Terminal, wireless communication method, and base station

Abstract

A terminal according to one embodiment of the present disclosure is provided with: a reception unit that receives NW training model information relating to a model (NW training model) trained by a network (NW), or reference model information relating to a reference model that is the basis of the NW training model; a control unit for further training a model determined on the basis of the NW training model information or the reference model information; and a transmission unit that transmits information relating to the trained model. According to one embodiment of the present disclosure, it is possible to achieve appropriate overhead reduction / channel estimation / resource utilization.

Description

Technical Field

[0001] The present disclosure relates to a terminal, a wireless communication method, and a base station in a next-generation mobile communication system. Background Art

[0002] In the Universal Mobile Telecommunications System (UMTS) network, Long Term Evolution (LTE) has been standardized for the purpose of further high-speed data rates and low latency (Non-Patent Document 1). In addition, LTE-Advanced (3GPP Rel. 10-14) has been standardized for the purpose of further increasing capacity and improving (evolution) of LTE (Release (Rel.) 8, 9 of the Third Generation Partnership Project (3GPP)).

[0003] Subsequent systems of LTE (for example, also known as the fifth generation mobile communication system (5G), 5G+(plus), the sixth generation mobile communication system (6G), New Radio (NR), 3GPP Rel.15 and later, etc.) are also discussed.

[0004] Prior art literature

[0005] Non-patent literature

[0006] Non-patent document 1: 3GPP TS 36.300 V8.12.0 “Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description; Stage 2 (Release 8)”, April 2010 Summary of the invention

[0007] Problems to be solved by the invention

[0008] Regarding future wireless communication technologies, the use of artificial intelligence (AI) technologies such as machine learning (ML) for network / device control and management is being discussed.

[0009] It is preferred to be able to appropriately control AI model training / model inference in terminals (user terminals, user equipment (UE)). However, the specific content of the control has not yet been discussed. If the control is not properly specified, appropriate overhead reduction / high-precision channel estimation / efficient resource utilization cannot be achieved, and there is a concern that the communication throughput / communication quality improvement may be suppressed.

[0010] Therefore, one of the objects of the present disclosure is to provide a terminal, a wireless communication method, and a base station that can achieve appropriate overhead reduction / channel estimation / resource utilization.

[0011] Means for solving problems

[0012] A terminal involved in one method of the present disclosure comprises: a receiving unit that receives NW training model information related to a model (NW training model) trained by a network (Network (NW)), or reference model information related to a reference model that serves as the basis of the NW training model; a control unit that further trains a model judged based on the NW training model information or the reference model information; and a sending unit that sends information related to the trained model.

[0013] Effects of the Invention

[0014] According to an aspect of the present disclosure, appropriate overhead reduction / channel estimation / resource utilization can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a diagram showing an example of a framework for managing AI models.

[0016] Figure 2 This is a diagram showing an example of an AI model.

[0017] Figure 3 This is a diagram showing an example of specifying the AI ​​model according to the first embodiment.

[0018] Figure 4 This is a diagram showing an example of updating of the AI ​​model according to the first embodiment.

[0019] Figure 5 This is a diagram showing an example of fine tuning.

[0020] Figure 6This is a diagram showing an example of transfer learning.

[0021] Figure 7 This is a diagram showing an example of updating of the AI ​​model according to the first embodiment.

[0022] Figure 8 This is a diagram showing an example of designation of an AI model according to the second embodiment.

[0023] Fig. 9 This is a diagram showing an example of updating of the AI ​​model according to the second embodiment.

[0024] Fig.10 This is a diagram showing an example of updating of the AI ​​model according to the second embodiment.

[0025] Fig.11 This is a diagram showing an example of designation of an AI model according to the third embodiment.

[0026] Fig.12 This is a diagram showing an example of designation of an AI model according to the third embodiment.

[0027] Figure 13A-13B This is a diagram showing an example of UE training AI model information involved in the third embodiment.

[0028] Figure 14A-14B This is a diagram showing an example of UE training AI model information involved in the third embodiment.

[0029] Fig.15 This is a diagram showing an example of the exchange of AI model information for federated learning.

[0030] Figure 16A-16C This is a diagram showing an example of updated AI model information.

[0031] Fig.17A as well as Fig. 17B This is a diagram showing an example of learning using knowledge distillation.

[0032] Fig.18 This is a diagram showing an example of a schematic configuration of a wireless communication system according to an embodiment.

[0033] Fig.19 This is a diagram showing an example of the configuration of a base station according to an embodiment.

[0034] Fig. 20 This is a diagram showing an example of a configuration of a user terminal according to an embodiment.

[0035] Fig.21 This is a diagram showing an example of the hardware configuration of a base station and a user terminal according to an embodiment.

[0036] Fig. 22 This is a diagram showing an example of a vehicle according to an embodiment. DETAILED DESCRIPTION

[0037] (Application of Artificial Intelligence (AI) technology to wireless communications)

[0038] Regarding future wireless communication technologies, the use of AI technologies such as machine learning (ML) for network / device control and management is being discussed.

[0039] For example, with regard to future wireless communication technologies, discussions are underway to use AI technology to improve channel state information (Channel State Information Reference Signal (CSI)) feedback (e.g., reducing overhead, improving accuracy, prediction), improving beam management (e.g., improving accuracy, prediction in the time / space domain), improving position measurement (e.g., improvement in position estimation / prediction), etc.

[0040] Figure 1 This is a diagram showing an example of a framework for managing an AI model. In this example, each stage associated with an AI model is shown by a framework. This example can also be described as life cycle management of an AI model.

[0041] The data collection stage is equivalent to the stage of collecting data for the generation / update of AI models. The data collection stage includes data collation (e.g., deciding which data to migrate for model training / model inference) and data migration (e.g., migrating data to entities that perform model training / model inference (e.g., UE, gNB)).

[0042] In the model training phase, model training is performed based on the data (training data) migrated from the collection phase. This phase may also include data preparation (e.g., implementation of data preprocessing, cleaning, formatting, conversion, etc.), model training / validation, model testing (e.g., confirmation of whether the trained model meets the performance threshold), model exchange (e.g., migration of models for distributed learning), model deployment / update (deployment / update of models to entities that perform model reasoning), etc.

[0043] In the model inference phase, model inference is performed based on the data (inference data) migrated from the collection phase. This phase may also include data preparation (for example, implementation of data preprocessing, cleaning, formatting, conversion, etc.), model inference, model performance feedback (feedback of model performance to the entity performing model training), output (providing model output to the actor), etc.

[0044] The actor stage may also include action triggers (e.g., decisions on whether to trigger actions on other entities), feedback (e.g., providing feedback on training data / inference data / information required for performance feedback), etc.

[0045] In addition, for example, the training of models for mobility optimization can also be performed in the operation, administration and maintenance (Operation, Administration and Maintenance) / gNodeB (gNB) in the network (NW). The former has advantages in interoperability, large-capacity storage, operator manageability, model flexibility (feature engineering, etc.). The latter has advantages in latency (latency) that does not require model updates, data exchange for model expansion (model deployment), etc. Reasoning of the above models can also be performed in the gNB, for example.

[0046] Furthermore, the entity doing the training / inference can also be different depending on the use case.

[0047] For example, for AI-assisted beam management based on measurement reports, model training can also be performed by OAM / gNB, and model inference can be performed by gNB.

[0048] For AI-assisted UE-assisted positioning, model training can also be performed by the Location Management Function (LMF), which performs model reasoning.

[0049] For CSI feedback / channel estimation using an autoencoder, OAM / gNB / UE can also perform model training, and gNB / UE (jointly) can perform model inference.

[0050] For AI-assisted beam management based on beam measurement or AI-assisted UE positioning, model training can also be performed by OAM / gNB / UE and model inference can be performed by UE.

[0051] As mentioned above, it is preferred that the AI ​​model training can be implemented in the UE. This is because sometimes the UE can directly obtain data that the gNB / OAM cannot directly use. For example, the model training of the autoencoder for CSI feedback requires complete downlink CSI information (or channel information), but only the UE can directly obtain this information.

[0052] If the autoencoder model is trained in gNB / OAM, the UE needs to send the complete downlink CSI information (or channel information), but the communication overhead is large. On the other hand, if the autoencoder model is trained in the UE, the UE needs to maintain the complete downlink CSI information (or channel information), but compared with gNB / OAM, the UE has scarce storage / computing resources.

[0053] Accordingly, it is preferred to be able to properly control model training / model inference in the UE. However, the specific content of the control has not yet been studied. If the control is not properly specified, appropriate overhead reduction / high-precision channel estimation / efficient resource utilization cannot be achieved, and there is a concern that the improvement of communication throughput / communication quality will be suppressed.

[0054] Therefore, the inventors of the present invention have come up with an appropriate control method for model training / model inference. In addition, each embodiment of the present disclosure can also be used in situations where AI / prediction is not used.

[0055] In one embodiment of the present disclosure, a terminal (user terminal, user equipment (UE)) / base station (Base Station (BS)) trains an ML model in a training mode and implements the ML model in an inference mode (also referred to as inference mode, inference mode, etc.). In the inference mode, the accuracy of the ML model trained in the training mode may also be verified (Validation).

[0056] In the present disclosure, UE / BS can also input channel state information, reference signal measurement values, etc. into the ML model, and output high-precision channel state information / measurement values / beam selection / position, future channel state information / radio link quality, etc.

[0057] In addition, in the present disclosure, AI can also be rewritten as an object (also referred to as object, object, data, function, program, etc.) having (implementing) at least one of the following characteristics:

[0058] Estimates based on observed or collected information;

[0059] · Selection based on observed or collected information;

[0060] Predictions based on observed or collected information.

[0061] In the present disclosure, an object may also be, for example, a device, such as a terminal or a base station, etc. In addition, in the present disclosure, an object may also correspond to a program / model / entity operating in the device.

[0062] In addition, in the present disclosure, the ML model may also be rewritten as an object having (implementing) at least one of the following features:

[0063] Producing estimates by feeding information;

[0064] ·Predict estimated values ​​by providing information;

[0065] Discover features by providing information;

[0066] · Select an action by providing information.

[0067] In addition, in the present disclosure, ML models, models, AI models, predictive analytics, predictive analytics models, etc. can also be rewritten to each other. In addition, ML models can also be derived using at least one of regression analysis (e.g., linear regression analysis, multivariate regression analysis, logistic regression analysis), support vector machine (Support Vector Machine), random forest (Random Forest), neural network (Neural Networks), deep learning (Deep Learning), etc. In the present disclosure, models can also be rewritten as at least one of encoders, decoders, tools, etc.

[0068] The ML model outputs at least one piece of information such as an estimated value, a predicted value, a selected operation, a classification, etc. based on the input information.

[0069] ML models can also include supervised learning, unsupervised learning, reinforcement learning, etc. Supervised learning can also be used to learn general rules for mapping inputs to outputs. Unsupervised learning can also be used to learn features of data. Reinforcement learning can also be used to learn operations for maximizing a goal.

