0853] Example 21 illustrates a schematic diagram of AI function deployment according to an embodiment of this application; as shown in the appendix. Figure 21 As shown.

[0854] In Example 21, both the AI ​​training function and the AI ​​inference function are located in the UE, wherein the UE provides the ability to train and infer.

[0855] In Example 21, RAN domain-specific management functions provide management capabilities for AI training functions and AI inference functions.

[0856] Example 22

[0857] Example 22 illustrates a schematic diagram of AI function deployment according to one embodiment of this application; as shown in the appendix. Figure 22 As shown.

[0858] In Example 22, both the AI ​​training function and the AI ​​inference function are located in the UE.

[0859] In Example 22, the management capabilities of both the AI ​​training function and the AI ​​inference function are provided locally by the UE.

[0860] In the appendix Figure 22 In this context, MnF refers to Management Function.

[0861] Example 23

[0862] Example 23 illustrates a structural block diagram of a processing apparatus for a first node according to an embodiment of this application; as shown in the appendix. Figure 23 As shown. In the appendix Figure 23 In the first node, the processing device 2300 includes a first receiver 2301 and a first transmitter 2302.

[0863] In embodiment 23, the first receiver 2301 receives the first dataset; the first transmitter 2302 sends the first feedback and the second feedback.

[0864] In Example 23, the first dataset includes multiple CSI groups, each of which includes a target CSI, a first CSI feedback, and a second CSI feedback. The first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback. The data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback. The generation of the second feedback depends on the first encoder. The training of the first encoder depends on the first dataset.

[0865] As an example, the first encoder is based on inference.

[0866] As an example, the first transmitter 2302 executes the first encoder.

[0867] As an example, at least one of the first transmitter 2302 and the first receiver 2301 is equipped with the first encoder.

[0868] As an example, the input data structure of the first encoder includes the data structure of the target CSI, and the output data structure of the first encoder includes the data structure fed back by the second CSI.

[0869] As an example, the first encoder satisfies the following consistency requirement: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group meets the performance requirements.

[0870] As an example, the first dataset includes multiple data subsets, any one of the multiple data subsets includes a portion of the multiple CSI groups, and all CSI groups in any one of the multiple data subsets include the same first CSI feedback.

[0871] As an example, the second feedback is conditional on the first feedback, which is associated with a first data subset, which is a subset of the first dataset.

[0872] As one embodiment, the first data subset includes a portion of the plurality of CSI groups, and the first feedback indicates the first CSI feedback included in the CSI group of the first data subset.

[0873] As an example, the first receiver 2301 receives a first information block; wherein, the data structure output by the first encoder includes the data structure of the second CSI feedback and the data structure of the first CSI feedback, and the first information block activates or deactivates at least one of the data structure of the second CSI feedback and the data structure of the first CSI feedback output by the first encoder.

[0874] As an example, when the first information block deactivates the data structure of the first CSI feedback output by the first encoder, the first information block must simultaneously deactivate the data structure of the second CSI feedback output by the first encoder.

[0875] As one embodiment, the first node includes a terminal.

[0876] As one embodiment, the first node includes a user equipment.

[0877] As one embodiment, the first node includes a relay node device.

[0878] As an example, the first receiver 2301 includes at least one of the following in embodiment 4: {antenna 452, receiver 454, receiver processor 456, multi-antenna receiver processor 458, controller / processor 459, memory 460, data source 467}.

[0879] As one embodiment, the first transmitter 2302 includes at least one of the following in embodiment 4: {antenna 452, transmitter 454, transmission processor 468, multi-antenna transmission processor 457, controller / processor 459, memory 460, data source 467}.

[0880] Example 24

[0881] Example 24 illustrates a structural block diagram of a processing apparatus for a second node according to an embodiment of this application; as shown in the appendix. Figure 24 As shown. In the appendix Figure 24 In the second node, the processing device 2400 includes a first processor 2401.

[0882] In embodiment 24, the first processor 2401 receives first feedback and second feedback.

[0883] In embodiment 24, the generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset, which includes multiple CSI groups, each of the multiple CSI groups including a target CSI, a first CSI feedback and a second CSI feedback; the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback.

[0884] As an example, the first encoder is based on inference.

[0885] As an example, the first processor 2401 sends the first dataset.

[0886] As an example, the first processor 2401 receives the first dataset.

[0887] As an example, the first processor 2401 recovers or reconstructs the first channel information based on the first feedback and the second feedback.

[0888] As one embodiment, the first processor 2401 deploys a first decoder, and the recovery or reconstruction of the first channel information depends on the first decoder.

[0889] As an example, the input data structure of the first encoder includes the data structure of the target CSI, and the output data structure of the first encoder includes the data structure fed back by the second CSI.

[0890] As an example, the first encoder satisfies the following consistency requirement: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group meets the performance requirements.

[0891] As an example, the first dataset includes multiple data subsets, any one of the multiple data subsets includes a portion of the multiple CSI groups, and all CSI groups in any one of the multiple data subsets include the same first CSI feedback.

[0892] As an example, the second feedback is conditional on the first feedback, which is associated with a first data subset, which is a subset of the first dataset.

[0893] As one embodiment, the first data subset includes a portion of the plurality of CSI groups, and the first feedback indicates the first CSI feedback included in the CSI group of the first data subset.

[0894] As one embodiment, the first processor 2401 sends a first information block; wherein the data structure output by the first encoder includes the data structure of the second CSI feedback and the data structure of the first CSI feedback, and the first information block activates or deactivates at least one of the data structure of the second CSI feedback and the data structure of the first CSI feedback output by the first encoder.