[0070] In the present disclosure, the words "generate", "calculate", "derive", etc. may be replaced by each other. In the present disclosure, the words "implement", "run", "operate", "execute", etc. may be replaced by each other. In the present disclosure, the words "train", "learn", "update", "retrain", etc. may be replaced by each other. In the present disclosure, the words "inference", "after-training", "formal use", "actual use", etc. may be replaced by each other. The words "signal" and "signal / channel" may be replaced by each other.

[0071] In the present disclosure, the training mode may also correspond to a mode in which the UE / BS sends / receives signals for the ML model (in other words, an operation mode during training). In the present disclosure, the inference mode may also correspond to a mode in which the UE / BS implements the ML model (e.g., implements the trained ML model to predict output) (in other words, an operation mode during inference).

[0072] In the present disclosure, the training mode may also mean a mode in which a specific signal with high overhead (eg, large amount of resources) is transmitted with respect to a specific signal transmitted in the inference mode.

[0073] In the present disclosure, the training mode may also mean a mode with reference to a first setting (e.g., a first DMRS setting, a first CSI-RS setting, a first CSI report setting). In the present disclosure, the inference mode may also mean a mode with reference to a second setting different from the first setting (e.g., a second DMRS setting, a second CSI-RS setting, a second CSI report setting). The first setting may also be set more than the second setting in terms of at least one of the time resources, frequency resources, code resources, and ports (antenna ports) related to the measurement. In addition, for example, the CSI report setting may also include settings related to the autoencoder.

[0074] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. The wireless communication methods according to the embodiments may be applied individually or in combination.

[0075] In the following embodiments, since the ML model related to the communication between UE and BS is described, the associated subjects are UE and BS, but the application of the embodiments of the present disclosure is not limited to this. For example, for communication between other subjects (for example, communication between UE and UE), the UE and BS in the following embodiments can also be rewritten as the first UE and the second UE. In other words, any one of the UE, BS, etc. of the present disclosure can be rewritten as any UE / BS.

[0076] In the present disclosure, "A / B" and "at least one of A and B" may be replaced with each other. In addition, in the present disclosure, "A / B / C" may also mean "at least one of A, B, and C".

[0077] In the present disclosure, activate, deactivate, indicate (or specify), select, configure, update, determine, etc. may also be mutually rewritten. In the present disclosure, support, control, controllable, operate, and can operate may also be mutually rewritten.

[0078] In the present disclosure, Radio Resource Control (RRC), RRC parameters, RRC messages, high-layer parameters, fields, Information Element (IE), settings, etc. may also be overwritten with each other. In the present disclosure, Medium Access Control (MAC) control elements (MAC Control Element (CE)), update commands, activation / deactivation commands, etc. may also be overwritten with each other.

[0079] In the present disclosure, the higher layer signaling may be, for example, any one of Radio Resource Control (RRC) signaling, Medium Access Control (MAC) signaling, broadcast information, etc., or a combination thereof.

[0080] In the present disclosure, MAC signaling may also use, for example, MAC Control Element (MACCE), MAC Protocol Data Unit (MAC Protocol Data Unit (PDU)), etc. Broadcast information may also be, for example, Master Information Block (MIB), System Information Block (SIB), minimum system information (Remaining Minimum System Information (RMSI)), Other System Information (Other System Information (OSI)), etc.

[0081] In the present disclosure, the physical layer signaling may be, for example, downlink control information (Downlink Control Information (DCI)), uplink control information (Uplink Control Information (UCI)), etc.

[0082] In the present disclosure, index, identifier (ID), indicator, resource ID, etc. may also be mutually overwritten. In the present disclosure, sequence, list, set, group, cluster, subset, etc. may also be mutually overwritten.

[0083] In the present disclosure, panel, UE panel, panel group, beam, beam group, precoder, uplink (UL) transmission entity, transmission / reception point (transmission / reception point (TRP))), base station, spatial relationship information (Spatial Relation Information (SRI)), spatial relationship, SRS resource indicator (SRS Resource Indicator (SRI)), control resource set (COntrol REsource SET (CORESET)), physical downlink shared channel (Physical Downlink Shared Channel (PDSCH)), codeword (Codeword (CW)), transport block (Transport Block (TB)), reference signal (Reference Signal (RS)), antenna port (e.g., demodulation reference signal (DeModulation Reference Signal, DeModulation Reference Signal (DMRS)) port), antenna port group (e.g., DMRS port group), group (e.g., spatial relationship group, code division multiplexing (Code Division Multiplexing (CDM)) group, reference signal group, CORESET group, physical uplink control channel (Physical Downlink Shared Channel (PDSCH)), codeword (Codeword (CW)), transport block (Transport Block (TB)), reference signal (Reference Signal (RS)), antenna port (e.g., demodulation reference signal (DeModulation Reference Signal, DeModulation Reference Signal (DMRS)) port), antenna port group (e.g., DMRS port group), group (e.g., spatial relationship group, code division multiplexing (Code Division Multiplexing (CDM)) group, reference signal group, CORESET group, physical uplink control channel (Physical Uplink Control Channel (PUCCH)) group, PUCCH resource group), resources (for example, reference signal resources, SRS resources), resource sets (for example, reference signal resource sets), CORESET pool, downlink transmission configuration indication state (Transmission Configuration Indication state (TCI state)) (DL TCI state), uplink TCI state (UL TCI state), unified TCI state (unified TCI state), common TCI state (common TCI state), Quasi-Co-Location (Quasi-Co-Location (QCL)), QCL assumptions, etc. can also be rewritten with each other.

[0084] In the present disclosure, CSI-RS, Non Zero Power (NZP) CSI-RS, Zero Power (ZeroPower (ZP)) CSI-RS and CSI Interference Measurement (CSI-IM) may also be replaced with each other. In addition, CSI-RS may also include other reference signals.

[0085] In the present disclosure, the measured / reported RS may also mean the RS measured / reported for CSI reporting.

[0086] In the present disclosure, timing, moment, time, time slot, sub-slot, symbol, sub-frame, etc. may be replaced with each other.

[0087] In the present disclosure, direction, axis, dimension, domain, polarization, polarization component, etc. can also be rewritten.

[0088] In the present disclosure, RS may be, for example, CSI-RS, SS / PBCH block (SS block (SSB)), etc. In addition, RS index may be CSI-RS resource indicator (CSI-RS Resource Indicator (CRI)), SS / PBCH block resource indicator (SS / PBCH Block Indicator (SSBRI)), etc.

[0089] In the present disclosure, estimation, prediction, and inference may be replaced by each other. In addition, in the present disclosure, estimation, prediction, and inference may be replaced by each other.

[0090] In the present disclosure, the autoencoder, encoder, decoder, etc. can also be rewritten as at least one of a model, an ML model, a neural network model, an AI model, an AI algorithm, etc. In addition, the autoencoder can also be rewritten with any autoencoder such as a stacked autoencoder (StackedAutoencoder) and a convolutional autoencoder. The encoder / decoder of the present disclosure can also adopt models such as a residual network (Residual Network (ResNet)), DenseNet, and RefineNet.

[0091] In addition, in the present disclosure, encoders, encoding processes, encoding, and corrections, changes, and controls based on encoders may be replaced with each other. In addition, in the present disclosure, decoders, decoding processes, decoding, and corrections, changes, and controls based on decoders may be replaced with each other.

[0092] In the present disclosure, channel measurement / estimation may also be performed using at least one of, for example, a channel state information reference signal (CSI-RS), a synchronization signal (Synchronization Signal (SS)), a synchronization signal / broadcast channel (Synchronization Signal / Physical Broadcast Channel (SS / PBCH)) block, a demodulation reference signal (DeModulation Reference Signal (DMRS)), a measurement reference signal (Sounding Reference Signal (SRS)), etc.

[0093] In the present disclosure, CSI may also include at least one of a channel quality indicator (Channel Quality Indicator (CQI)), a precoding matrix indicator (Precoding Matrix Indicator (PMI)), a CSI-RS resource indicator (CSI-RS Resource Indicator (CRI)), a SS / PBCH block resource indicator (SS / PBCH Block Resource Indicator (SSBRI)), a layer indicator (Layer Indicator (LI)), a rank indicator (Rank Indicator (RI)), L1-RSRP (Layer 1 Reference Signal Received Power), L1-RSRQ (Reference Signal Received Quality), L1-SINR (Signal to Interference plus Noise Ratio), L1-SNR (Signal to Noise Ratio), etc.

[0094] In the present disclosure, UCI, CSI report, CSI feedback, feedback information, feedback bit, etc. may also be replaced by each other. In addition, in the present disclosure, bit, bit string, bit sequence, sequence, value, information, value obtained from bit, information obtained from bit, etc. may also be replaced by each other.

[0095] In the present disclosure, the layers (for the encoder) may also be mutually rewritten with the layers (input layer, intermediate layer, etc.) used in the AI ​​model. The layers of the present disclosure may also be equivalent to at least one of the input layer, intermediate layer, output layer, batch normalization layer, convolution layer, activation layer, dense layer, normalization layer, pooling layer, attention layer, dropout layer, fully connected layer, etc.

[0096] (Wireless Communication Method)

[0097] In the present disclosure, AI model information may also mean information including at least one of the following:

[0098] Information about the input / output of the AI ​​model;

[0099] Information for pre-processing / post-processing of AI model input / output;

[0100] Information about the parameters of the AI ​​model;

[0101] Training information for AI models (training information);

[0102] Reasoning information for AI models;

[0103] Performance information related to AI models.

[0104] Here, the input / output information of the above AI model may also include information related to at least one of the following:

[0105] Content of input / output data (e.g., RSRP, SINR, amplitude / phase information in the channel matrix (or precoding matrix), information related to the angle of arrival (AOA), information related to the angle of departure (AoD), location information);

[0106] The type of input / output data (e.g., immutable value, floating-point number);

[0107] Quantization interval (Quantization Step Size) of input / output data (e.g., 1 dBm for L1-RSRP);

[0108] The range that the input / output data can take (e.g., [0, 1]).

[0109] In addition, in the present disclosure, information related to AoA may also include information related to at least one of an azimuth angle of arrival and a zenith angle of arrival (ZoA). In addition, information related to AoD may also include information related to at least one of an azimuth angle of departure and a zenith angle of departure (ZoD).

[0110] In the present disclosure, the location information may also be location information related to the UE / NW. The location information may also include at least one of information obtained by a positioning system (e.g., using a satellite positioning system (Global Navigation Satellite System (GNSS)), a Global Positioning System (GPS), etc.) (e.g., latitude, longitude, altitude), information of a base station adjacent to the UE (or in a serving cell) (e.g., an identifier (ID) of a base station / cell), the distance between the BS and the UE, the direction / angle of the BS (UE) seen from the UE (BS), the coordinates of the BS (UE) seen from the UE (BS) (e.g., coordinates of the X / Y / Z axes), etc.), and a specific address of the UE (e.g., an Internet Protocol (IP) address). The location information of the UE is not limited to information based on the location of the BS, but may also be information based on a specific point.

[0111] The location information may also include information related to the actual installation thereof (eg, the location / position / orientation of the antenna, the location / orientation of the antenna panel, the number of antennas, the number of antenna panels, etc.).

[0112] The location information may also include mobility information. The mobility information may also include information indicating a mobility type, and information indicating at least one of a moving speed of the UE, an acceleration of the UE, and a moving direction of the UE.

[0113] Here, the mobility type may also be equivalent to at least one of a fixed location UE, a movable / moving UE, a no mobility UE, a low mobility UE, a middle mobility UE, a high mobility UE, a cell-edge UE, a non-cell-edge UE, etc.

[0114] The information for pre-processing / post-processing of the input / output of the above AI model may also include information related to at least one of the following:

[0115] Whether to apply normalization (e.g., Z-Score normalization, min-max normalization);

[0116] Parameters used for normalization (e.g., mean / variance for Z-score normalization, min / max for min-max normalization);

[0117] Whether to apply specific numerical transformation methods (e.g., one hot encoding, label encoding, etc.);

[0118] · Selection rules for whether to use as training data.

[0119] For example, the input information x can also be normalized by Z score (x new =(x-μ) / σ. Here, μ is the mean of x and σ is the standard deviation) as the preprocessed normalized input information x new The input is sent to the AI ​​model, and the output y from the AI ​​model can also be out Post-processing is applied to obtain the final output y.

[0120] The information of the parameters of the AI ​​model may also include information related to at least one of the following:

[0121] Information about weights in the AI ​​model (e.g., coefficients (binding coefficients) of neurons);

[0122] The structure of the AI ​​model

[0123] The type of AI model as a model component (e.g., Residual Network (ResNet), Dense Network (DenseNet), RefineNet, transformer model, CRBlock, Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU));

[0124] The functionality of the AI ​​model as a model component (e.g., decoder, encoder).

[0125] In addition, the weight information in the above AI model may also include information related to at least one of the following:

[0126] The bit width (size) of the weight information;

[0127] ·The quantization interval of weight information;

[0128] The range of weight information that can be taken;

[0129] Parameters of weights in AI models;

[0130] Information about the differences between the AI ​​model and the one before the update (when updating);

[0131] Weight initialization methods (e.g., zero initialization, random initialization (based on normal distribution / uniform distribution / truncated normal distribution), Xavier initialization (for sigmoid function), He initialization (for Rectified Linear Units (ReLU)))).

[0132] In addition, the structure of the above AI model may also include information related to at least one of the following:

[0133] Number of layers;

[0134] The type of layer (e.g., convolutional, activation, dense, normalization, pooling, attention);

[0135] Layer information;

[0136] Time series-specific parameters (e.g., bidirectionality, time step);

[0137] Parameters used for training (e.g., type of features (L2 regularization, dropout)

[0138] function, etc.), where to place the function (e.g., after which layer)).

[0139] The layer information may also include information related to at least one of the following:

[0140] The number of neurons in each layer;

[0141] Kernel size

[0142] The stride used for pooling / convolutional layers;

[0143] Pooling method (MaxPooling, AveragePooling, etc.)

[0144] Information of residual blocks;

[0145] Number of heads;

[0146] Normalization methods (quantity normalization, instance normalization, layer normalization, etc.);

[0147] Activation functions (Sigmoid, tanh function, ReLU, leaky ReLU information, Maxout, Softmax).

[0148] Figure 2 This is a diagram showing an example of an AI model. This example shows a residual network (ResNet) as model component #1, a converter model as model component #2, and an AI model including a dense layer and a normalization layer. As mentioned above, an AI model can be included as a component of another AI model. In addition, Figure 2 It can also be an AI model that processes from left to right.

[0149] The training information for the above AI model may also include information related to at least one of the following:

[0150] Information about the optimization algorithm used (e.g., optimization type (Stochastic Gradient Descent (SGD)), AdaGrad, Adam, etc.), optimized parameters (learning rate, momentum information, etc.);

[0151] Information about the loss function (e.g., information about the metrics of the loss function (Mean Absolute Error (MAE)), Mean Square Error (MSE), Cross Entropy Loss, NLLoss, KL Divergence)

[0152] wait));

[0153] Parameters (e.g., layers, weights) that should be frozen for training;

[0154] Parameters that should be updated (e.g., layers, weights);

[0155] Parameters that should be used as initial parameters for training (e.g.

[0156] layers, weights);

[0157] The training / updating iterations of the AI ​​model (e.g., (recommended) number of epochs, batch size, amount of data used for training).

[0158] The reasoning information used for the above-mentioned AI model may also include information related to decision tree branch pruning, parameter quantization, etc.

[0159] The performance information related to the above-mentioned AI model may also include information related to the expected value of the loss function defined for the AI ​​model.

[0160] The AI ​​model information related to a specific AI model may be predetermined in the specification or notified to the UE from the network (NW). The AI ​​model specified in the specification may also be referred to as a reference AI model. The AI ​​model information related to the reference AI model may also be referred to as reference AI model information.

[0161] In addition, the AI ​​model information in the present disclosure may also include an index for determining the AI ​​model (for example, it may also be referred to as an AI model index). The AI ​​model information in the present disclosure may also include an AI model index instead of the input / output information of the above-mentioned AI model, or include an AI model index in addition to the input / output information of the above-mentioned AI model. The association between the AI ​​model index and the AI ​​model information (for example, the input / output information of the AI ​​model) may also be specified in a pre-specification, or may be notified to the UE from the NW.

[0162] <First Embodiment>

[0163] The first embodiment relates to the control of utilization of a reference AI model.

[0164] The UE may also receive reference AI model information from the NW. The UE may also deploy the reference AI model (which may be set to be available or configured) based on the received reference AI model information.

[0165] Figure 3 This is a diagram showing an example of specifying an AI model according to the first embodiment. In this example, AI models #1 to #3 are specified in the specification. A UE that receives AI model information indicating AI model #3 from the NW may also deploy AI model #3.

[0166] The UE may also directly use a reference AI model based on (eg, specified by) the received reference AI model information for inference.

[0167] In addition, the UE may also train a reference AI model based on the received reference AI model information.

[0168] In addition, in the present disclosure, the training / updating of the AI ​​model in the UE may also be performed based on a data set possessed by the UE (eg, data obtained from measurement results).

[0169] Figure 4 This is a diagram showing an example of updating the AI ​​model in the first embodiment. In this example, AI models #1-#2 are specified in the specification. A UE that receives AI model information representing AI model #2 from the NW may also deploy AI model #2. In addition, the UE may also update the AI ​​model #2.

[0170] Here, we explain fine-tuning and transfer learning for updating AI models. Figure 5 This is a diagram showing an example of fine-tuning. Figure 6 This is a diagram showing an example of transfer learning.

[0171] like Figure 5 As shown, in fine-tuning, the entity implementing the training uses the parameters of the trained AI model in the source domain as the initial parameters of the AI ​​model in the target domain (pre-training AI model) and performs training to derive the trained AI model in the target domain. In other words, all layers of the trained AI model in the source domain can also be migrated to all layers of the AI ​​model in the target domain, or part of it can be directly used without being updated (fixed (frozen)).

[0172] like Figure 6As shown, in transfer learning (which may also be referred to as network-based deep transfer learning, etc.), the entity implementing the training reuses a portion of the trained layers in the source domain as layers of the target domain. These layers (in the figure, the front (left) layer) can be used as feature extractors. That is, only a portion of the layers of the trained AI model in the source domain may be migrated to the corresponding layers of the AI ​​model in the target domain, and the layers may also be fixed (frozen) or fine-tuned. In addition, the remaining layers in the target domain may not be layers corresponding to the source domain, and may also be retrained (or fine-tuned) in the target domain.

[0173] The UE may not update the parameters (e.g., layers, weights) that should be frozen for training obtained from the reference AI model information during training. The UE may also update the parameters that should be updated (fine-tuned) obtained from the reference AI model information during training.

[0174] The UE may perform training using the parameters that should be used as initial parameters for training (that should be used as initial parameters) obtained from the reference AI model information.

[0175] Figure 7 This is a diagram showing an example of updating the AI ​​model involved in the first embodiment. In this example, the UE is instructed by the NW to update the reference AI model #2. The instruction includes information indicating that the first three layers should be frozen and the last two layers should be fine-tuned. Based on the instruction, the UE can derive an updated AI model #2 in which the first three layers of the reference AI model #2 are frozen and the last two layers are fine-tuned.

[0176] According to the first embodiment described above, the reference AI model can be appropriately used in the UE.

[0177] <Second Embodiment>

[0178] The second embodiment relates to control of utilization of an AI model trained by a NW (hereinafter also referred to as a NW-trained AI model).

[0179] The NW may train a reference AI model. The NW may send AI model information related to the trained reference AI model (in other words, the NW trained AI model) to the UE. The AI ​​model information may also be referred to as NW trained AI model information.

[0180] The UE may also receive reference AI model information from the NW that serves as the basis for the NW training AI model. In addition, the UE may also receive the NW training AI model from the NW. The NW training AI model information may also include any one or a combination of the following information instead of the AI ​​model information already described, or include any one or a combination of the following information in addition to the AI ​​model information already described:

[0181] Reference AI models that serve as the basis for NW training AI models (indexes);

[0182] Information on the differences between the NW trained AI model and the reference AI model.

[0183] The UE may also deploy the NW training AI model (which may be set to be available or equipped) based on the received NW training AI model information.

[0184] Figure 8 1 is a diagram showing an example of specifying an AI model according to the second embodiment. In this example, reference AI models #1-#2 are specified in advance in the specification. NW trains reference AI model #2 and derives NW training AI model #2.

[0185] The NW notifies the UE of the AI ​​model information used to represent the reference AI model #2, and also notifies the WN of the information for training the AI ​​model #2 (e.g., the information of the updated difference). The UE that receives this information can decide the NW training AI model #2 for the reference AI model #2 by considering the information of the updated difference.

[0186] The UE may also directly use the NW trained AI model based on the received NW trained AI model information for reasoning.

[0187] In addition, the UE may also train the NW training AI model based on the received NW training AI model information.

[0188] Fig. 9 is a diagram showing an example of updating the AI ​​model involved in the second embodiment. Figure 8 Similarly, the UE assumes that the NW training AI model #2 is determined. The UE may further update the NW training AI model #2. For example, the UE may also update the same layers (the first three layers shown in the figure) in the NW training AI model #2 as the original reference AI model #2 based on the data set it has.

[0189] The UE may not update the parameters (e.g., layers, weights) that should be frozen for training obtained from the NW training AI model information during training. The UE may also update the parameters that should be updated obtained from the NW training AI model information during training.

[0190] The UE may also perform training using the parameters that should be used as initial parameters for training (that should be used as initial parameters) obtained from the NW training AI model information.

[0191] Fig.10 : is a diagram showing an example of updating the AI ​​model involved in the second embodiment. In this example, the UE is instructed by the NW to update the NW-trained AI model #2. The instruction includes information indicating that the first three layers should be frozen and the last two layers should be fine-tuned. Based on the instruction, the UE can derive the updated AI model #2 after freezing the first three layers of the NW-trained AI model #2 and fine-tuning the last two layers.