[0895] As an example, when the first information block deactivates the data structure of the first CSI feedback output by the first encoder, the first information block must simultaneously deactivate the data structure of the second CSI feedback output by the first encoder.

[0896] As one embodiment, the second node includes a base station.

[0897] As one embodiment, the second node includes a base station device.

[0898] As one embodiment, the second node includes a relay node device.

[0899] As one embodiment, the second node includes the sustaining base station of the serving cell of the first node.

[0900] As one embodiment, the second node includes an OTT (Over-The-Top) server.

[0901] As an example, the second node provides OAM (Operation Administration and Maintenance).

[0902] As one embodiment, the second node includes a NAS (Network Access Server).

[0903] As one embodiment, the second node includes a NAS device.

[0904] As one example, the second node provides network access services.

[0905] As one embodiment, the second node includes core network equipment.

[0906] As one embodiment, the second node includes base station equipment and core network equipment.

[0907] As one embodiment, the second node includes a base station device and a NAS device.

[0908] As an example, the first processor 2401 includes at least one of the following in embodiment 4: {antenna 420, transmitter / receiver 418, transmitter processor 416, receiver processor 470, multi-antenna transmitter processor 471, multi-antenna receiver processor 472, controller / processor 475, memory 476}.

[0909] Those skilled in the art will understand that all or part of the steps in the above methods can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium, such as a read-only memory, hard disk, or optical disk. Optionally, all or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Accordingly, each module unit in the above embodiments can be implemented in hardware or in the form of software functional modules. This application is not limited to any specific combination of software and hardware. The user equipment, terminal, and UE in this application include, but are not limited to, drones, communication modules on drones, remote-controlled aircraft, aircraft, small aircraft, mobile phones, tablets, laptops, vehicle-mounted communication equipment, vehicles, RSUs, wireless sensors, internet access cards, IoT terminals, RFID terminals, NB-IoT terminals, MTC (Machine Type Communication) terminals, eMTC (enhanced MTC) terminals, data cards, internet access cards, vehicle-mounted communication equipment, low-cost mobile phones, low-cost tablets, and other wireless communication devices. The base stations or system equipment in this application include, but are not limited to, macrocell base stations, microcell base stations, small cell base stations, home base stations, relay base stations, eNBs, gNBs, TRPs (Transmitter Receiver Points), GNSS, relay satellites, satellite base stations, airborne base stations, RSUs (Road Side Units), drones, and testing equipment, such as transceivers or signaling testers that simulate some functions of a base station, and other wireless communication equipment.

[0910] Those skilled in the art will understand that the present invention can be practiced in other specified forms without departing from its core or essential characteristics. Therefore, the embodiments disclosed herein should in any way be considered descriptive rather than restrictive. The scope of the invention is defined by the appended claims rather than the foregoing description, and all modifications within their equivalent meaning and scope are considered to be included therein.

Claims

1. A first node used for wireless communication, characterized in that, include: A first receiver receives a first dataset, which includes multiple CSI groups. Each CSI group includes a target CSI, a first CSI feedback, and a second CSI feedback. The first transmitter sends the first and second feedback signals. Wherein, the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback; the generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset.

2. The first node according to claim 1, characterized in that, The input data structure of the first encoder includes the data structure of the target CSI, and the output data structure of the first encoder includes the data structure fed back by the second CSI.

3. The first node according to claim 1 or 2, characterized in that, The first encoder satisfies the following consistency requirements: the input includes the target CSI in a CSI group, and the error between the output and the second CSI feedback in the CSI group meets the performance requirements.

4. The first node according to any one of claims 1 to 3, characterized in that, The first dataset includes multiple data subsets, any one of which includes a portion of the multiple CSI groups, and all CSI groups in any one of the multiple data subsets include the same first CSI feedback.

5. The first node according to any one of claims 1 to 4, characterized in that, The second feedback is conditional on the first feedback, which is associated with a first data subset, which is a subset of the first dataset.

6. The first node according to claim 5, characterized in that, The first data subset includes a portion of the plurality of CSI groups, and the first feedback indicates the first CSI feedback included in the CSI group of the first data subset.

7. The first node according to any one of claims 1 to 6, characterized in that, The first receiver receives a first information block; wherein the data structure output by the first encoder includes the data structure of the second CSI feedback and the data structure of the first CSI feedback, and the first information block activates or deactivates at least one of the data structure of the second CSI feedback and the data structure of the first CSI feedback output by the first encoder.

8. A second node used for wireless communication, characterized in that, include: The first processor receives the first and second feedback; The generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset, which includes multiple CSI groups, each of which includes a target CSI, a first CSI feedback, and a second CSI feedback; the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback.

9. A method used in a first node of wireless communication, characterized in that, include: Receive a first dataset, the first dataset comprising multiple CSI groups, each of the multiple CSI groups comprising a target CSI, a first CSI feedback and a second CSI feedback; Send the first and second feedback; Wherein, the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback; the generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset.

10. A method used in a second node of wireless communication, characterized in that, include: Receive the first and second feedback; The generation of the second feedback depends on the first encoder; the training of the first encoder depends on the first dataset, which includes multiple CSI groups, each of which includes a target CSI, a first CSI feedback, and a second CSI feedback; the first feedback has the same data structure as the first CSI feedback, and the second feedback has the same data structure as the second CSI feedback; the data structure of the second CSI feedback includes more bits than the data structure of the first CSI feedback.

Images