[0192] According to the first embodiment described above, in the UE, the NW can be appropriately used to train the AI ​​model.

[0193] <Third Embodiment>

[0194] The third embodiment relates to control of utilization of an AI model trained by a UE (hereinafter also referred to as a UE-trained AI model).

[0195] UE can train the reference AI model / NW training AI model and export the UE training AI model.

[0196] The UE may also use physical layer signaling (e.g., UCI), high-layer signaling (e.g., RRC signaling, MAC CE), specific signals / channels, or a combination thereof to send AI model information related to the UE training AI model (hereinafter also referred to as UE training AI model information) to the NW. The UE training AI model information may also include any one or a combination of the following information instead of the AI ​​model information already described, or include any one or a combination of the following information in addition to the AI ​​model information already described:

[0197] Reference AI model / NW training AI model (index) that forms the basis for UE training AI model;

[0198] Information on the differences between the UE-trained AI model and the reference AI model / NW-trained AI model.

[0199] Fig.11 : is a diagram showing an example of specifying an AI model involved in the third embodiment. In this example, the UE exports the UE-trained AI model. The UE sends the UE-trained AI model information (which may also include the performance information of the AI ​​model) to the NW. The NW may also apply the UE-trained AI model represented by the information.

[0200] Fig.12 This is a diagram showing an example of specifying an AI model involved in the third embodiment. In this example, the UE derives a UE training AI model from a reference AI model. The differences between these models are part of the layers colored in the figure. The UE sends UE training AI model information to the NW. The information may also include information about the reference AI model and information about the differences (performance information of the AI ​​model). The NW can determine the UE training AI model by considering the updated difference information for the reference AI model represented by the information.

[0201] The UE may also report information regarding which weights / layers are updated (contained in, for example, UE training AI model information).

[0202] Figure 13A-13B as well as Figure 14A-14B : is a diagram showing an example of UE training AI model information involved in the third embodiment. In this example, the UE trains the self-encoder (encoder / decoder) and reports the UE training AI model information about the decoder to the NW as needed.

[0203] Fig.13A Corresponding to the case where the UE does not train the AI ​​model, since both the encoder and the decoder are not updated, the UE training AI model information is not sent to the NW.

[0204] Fig. 13B Corresponding to the case where only the encoder is updated, the UE may not include information about the decoder in the UE training AI model information, but may also include information related to the estimated performance of the updated model, for example. Fig. 13B The source can be used for situations where changes in the statistical properties of the input can be adequately addressed by modifications to the encoder alone.

[0205] Fig.14A Corresponding to the case where part of the encoder and decoder are updated. The UE can include information about the decoder (information about the updated part) in the UE training AI model information, for example, including information related to the estimated performance of the updated model and send it. Fig.14A The situation can be used for situations where changes in the statistical properties of the input can be adequately coped with by only relatively small modifications to the decoder.

[0206] Fig. 14B Corresponding to the situation where all encoders and decoders are updated, the UE may include information about the decoder in the UE training AI model information, for example, information related to the estimated performance of the updated model, a reference model report, etc., and send it. Fig. 14B The situation can be used in cases where changes in the statistical properties of the input require relatively large modifications of the decoder.

[0207] From the perspective of reducing communication overhead, the UE preferably Fig. 13B Applicable when medium performance is sufficient Fig. 13B The report, in Fig.14A Applicable when medium performance is sufficient Fig.14A The report, in Fig.14A Application in cases where performance is insufficient Fig. 14B As described above, the UE may also dynamically control the communication overhead involved in the notification of the AI ​​model.

[0208] [Distributed Learning]

[0209] The sending of the UE training AI model information of the third embodiment can be used for distributed learning using a server and a large number of clients in the NW. In this case, the UE training AI model information can also be referred to as client training AI model information.

[0210] Distributed learning can also be federated learning, which includes at least one of the following steps:

[0211] Each client reports its computational resources to the server.

[0212] The server provides the global AI model to the selected clients;

[0213] The selected clients train local AI models based on the global AI models and their own data;

[0214] The selected clients report the parameters of the local AI model to the server respectively;

[0215] The server updates the global AI model in combination with the reported local AI model, and provides the updated global AI model to one or more clients (eg, all clients).

[0216] At least part of the above steps may be repeated. Fig.15 This is a diagram showing an example of the exchange of AI model information for federated learning.

[0217] For example, in the i-th iteration (i is an integer), the server (NW in this example) sends the global AI model G to the client (UE in this example). i The client updates the global AI model G based on its own data set. i And get the local AI model L i The client will communicate with the local AI model L i The relevant update information is fed back (reported) to the server.

[0218] The server updates the global AI model G based on the update information i from more than one client i And obtain the global AI model G i+1 . Federated learning can also be done in this way.

[0219] The NW may also send AI model information related to the global AI model to the UE. The AI ​​model information may also be referred to as global AI model information. The global AI model information may also include reference AI model information.

[0220] The global AI model information may also include only one or a combination of the following information:

[0221] A reference AI model that serves as the basis for global AI models;

[0222] Global AI models (index);

[0223] Information on how the global AI model compares to the reference AI model.

[0224] In addition, the global AI model G i+1 The AI ​​model information may also include a comparison with the global AI model G i Information about the differences in AI models (information).

[0225] The update information i may also include a model ID related to the global AI model / local AI model, and may also include a model ID compared to the global AI model G i The information on the difference in AI models (information) may also correspond to the updated AI model information which will be described later in the fourth embodiment.

[0226] The UE may also report the local AI model L in the update information i or instead thereof. i The intermediate information (e.g., gradients, features, etc.)

[0227] In addition, the server may also be gNB / OAM / LMF / UE, and the client may also be UE / gNB. For example, when the UE is the client, there is an advantage that a large amount of data for training does not need to be sent to the gNB. For example, when the gNB is the client, there is an advantage that a large amount of data for training does not need to be sent to the OAM / LMF.

[0228] In addition, in the present disclosure, the server, NW, gNB, OAM, LMF and UE can also be rewritten to each other. In the present disclosure, the client, NW, UE, gNB can also be rewritten to each other. In addition, in the present disclosure, gNB / UE can also be rewritten to devices in other networks / core networks.

[0229] The client (e.g., UE) may also report information about resources used for AI model training in the current iteration (hereinafter referred to as resource information) to the server (e.g., gNB). The resource information may also include information related to at least one of the following:

[0230] Capabilities related to the training of AI models (e.g., computing resources);

[0231] Information about available datasets (e.g., structure / type / format / volume of datasets);

[0232] • Information related to the client (eg, the status of the UE).

[0233] The computing resources mentioned above may also represent, for example, at least one of the available floating-point operations per second (FLOPS), the performance of the central processing unit (CPU) / graphics processing unit (GPU), the available storage capacity, etc.

[0234] The information related to the client may also include information related to at least one of the following:

[0235] Information about the type of client (device) (e.g., the model name / model number of the device, the name / model number of the communication chip, etc.);

[0236] The result / quality of the estimate (e.g., Block Error Rate (BLER))

[0237] / CQI / RSRP / RSRQ / SINR);

[0238] The client's willingness to process training (e.g., whether more than one client is willing to process training the AI ​​model);

[0239] The remaining amount of data that the client can exchange with the server (i.e., the amount of data that the client can exchange);

[0240] Remaining battery power.

[0241] According to the third embodiment described above, the UE can be appropriately utilized to train the AI ​​model.

[0242] <Fourth Embodiment>

[0243] The fourth embodiment relates to information about differences between other AI models and a certain AI model (information about differences between other AI models with respect to a certain AI model). This information about differences may also be referred to as updated AI model information.

[0244] The updated AI model information may also be used as at least one of the following information as described above, for example:

[0245] Information about the differences between the updated AI model and the AI ​​model before the update;

[0246] Information about the differences between the NW trained AI model and the reference AI model;

[0247] Information on the differences between the UE-trained AI model and the reference AI model / NW-trained AI model.

[0248] The updated AI model information may include weight information represented by an absolute value, or may include weight information represented by a difference value between the updated AI model and the AI ​​model before the update.

[0249] Figure 16A-16C is a diagram showing an example of updated AI model information. In this example, Fig.16A The reference model shown trains the updated AI model information when the NW trains the AI ​​model.

[0250] Fig. 16B Indicates the weight information contained in the updated AI model information, expressed by an absolute value. In this example, the last layer may not be reported because it is not updated. The UE may not expect to report / receive AI model information that has not been updated.

[0251] Fig. 16C Indicates the weight information contained in the updated AI model information and represented by the difference value. The difference value is basically expected to be smaller than the absolute value, so it is expected to reduce the amount of information. In addition, if Fig. 16C The amount of information and Fig. 16B If the weight information is the same, the granularity of each weight information can be reduced, and a more detailed model description can be achieved.

[0252] In addition, when the reference model information also includes information about the structure of the AI ​​model, the updated AI model information may not include information about the structure of the AI ​​model. This is because the UE and the base station can determine the AI ​​model based on the structure of the same AI model as long as they exchange weight information.

[0253] According to the fourth embodiment described above, the signaling overhead for AI model deployment can be appropriately reduced.

[0254] <Fifth Embodiment>

[0255] The fifth embodiment relates to knowledge distillation.

[0256] In knowledge distillation, a single large model or multiple integrated models with good prediction accuracy is prepared as a teacher model, and the knowledge possessed by the teacher model is used to learn a student model that is lightweight and easy to deploy. The student model obtained through distillation is expected to be a model with the same degree of accuracy as the teacher model but lighter.

[0257] Fig.17A as well as Fig. 17B This is a diagram showing an example of learning using knowledge distillation. In normal learning, training is performed so that the output (output from layer m) obtained by inputting a certain data set (learning data) to the AI ​​model is close to the correct answer.

[0258] In addition, for the output from layer m, the variables / values ​​before applying one or more functions (e.g., Sigmoid function / Softmax function) can also be called logical values ​​(logits), and the variables / values ​​after applying one or more functions can also be called Probs. If the sum of Probs is taken for all possible categories of the correct answer, it becomes 1.

[0259] exist Fig.17A In the conventional learning shown in FIG. 1 , the correct answer label of the learning data is used as the target, and learning is performed so that the loss of Probs and hard targets (also called target loss, inference reference loss, etc.) becomes small.

[0260] On the other hand, using Fig. 17B In the learning of knowledge distillation shown in the figure, the teacher's output (Probs) is used as the source target, and learning is performed so that the loss of the student's Probs and the soft target (also called soft target loss, distillation loss, etc.) becomes smaller. In knowledge distillation, the above-mentioned one or more functions can also be, for example, a softmax function with temperature. In addition, the soft target can also correspond to information related to the probability that the input belongs to each level.

[0261] If the logical value of the teacher model is represented by v, Probs is represented by p, the logical value of the student model is represented by z, and Probs is represented by q, then the probs p of level i of the teacher model when the temperature parameter is T i Through p i =exp(vi / T) / Σ j exp(v j / T) to find the probsq of level i of the teacher model when the temperature parameter is T i You can also use q i =exp(z i / T) / Σ j exp(z j / T) to find the answer.

[0262] The soft target loss can be based on the teacher model’s p i With the student model i By obtaining the cross entropy, for example, it can also be obtained as soft target loss = -Σ i p i log(q i ).

[0263] Alternatively, the hard target loss can also be based on q for T = 1 i And find out.

[0264] In the fifth embodiment, the UE may also receive AI model information related to the student model (hereinafter, also referred to as student model information). The UE may also obtain output (train the student model) using the student model specified based on the received student model information, thereby reporting information related to the output to the NW. This step may also be referred to as retraining of the student model.

[0265] The student model information may include any one of the following information or a combination thereof instead of the AI ​​model information already described, or may include any one of the following information or a combination thereof in addition to the AI ​​model information already described:

[0266] Student model (index);

[0267] Teacher model (index);

[0268] Information about the dataset used for the computation of the teacher / student models;

[0269] The value of the soft target for the teacher model / student model;

[0270] The value of the output from a layer of the teacher model / student model;

[0271] The value of the loss function of the teacher model / student model;

[0272] The values ​​of the soft targets used for the teacher / student models;

[0273] The value of the output from a layer for the teacher model / student model;

[0274] The value of the loss function used for the teacher model / student model.

[0275] In addition, the information related to the data set may also correspond to at least one of, for example, a batch of data, an index representing the data set (or data), the data set (or data) itself, etc. The information on the value of the loss function may also include information on one or both of the soft target loss and the hard target loss.

[0276] The UE may also send information related to the retrained student model (hereinafter also referred to as retrained student model information) to the NW using physical layer signaling (e.g., UCI), high-layer signaling (e.g., RRC signaling, MAC CE), a specific signal / channel, or a combination thereof. The retrained student model information may also include any one or a combination of the following information instead of the AI ​​model information already described, or include any one or a combination of the following information in addition to the AI ​​model information already described:

[0277] The (index of) the student model that became the basis for the retrained student model;

[0278] Information related to the knowledge needed to retrain the student model (e.g., soft targets, output from a certain layer);

[0279] Information about the dataset used for retraining;

[0280] Intermediate information (e.g., gradients, features, etc.) for retraining the student model;

[0281] Information about the hard target loss for retraining the student model.

[0282] The above-mentioned information related to the data set may also include information related to the content of the data included in the data set.

[0283] The NW may further update the student model based on the soft target determined by the information related to the retrained student model and the soft target of the teacher data. In addition, including this update of the NW may be referred to as retraining.

[0284] In addition, when the UE receives information related to the teacher model (hereinafter also referred to as teacher model information) from the NW, it can also use the teacher model information for retraining the above-mentioned student model.

[0285] According to the fifth embodiment, even when the UE uses a lightweight student model in consideration of computing power, by reporting the retrained student model information to the NW, knowledge distillation can be re-performed in the NW to generate a higher-precision / lightweight student model.

[0286] <Supplement>

[0287] Any notification from the NW to the UE in the above-mentioned embodiments may be performed using physical layer signaling (e.g., DCI), high-layer signaling (e.g., RRC signaling, MAC CE), specific signals / channels (e.g., PDCCH, PDSCH, reference signals), or a combination thereof.

[0288] When the above notification is performed through MAC CE, the MAC CE is identified by including a new logical channel ID (Logical Channel ID (LCID)) in the MAC subheader.

[0289] When the above notification is performed through DCI, the above notification can also be performed through a specific field of the DCI, a Radio Network Temporary Identifier (RNTI) used in scrambling the cyclic redundancy check (CRC) bits assigned to the DCI, the format of the DCI, etc.

[0290] In addition, any notification from the NW to the UE in the above-mentioned embodiments may be performed periodically, semi-continuously, or aperiodically.

[0291] The encoder / decoder in the above-mentioned implementation mode may also be rewritten with the AI ​​model equipped in the UE / base station. That is to say, the present disclosure is not limited to the case of using an autoencoder, but may also be applied to the case of using any model for reasoning. In addition, the object compressed by the UE / base station of the present disclosure using the encoder is not limited to CSI (or channel / precoding matrix), but may also be any information.

[0292] At least one of the above-mentioned embodiments may also be applied only to a UE that reports a specific UE capability (UE capability) or supports the specific UE capability.

[0293] The specific UE capability may also represent at least one of the following:

[0294] Supporting specific processing / operation / control / information for at least one of the above embodiments;

[0295] Support acquisition / reporting of channel specific information (e.g., LOS, NLOS, location information);

[0296] The maximum FLOPs of the AI ​​model that UE can fine-tune;

[0297] The maximum number of parameters of the AI ​​model that the UE can fine-tune;

[0298] Parameter models supported by the UE;

[0299] UE supported layers / algorithms / functions;

[0300] Computing power;

[0301] Data collection capabilities.

[0302] In addition, the above-mentioned specific UE capabilities can be capabilities that are applied across all frequencies (commonly regardless of frequency), or they can be capabilities by frequency (for example, cell, band, BWP), or they can be capabilities by frequency range (for example, Frequency Range 1 (FR1), FR2, FR3, FR4, FR5, FR2-1, FR2-2), or they can be capabilities by subcarrier spacing (SubCarrier Spacing (SCS)).

[0303] In addition, the above-mentioned specific UE capabilities can be capabilities applied across full-duplex modes (commonly regardless of the duplex mode) or capabilities according to duplex modes (for example, time division duplex (TDD) and frequency division duplex (FDD)).

[0304] In addition, at least one of the above embodiments may also be applied when the UE is set with specific information associated with the above embodiments through high-level signaling. For example, the specific information may also be information indicating the use of the activated AI model, any RRC parameter for a specific version (e.g., Rel.18), etc.

[0305] The UE may also apply operations such as Rel.15 / 16 without supporting at least one of the above-mentioned specific UE capabilities or without being set with the above-mentioned specific information.

[0306] In addition, at least one of the above-mentioned embodiments may also be used for transmission (compression) of information between UE and base station other than CSI feedback. For example, the UE may also report information related to position (or positioning) / information related to position estimation in the location management function (LMF) according to at least one of the above-mentioned embodiments (for example, generated by an encoder) to the network. The information may also be information on the channel impulse response (Channel Impulse Response (CIR)) by subband / per antenna port. If it is reported, the base station can estimate the position of the UE even if the angle / time difference of the received signal is not reported.

[0307] (Note)

[0308] The following inventions are added to one embodiment of the present disclosure.

[0309] [Note 1]

[0310] A terminal having:

[0311] A receiving unit that receives NW training model information related to a model (NW training model) trained by a network (Network (NW)) and reference model information related to a reference model that is a basis for the NW training model; and

[0312] The control unit determines the NW training model based on the NW training model information and the reference model information.

[0313] [Note 2]

[0314] A terminal as described in Appendix 1, wherein:

[0315] The NW training model information indicates a difference between the NW training model and the reference model, and the reference model information indicates an index related to the reference model.

[0316] [Note 3]

[0317] A terminal as described in Supplement 1 or Supplement 2, wherein:

[0318] The control unit further trains the NW training model.

[0319] [Note 4]

[0320] A terminal as described in any one of Notes 1 to 3, wherein:

[0321] The NW training model information includes information about parameters that should be frozen for training.

[0322] The control unit further trains the NW training model, wherein the parameters are not updated.

[0323] (Note)

[0324] The following inventions are added to one embodiment of the present disclosure.

[0325] [Note 1]

[0326] A terminal having:

[0327] A receiving unit, receiving NW training model information related to a model (NW training model) trained by a network (Network (NW)) or reference model information related to a reference model that is a basis for the NW training model;

[0328] a control unit, further training a model determined based on the NW training model information or the reference model information; and

[0329] A sending unit sends information related to the trained model.

[0330] [Note 2]

[0331] A terminal as described in Appendix 1, wherein:

[0332] The information about the trained model includes information about which weights or layers are updated.

[0333] [Note 3]

[0334] A terminal as described in Supplement 1 or Supplement 2, wherein:

[0335] The information related to the trained model may also include update information of the global model in federated learning.

[0336] [Note 4]

[0337] A terminal as described in any one of Notes 1 to 3, wherein:

[0338] The sending unit sends information about resources used for model training in a current iteration in federated learning.

[0339] (Note)

[0340] The following inventions are added to one embodiment of the present disclosure.

[0341] [Note 1]

[0342] A terminal having:

[0343] A receiving unit, receiving student model information related to a student model in knowledge distillation; and

[0344] A sending unit sends information related to the output based on the student model information.

[0345] [Note 2]

[0346] A terminal as described in Appendix 1, wherein:

[0347] The information about the output includes information about the soft target of the student model.

[0348] [Note 3]

[0349] A terminal as described in Supplement 1 or Supplement 2, wherein:

[0350] The student model information may also include the value of the loss function.

[0351] [Note 4]

[0352] The student model information includes information of one or both of soft target loss and hard target loss.

[0353] (Wireless Communication System)

[0354] Hereinafter, a configuration of a wireless communication system according to an embodiment of the present disclosure will be described. In the wireless communication system, communication is performed using any one of the wireless communication methods according to the above-mentioned embodiments of the present disclosure or a combination thereof.

[0355] Fig.18 1 is a diagram showing an example of a schematic structure of a wireless communication system involved in one embodiment. The wireless communication system 1 may also be a system that implements communication using Long Term Evolution (LTE) standardized by the Third Generation Partnership Project (3GPP), the fifth generation mobile communication system New Radio (5G NR), and the like.

[0356] In addition, the wireless communication system 1 can also support dual connectivity (Multi-RAT Dual Connectivity (MR-DC)) between multiple radio access technologies (Radio Access Technology (RAT)). MR-DC can also include dual connectivity between LTE (Evolved Universal Terrestrial Radio Access (E-UTRA))) and NR (E-UTRA-NR Dual Connectivity (E-UTRA-NR Dual Connectivity (EN-DC))), dual connectivity between NR and LTE (NR-E-UTRA Dual Connectivity (NR-E-UTRA Dual Connectivity (NE-DC))), etc.

[0357] In EN-DC, the base station (eNB) of LTE (E-UTRA) is the master node (Master Node (MN)), and the base station (gNB) of NR is the secondary node (Secondary Node (SN)). In NE-DC, the base station (gNB) of NR is the MN, and the base station (eNB) of LTE (E-UTRA) is the SN.

[0358] The wireless communication system 1 can also support dual connectivity between multiple base stations within the same RAT (for example, dual connectivity (NR-NR Dual Connectivity (NN-DC)) where both MN and SN are NR base stations (gNB)).

[0359] The wireless communication system 1 may also include a base station 11 that forms a macro cell C1 with a relatively wide coverage, and a base station 12 (12a-12c) that is configured in the macro cell C1 and forms a small cell C2 that is narrower than the macro cell C1. The user terminal 20 may also be located in at least one cell. The configuration and number of each cell and user terminal 20 are not limited to the method shown in the figure. Hereinafter, when the base stations 11 and 12 are not distinguished, they are collectively referred to as base stations 10.

[0360] The user terminal 20 may be connected to at least one of the plurality of base stations 10. The user terminal 20 may use at least one of carrier aggregation (CA) using a plurality of component carriers (CC) and dual connectivity (DC).

[0361] Each CC may also be included in at least one of the first frequency band (Frequency Range 1 (FR1)) and the second frequency band (Frequency Range 2 (FR2)). The macro cell C1 may also be included in FR1, and the small cell C2 may also be included in FR2. For example, FR1 may be a frequency band below 6 GHz (below 6 GHz (sub-6 GHz)), and FR2 may be a frequency band higher than 24 GHz (above-24 GHz). In addition, the frequency bands and definitions of FR1 and FR2 are not limited to these. For example, FR1 may also be equivalent to a frequency band higher than FR2.

[0362] Furthermore, the user terminal 20 may perform communication in each CC using at least one of time division duplex (TDD) and frequency division duplex (FDD).

[0363] Multiple base stations 10 may also be connected via wired (e.g., optical fiber based on Common Public Radio Interface (CPRI)), X2 interface, etc.) or wireless (e.g., NR communication). For example, when NR communication between base stations 11 and 12 is used as a backhaul, the base station 11 equivalent to the upper station may also be referred to as an Integrated Access Backhaul (IAB) donor, and the base station 12 equivalent to a relay station (relay) may also be referred to as an IAB node.

[0364] The base station 10 may also be connected to the core network 30 via other base stations 10 or directly. The core network 30 may also include, for example, at least one of an Evolved Packet Core (EPC), a 5G Core Network (5GCN), a Next Generation Core (NGC), and the like.

[0365] The user terminal 20 may also be a terminal that supports at least one of communication modes such as LTE, LTE-A, and 5G.

[0366] In the wireless communication system 1, a wireless access method based on orthogonal frequency division multiplexing (OFDM) may be used. For example, in at least one of the downlink (DL) and the uplink (UL), cyclic prefix OFDM (CP-OFDM), discrete Fourier transform spread OFDM (DFT-s-OFDM), orthogonal frequency division multiple access (OFDMA), single carrier frequency division multiple access (SC-FDMA), etc. may be used.

[0367] The radio access scheme may also be referred to as a waveform. In addition, in the wireless communication system 1, other radio access schemes (for example, other single-carrier transmission schemes, other multi-carrier transmission schemes) may be used as the radio access schemes for UL and DL.

[0368] As downlink channels, the wireless communication system 1 may use a downlink shared channel (Physical Downlink Shared Channel (PDSCH)) shared by each user terminal 20, a broadcast channel (Physical Broadcast Channel (PBCH))), a downlink control channel (Physical Downlink Control Channel (PDCCH)), etc.

[0369] In addition, as uplink channels, an uplink shared channel (Physical Uplink Shared Channel (PUSCH)) shared by each user terminal 20), an uplink control channel (Physical Uplink Control Channel (PUCCH))), a random access channel (Physical Random Access Channel (PRACH)) and the like can also be used in the wireless communication system 1.

[0370] User data, high-layer control information, system information block (SIB), etc. are transmitted through PDSCH. User data, high-layer control information, etc. can also be transmitted through PUSCH. In addition, master information block (MIB) can also be transmitted through PBCH.

[0371] The lower layer control information may also be transmitted via the PDCCH. The lower layer control information may include, for example, downlink control information (Downlink Control Information (DCI)) including scheduling information of at least one of the PDSCH and the PUSCH.

[0372] In addition, the DCI for scheduling the PDSCH may also be referred to as DL allocation, DL DCI, etc., and the DCI for scheduling the PUSCH may also be referred to as UL grant, UL DCI, etc. In addition, the PDSCH may also be rewritten as DL data, and the PUSCH may also be rewritten as UL data.

[0373] In the detection of PDCCH, the control resource set (CORESET) and the search space can also be used. CORESET corresponds to the resources for searching DCI. The search space corresponds to the search area and search method of PDCCH candidates. A CORESET can also be associated with one or more search spaces. The UE can also monitor the CORESET associated with a search space based on the search space setting.

[0374] A search space may also correspond to a PDCCH candidate corresponding to one or more aggregation levels. One or more search spaces may also be referred to as a search space set. In addition, the "search space", "search space set", "search space setting", "search space set setting", "CORESET", "CORESET setting" and the like in the present disclosure may also be rewritten mutually.

[0375] Uplink control information (uplink control information (UCI)) including at least one of channel state information (CSI), delivery confirmation information (e.g., also known as hybrid automatic repeat request confirmation (HARQ-ACK), ACK / NACK, etc.), and scheduling request (SR) can also be transmitted through PUCCH. The random access preamble used to establish a connection with a cell can also be transmitted through PRACH.

[0376] In the present disclosure, downlink, uplink, etc. may be expressed without the word “link.” In addition, various channels may be expressed without the word “physical” at the beginning.

[0377] In the wireless communication system 1, a synchronization signal (Synchronization Signal (SS)), a downlink reference signal (Downlink Reference Signal (DL-RS)), etc. can also be transmitted. As DL-RS, in the wireless communication system 1, a cell-specific reference signal (Cell-specific Reference Signal (CRS)), a channel state information reference signal (Channel State Information Reference Signal (CSI-RS)), a demodulation reference signal (DeModulation Reference Signal (DMRS)), a positioning reference signal (Positioning Reference Signal (PRS)), a phase tracking reference signal (Phase Tracking Reference Signal (PTRS)), etc. can also be transmitted.

[0378] The synchronization signal may be, for example, at least one of a primary synchronization signal (PSS) and a secondary synchronization signal (SSS). A signal block including SS (PSS, SSS) and PBCH (and DMRS for PBCH) may also be referred to as SS / PBCH block, SS block (SSB), etc. In addition, SS, SSB, etc. may also be referred to as reference signals.

[0379] In addition, in the wireless communication system 1, as an uplink reference signal (Uplink Reference Signal (UL-RS)), a measurement reference signal (Sounding Reference Signal (SRS)), a demodulation reference signal (DMRS), etc. can also be transmitted. In addition, DMRS can also be called a user terminal specific reference signal (UE-specific Reference Signal).

[0380] (Base Station)

[0381] Fig.19 1 is a diagram showing an example of a structure of a base station involved in one embodiment. The base station 10 includes a control unit 110, a transmitting and receiving unit 120, a transmitting and receiving antenna 130, and a transmission path interface (transmission line interface) 140. In addition, the control unit 110, the transmitting and receiving unit 120, the transmitting and receiving antenna 130, and the transmission path interface 140 may each be provided with more than one.

[0382] In addition, in this example, the functional blocks of the characteristic parts in this embodiment are mainly shown, and it is also conceivable that the base station 10 also has other functional blocks required for wireless communication. Part of the processing of each unit described below may also be omitted.

[0383] The control unit 110 controls the entire base station 10. The control unit 110 can be composed of a controller, a control circuit, and the like that are described based on common knowledge in the technical field to which the present disclosure relates.

[0384] The control unit 110 may also control signal generation, scheduling (e.g., resource allocation, mapping), etc. The control unit 110 may also control transmission and reception, measurement, etc. using the transmission and reception unit 120, the transmission and reception antenna 130, and the transmission path interface 140. The control unit 110 may also generate data, control information, sequences, etc. to be sent as signals, and forward them to the transmission and reception unit 120. The control unit 110 may also perform call processing (setting, release, etc.) of communication channels, state management of the base station 10, management of wireless resources, etc.

[0385] The transmitting and receiving unit 120 may also include a baseband unit 121, a radio frequency (RF) unit 122, and a measuring unit 123. The baseband unit 121 may also include a transmitting processing unit 1211 and a receiving processing unit 1212. The transmitting and receiving unit 120 may be composed of a transmitter / receiver, an RF circuit, a baseband circuit, a filter, a phase shifter (phase shifter), a measuring circuit, a transmitting and receiving circuit, etc., which are described based on the common knowledge in the technical field involved in the present disclosure.

[0386] The transmitting and receiving unit 120 may be configured as an integrated transmitting and receiving unit, or may be configured as a transmitting unit and a receiving unit. The transmitting unit may also be configured as a transmitting processing unit 1211 and an RF unit 122. The receiving unit may also be configured as a receiving processing unit 1212, an RF unit 122, and a measuring unit 123.

[0387] The transmitting / receiving antenna 130 can be constituted by an antenna described based on common knowledge in the technical field involved in the present disclosure, such as an array antenna.

[0388] The transmitting and receiving unit 120 may also transmit the above-mentioned downlink channel, synchronization signal, downlink reference signal, etc. The transmitting and receiving unit 120 may also receive the above-mentioned uplink channel, uplink reference signal, etc.

[0389] The transmitting and receiving unit 120 may also use digital beamforming (eg, precoding), analog beamforming (eg, phase rotation), etc. to form at least one of a transmitting beam and a receiving beam.

[0390] The sending and receiving unit 120 (the sending processing unit 1211 may also) performs processing on the Packet Data Convergence Protocol (PDCP) layer, the Radio Link Control (RLC) layer (for example, RLC retransmission control), the Medium Access Control (MAC) layer (for example, HARQ retransmission control), etc., for data and control information obtained from the control unit 110, and generates a bit string to be sent.

[0391] The transmitting and receiving unit 120 (transmitting processing unit 1211) can also perform channel coding (which may also include error correction coding), modulation, mapping, filter processing (filtering processing), discrete Fourier transform (Discrete Fourier Transform (DFT)) processing (as needed), inverse fast Fourier transform (Inverse Fast Fourier Transform (IFFT)) processing, precoding, digital-to-analog conversion and other transmission processing on the bit string to be sent, and output a baseband signal.

[0392] The transmitting and receiving unit 120 (RF unit 122 ) may also perform modulation, filter processing, amplification, etc. on the baseband signal to a radio frequency band, and transmit the signal in the radio frequency band via the transmitting and receiving antenna 130 .

[0393] On the other hand, the transmitting and receiving unit 120 (RF unit 122 ) may also perform amplification, filter processing, demodulation to a baseband signal, etc. on the signal in the radio frequency band received by the transmitting and receiving antenna 130 .

[0394] The sending and receiving unit 120 (receiving processing unit 1212) can also apply analog-to-digital conversion, fast Fourier transform (FFT) processing, inverse discrete Fourier transform (IDFT) processing (as needed), filter processing (filtering processing), demapping, demodulation, decoding (which may also include error correction decoding), MAC layer processing, RLC layer processing, and PDCP layer processing to the acquired baseband signal to obtain user data, etc.

[0395] The transmitting and receiving unit 120 (measuring unit 123) may also implement measurements related to the received signal. For example, the measuring unit 123 may also perform radio resource management (Radio Resource Management (RRM)) measurements, channel state information (CSI) measurements, etc. based on the received signal. The measuring unit 123 may also measure received power (e.g., reference signal received power (Reference Signal Received Power (RSRP))), received quality (e.g., reference signal received quality (Reference Signal Received Quality (RSRQ)), signal to interference plus noise ratio (Signal to Interference plus Noise Ratio (SINR)), signal to noise ratio (Signal to Noise Ratio (SNR))), signal strength (e.g., received signal strength indicator (Received Signal Strength Indicator (RSSI))), propagation path information (e.g., CSI), etc. The measurement results may also be output to the control unit 110.

[0396] The transmission path interface 140 can also send and receive signals (return signaling) between devices included in the core network 30, other base stations 10, etc., and can also obtain and transmit user data (user plane data) and control plane data for the user terminal 20.

[0397] In addition, the transmission unit and the reception unit of the base station 10 in the present disclosure may also be constituted by at least one of the transmission and reception unit 120 , the transmission and reception antenna 130 , and the transmission path interface 140 .

[0398] In addition, the control unit 110 may also train the reference model. The transmitting and receiving unit 120 may also send information related to the trained model and reference model information related to the reference model to the user terminal 20.

[0399] In addition, the transmitting and receiving unit 120 may also transmit reference model information related to the reference model or information related to a model trained based on the reference model to the user terminal 20. The transmitting and receiving unit 120 may also receive information related to a model further trained based on the reference model or the model.

[0400] In addition, the transmitting and receiving unit 120 may also transmit student model information related to the student model in the knowledge distillation to the user terminal 20. The transmitting and receiving unit 120 may also receive information related to the output based on the student model information.

[0401] (User Terminal)

[0402] Fig. 20 2 is a diagram showing an example of a configuration of a user terminal according to an embodiment. The user terminal 20 includes a control unit 210, a transmitting / receiving unit 220, and a transmitting / receiving antenna 230. In addition, the control unit 210, the transmitting / receiving unit 220, and the transmitting / receiving antenna 230 may be provided in one or more portions.

[0403] In addition, in this example, the functional blocks of the characteristic parts in this embodiment are mainly shown, and it is also conceivable that the user terminal 20 also has other functional blocks required for wireless communication. Part of the processing of each unit described below may be omitted.

[0404] The control unit 210 controls the entire user terminal 20. The control unit 210 can be composed of a controller, a control circuit, and the like that are described based on common knowledge in the technical field to which the present disclosure relates.

[0405] The control unit 210 may also control signal generation, mapping, etc. The control unit 210 may also control transmission and reception, measurement, etc. using the transmission and reception unit 220 and the transmission and reception antenna 230. The control unit 210 may also generate data, control information, sequences, etc. to be transmitted as signals, and forward them to the transmission and reception unit 220.

[0406] The transmitting and receiving unit 220 may also include a baseband unit 221, an RF unit 222, and a measuring unit 223. The baseband unit 221 may also include a transmitting processing unit 2211 and a receiving processing unit 2212. The transmitting and receiving unit 220 may be composed of a transmitter / receiver, an RF circuit, a baseband circuit, a filter, a phase shifter, a measuring circuit, a transmitting and receiving circuit, etc., which are described based on the common knowledge in the technical field involved in the present disclosure.

[0407] The transmitting and receiving unit 220 may be configured as an integrated transmitting and receiving unit, or may be configured as a transmitting unit and a receiving unit. The transmitting unit may also be configured as a transmitting processing unit 2211 and an RF unit 222. The receiving unit may also be configured as a receiving processing unit 2212, an RF unit 222, and a measuring unit 223.

[0408] The transmitting / receiving antenna 230 can be constituted by an antenna described based on common knowledge in the technical field involved in the present disclosure, such as an array antenna.

[0409] The transmitting and receiving unit 220 may also receive the above-mentioned downlink channel, synchronization signal, downlink reference signal, etc. The transmitting and receiving unit 220 may also transmit the above-mentioned uplink channel, uplink reference signal, etc.

[0410] The transmitting and receiving unit 220 may also use digital beamforming (eg, precoding), analog beamforming (eg, phase rotation), etc. to form at least one of a transmitting beam and a receiving beam.

[0411] The sending and receiving unit 220 (sending processing unit 2211) can also perform PDCP layer processing, RLC layer processing (for example, RLC retransmission control), MAC layer processing (for example, HARQ retransmission control), etc. on the data, control information, etc. obtained from the control unit 210 to generate a bit string to be sent.

[0412] The transmitting and receiving unit 220 (transmitting processing unit 2211) can also perform channel coding (which may also include error correction coding), modulation, mapping, filter processing (filtering processing), DFT processing (as needed), IFFT processing, precoding, digital-to-analog conversion and other transmission processing on the bit string to be sent, and output a baseband signal.

[0413] In addition, whether to apply DFT processing may also be based on the setting of transform precoding. For a certain channel (e.g., PUSCH), when transform precoding is valid (enabled), the transmitting and receiving unit 220 (transmitting processing unit 2211) may also perform DFT processing as the above-mentioned transmission processing in order to transmit the channel using a DFT-s-OFDM waveform. Otherwise, the transmitting and receiving unit 220 (transmitting processing unit 2211) may also perform DFT processing as the above-mentioned transmission processing without performing DFT processing.

[0414] The transmitting and receiving unit 220 (RF unit 222 ) may perform modulation, filter processing (filtering processing), amplification, etc. on the baseband signal to a radio frequency band, and transmit the signal in the radio frequency band via the transmitting and receiving antenna 230 .

[0415] On the other hand, the transmitting and receiving unit 220 (RF unit 222) may perform amplification, filter processing (filter processing), demodulation into a baseband signal, etc. on the signal in the radio frequency band received by the transmitting and receiving antenna 230.

[0416] The sending and receiving unit 220 (receiving processing unit 2212) can also apply analog-to-digital conversion, FFT processing, IDFT processing (as needed), filter processing (filtering processing), demapping, demodulation, decoding (which may also include error correction decoding), MAC layer processing, RLC layer processing, and PDCP layer processing to the obtained baseband signal to obtain user data, etc.

[0417] The transmitting and receiving unit 220 (measuring unit 223) may also perform measurements related to the received signal. For example, the measuring unit 223 may also perform RRM measurements, CSI measurements, etc. based on the received signal. The measuring unit 223 may also measure received power (e.g., RSRP), received quality (e.g., RSRQ, SINR, SNR), signal strength (e.g., RSSI), propagation path information (e.g., CSI), etc. The measurement results may also be output to the control unit 210.

[0418] In addition, the transmitting unit and the receiving unit of the user terminal 20 in the present disclosure may also be composed of at least one of the transmitting and receiving unit 220 and the transmitting and receiving antenna 230 .

[0419] In addition, the transmitting and receiving unit 220 may also receive NW training model information related to a model (NW training model) trained by a network (Network (NW)) and reference model information related to a reference model that is a basis for the NW training model. The control unit 210 may also determine the NW training model based on the NW training model information and the reference model information.

[0420] The NW training model information may also indicate a difference between the NW training model and the reference model, and the reference model information indicates an index related to the reference model.

[0421] The control unit 210 may also further train the NW training model.

[0422] Alternatively, the NW training model information includes information about parameters that should be frozen for training, and the control unit 210 further trains the NW training model, wherein the parameters are not updated.

[0423] In addition, the transmitting and receiving unit 220 may also receive NW training model information related to a model (NW training model) trained by a network (Network (NW)) or reference model information related to a reference model that is the basis of the NW training model. The control unit 210 may further train a model determined based on the NW training model information or the reference model information. The transmitting and receiving unit 220 may also send information related to the trained model.

[0424] The information about the trained model may also include information about which weights or layers were updated.

[0425] The information related to the trained model may also include update information of the global model in federated learning.

[0426] The sending and receiving unit 220 may also send information regarding resources used for the current iterative model training in federated learning.

[0427] In addition, the sending and receiving unit 220 may also receive student model information related to the student model in the knowledge distillation. The sending and receiving unit 220 may also send information related to the output based on the student model information.

[0428] The information about the output may also include information about the soft target of the student model.

[0429] The student model information may also include the value of the loss function.

[0430] The student model information may also include information of one or both of soft target loss and hard target loss.

[0431] (Hardware Structure)

[0432] In addition, the block diagram used in the description of the above-mentioned embodiment shows a block of a functional unit. These functional blocks (structural units) are implemented by any combination of at least one of hardware and software. In addition, the implementation method of each functional block is not particularly limited. That is, each functional block can be implemented by a device that is physically or logically combined, or two or more devices that are physically or logically separated can be directly or indirectly (for example, by wired, wireless, etc.) connected and implemented by these multiple devices. The functional block can also be implemented by combining the above-mentioned one device or the above-mentioned multiple devices with software.

[0433] Here, the functions include judging, deciding, determining, calculating, calculating, processing, deriving, investigating, searching, confirming, receiving, sending, outputting, accessing, solving, selecting, selecting, establishing, comparing, assuming, expecting, regarding, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, assigning, etc., but are not limited to these. For example, a functional block (structural unit) that implements a sending function may also be referred to as a transmitting unit, a transmitter, etc. Any one of them is as described above, and the implementation method thereof is not particularly limited.

[0434] For example, a base station, a user terminal, etc. in one embodiment of the present disclosure may also function as a computer that performs processing of the wireless communication method of the present disclosure. Fig.21 1001, memory 1002, storage 1003, communication device 1004, input device 1005, output device 1006, bus 1007, etc.

[0435] In addition, in the present disclosure, the terms such as device, circuit, equipment, section, unit, etc. can be interchangeably used. The hardware structure of the base station 10 and the user terminal 20 may be configured to include one or more of the devices shown in the figure, or may be configured to exclude some of the devices.

[0436] For example, only one processor 1001 is shown, but there may be multiple processors. In addition, the processing may be performed by one processor, or may be performed by two or more processors simultaneously, sequentially, or in other ways. In addition, the processor 1001 may also be implemented by one or more chips.

[0437] The functions of the base station 10 and the user terminal 20 are realized, for example, by reading specific software (program) into hardware such as the processor 1001 and the memory 1002, so that the processor 1001 performs calculations and controls communication via the communication device 1004, or controls at least one of the reading and writing of data in the memory 1002 and the storage 1003.

[0438] The processor 1001 controls the entire computer by, for example, operating an operating system. The processor 1001 may also be composed of a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic device, a register, etc. For example, at least a part of the above-mentioned control unit 110 (210), the sending and receiving unit 120 (220), etc. may also be implemented by the processor 1001.

[0439] In addition, the processor 1001 reads the program (program code), software module, data, etc. from at least one of the storage 1003 and the communication device 1004 to the memory 1002, and performs various processes based on them. As a program, a program that causes a computer to perform at least a part of the operations described in the above-mentioned embodiments can be used. For example, the control unit 110 (210) can also be implemented by a control program stored in the memory 1002 and operated in the processor 1001, and the same can be implemented for other functional blocks.

[0440] The memory 1002 may also be a computer-readable recording medium, for example, composed of at least one of a read-only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a random access memory (RAM), or other appropriate storage media. The memory 1002 may also be referred to as a register, a cache, a main memory (main storage device), etc. The memory 1002 can store executable programs (program codes), software modules, etc. for implementing the wireless communication method involved in one embodiment of the present disclosure.

[0441] The storage 1003 may also be a computer-readable recording medium, such as a flexible disk, a floppy disk, an optical disk (such as a compact disk (Compact Disc ROM (CD-ROM)), a digital versatile disk, a Blu-ray (Blu-ray) (registered trademark) disk), a removable disk, a hard disk drive, a smart card, a flash memory device (such as a card, a stick, a key drive), a magnetic stripe, a database, a server, or at least one of other suitable storage media. The storage 1003 may also be referred to as an auxiliary storage device.

[0442] The communication device 1004 is hardware (transmitting and receiving device) for communicating between computers via at least one of a wired network and a wireless network, and is also called a network device, a network controller, a network card, a communication module, etc. In order to realize at least one of frequency division duplex (Frequency Division Duplex (FDD)) and time division duplex (Time Division Duplex (TDD)), the communication device 1004 can also be configured to include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc. For example, the above-mentioned transmitting and receiving unit 120 (220), the transmitting and receiving antenna 130 (230), etc. can also be realized by the communication device 1004. The transmitting and receiving unit 120 (220) can also be realized by physically or logically separating the transmitting unit 120a (220a) and the receiving unit 120b (220b).

[0443] The input device 1005 is an input device that receives input from the outside (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.). The output device 1006 is an output device that implements output to the outside (e.g., a display, a speaker, a light emitting diode (LED) lamp, etc.). In addition, the input device 1005 and the output device 1006 may also be an integrated structure (e.g., a touch panel).

[0444] In addition, the processor 1001, the memory 1002 and other devices are connected via a bus 1007 for communicating information. The bus 1007 may be configured as a single bus or may be configured as different buses between the devices.

[0445] In addition, the base station 10 and the user terminal 20 may also be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), and a field programmable gate array (FPGA), and the hardware may be used to implement a part or all of each functional block. For example, the processor 1001 may also be implemented using at least one of these hardware.

[0446] (Variation Example)

[0447] In addition, the terms described in the present disclosure and the terms required for understanding the present disclosure may also be replaced with terms having the same or similar meanings. For example, channels, code elements, and signals (signals or signaling) may also be rewritten with each other. In addition, a signal may also be a message. A reference signal may also be referred to as RS, and may also be referred to as a pilot, a pilot signal, etc. depending on the standard applied. In addition, a component carrier (CC) may also be referred to as a cell, a frequency carrier, a carrier frequency, etc.

[0448] A wireless frame may also be composed of one or more periods (frames) in the time domain. Each period (frame) of the one or more periods (frames) constituting a wireless frame may also be referred to as a subframe. Further, a subframe may also be composed of one or more time slots in the time domain. A subframe may also be a fixed time length (e.g., 1 ms) that is not dependent on a parameter set (numerology).

[0449] Here, the parameter set may also be a communication parameter applied in at least one of the transmission and reception of a certain signal or channel. For example, the parameter set may also represent at least one of a subcarrier spacing (SCS), bandwidth, symbol length, cyclic prefix length, transmission time interval (TTI), the number of symbols per TTI, a wireless frame structure, a specific filter processing performed by a transmitter and receiver in the frequency domain, a specific windowing processing performed by a transmitter and receiver in the time domain, and the like.

[0450] A time slot may also be composed of one or more symbols (Orthogonal Frequency Division Multiplexing (OFDM) symbols, Single Carrier Frequency Division Multiple Access (SC-FDMA) symbols, etc.) in the time domain. In addition, a time slot may also be a time unit based on a parameter set.

[0451] A time slot may also include multiple mini-slots. Each mini-slot may also be composed of one or more symbols in the time domain. In addition, a mini-slot may also be referred to as a sub-slot. A mini-slot may also be composed of fewer symbols than a time slot. A PDSCH (or PUSCH) transmitted in a time unit larger than a mini-slot may also be referred to as a PDSCH (PUSCH) mapping type A. A PDSCH (or PUSCH) transmitted using a mini-slot may also be referred to as a PDSCH (PUSCH) mapping type B.

[0452] Radio frames, subframes, time slots, mini-time slots, and symbols all represent time units for transmitting signals. Radio frames, subframes, time slots, mini-time slots, and symbols may also be referred to by their respective names. In addition, time units such as frames, subframes, time slots, mini-time slots, and symbols in the present disclosure may also be replaced with each other.

[0453] For example, a subframe may be referred to as a TTI, a plurality of consecutive subframes may be referred to as a TTI, and a time slot or a mini time slot may be referred to as a TTI. That is, at least one of a subframe and a TTI may be a subframe (1 ms) in existing LTE, a period shorter than 1 ms (e.g., 1-13 symbols), or a period longer than 1 ms. In addition, the unit representing a TTI may be referred to as a time slot, a mini time slot, etc. instead of a subframe.

[0454] Here, TTI refers to, for example, the minimum time unit for scheduling in wireless communication. For example, in the LTE system, the base station schedules each user terminal to allocate wireless resources (frequency bandwidth, transmission power, etc. that can be used in each user terminal) in TTI units. In addition, the definition of TTI is not limited to this.

[0455] TTI may also be a transmission time unit for a data packet (transport block), a code block, a code word, etc. that has been channel-coded, and may also be a processing unit for scheduling, link adaptation, etc. In addition, when TTI is given, the time interval (e.g., the number of symbols) to which a transport block, a code block, a code word, etc. is actually mapped may be shorter than the TTI.

[0456] In addition, when a time slot or a mini time slot is called TTI, one or more TTIs (i.e., one or more time slots or one or more mini time slots) can also be the minimum time unit of scheduling. In addition, the number of time slots (mini time slots) constituting the minimum time unit of scheduling can also be controlled.

[0457] A TTI having a time length of 1 ms may also be referred to as a normal TTI (TTI in 3GPP Rel.8-12), a standard TTI, a long TTI, a normal subframe, a standard subframe, a long subframe, a time slot, etc. A TTI shorter than a normal TTI may also be referred to as a shortened TTI, a short TTI, a partial TTI (partial or fractional TTI), a shortened subframe, a short subframe, a mini time slot, a sub time slot, a time slot, etc.

[0458] In addition, a long TTI (e.g., normal TTI, subframe, etc.) can also be rewritten as a TTI with a time length exceeding 1 ms, and a short TTI (e.g., shortened TTI, etc.) can also be rewritten as a TTI with a TTI length shorter than that of the long TTI and longer than 1 ms.

[0459] A resource block (RB) is a resource allocation unit in the time domain and the frequency domain, and may include one or more consecutive subcarriers (subcarriers) in the frequency domain. The number of subcarriers included in an RB may also be the same regardless of the parameter set, for example, 12. The number of subcarriers included in an RB may also be determined based on the parameter set.

[0460] In addition, an RB may also include one or more symbols in the time domain, and may also be the length of a slot, a mini-slot, a subframe, or a TTI. A TTI, a subframe, etc. may also be composed of one or more resource blocks, respectively.

[0461] In addition, one or more RBs may also be referred to as a physical resource block (Physical RB (PRB)), a sub-carrier group (Sub-Carrier Group (SCG)), a resource element group (Resource Element Group (REG)), a PRB pair, an RB pair, etc.

[0462] In addition, a resource block may be composed of one or more resource elements (RE). For example, one RE may be a radio resource region of one subcarrier and one symbol.

[0463] A Bandwidth Part (BWP) (also referred to as a partial bandwidth, etc.) may also represent a subset of contiguous common RBs (common resource blocks) for a parameter set in a carrier. Here, a common RB may also be identified by an index of the RB relative to a common reference point of the carrier. PRBs may also be defined in a BWP and numbered within the BWP.

[0464] The BWP may include a UL BWP (BWP for UL) and a DL BWP (BWP for DL). For a UE, one or more BWPs may be configured in one carrier.

[0465] At least one of the set BWPs may be activated, and the UE may not assume that a specific signal / channel is transmitted or received outside the activated BWP. In addition, "cell", "carrier", etc. in the present disclosure may also be rewritten as "BWP".

[0466] In addition, the above-mentioned structures such as radio frames, subframes, time slots, mini-time slots and symbols are only examples. For example, the number of subframes included in a radio frame, the number of time slots in each subframe or radio frame, the number of mini-time slots included in a time slot, the number of symbols and RBs included in a time slot or mini-time slot, the number of subcarriers included in an RB, the number of symbols in a TTI, the symbol length, and the cyclic prefix (CP) length can be changed in various ways.

[0467] In addition, the information, parameters, etc. described in the present disclosure may be represented by absolute values, relative values ​​relative to a specific value, or other corresponding information. For example, wireless resources may also be indicated by a specific index.

[0468] In the present disclosure, the names used for parameters, etc. are not limiting in all respects. Furthermore, the mathematical formulas, etc. using these parameters may be different from those explicitly disclosed in the present disclosure. Various channels (PUCCH, PDCCH, etc.) and information elements can be identified by any suitable names, and therefore, the various names assigned to these various channels and information elements are not limiting in all respects.

[0469] Information, signals, etc. described in this disclosure may also be represented using any of a variety of different techniques. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be mentioned throughout the above description may also be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or photons, or any combination thereof.

[0470] Furthermore, information, signals, etc. can be output in at least one of the following directions: from a higher layer (upper layer) to a lower layer (lower layer), and from a lower layer to a higher layer. Information, signals, etc. can also be input and output via multiple network nodes.

[0471] The input and output information, signals, etc. may be stored in a specific location (e.g., a memory), or may be managed using a management table. The input and output information, signals, etc. may be overwritten, updated, or appended. The output information, signals, etc. may also be deleted. The input information, signals, etc. may also be sent to other devices.

[0472] The notification of information is not limited to the methods / implementations described in the present disclosure, and may also be performed by other methods. For example, the notification of information in the present disclosure may also be implemented through physical layer signaling (e.g., downlink control information (Downlink Control Information (DCI))), uplink control information (Uplink Control Information (UCI))), high-layer signaling (e.g., Radio Resource Control (RRC) signaling, broadcast information (Master Information Block (MIB)), System Information Block (SIB)), etc.), Medium Access Control (MAC) signaling), other signals or a combination thereof.

[0473] In addition, physical layer signaling may also be referred to as layer 1 / layer 2 (Layer 1 / Layer 2 (L1 / L2)) control information (L1 / L2 control signal), L1 control information (L1 control signal), etc. In addition, RRC signaling may also be referred to as an RRC message, such as an RRC connection establishment (RRC Connection Setup) message, an RRC connection reconstruction (RRC Connection Reconfiguration) message, etc. In addition, MAC signaling may also be notified using, for example, a MAC control element (MACControl Element (CE)).

[0474] Furthermore, notification of specific information (eg, notification of “it is X”) is not limited to explicit notification, but may be performed implicitly (eg, by not notifying the specific information or by notifying other information).

[0475] The determination can be made by a value represented by a bit (0 or 1), by a true or false value (Boolean value) represented by true or false, or by comparison of numerical values ​​(for example, comparison with a specific value).

[0476] Whether software is called software, firmware, middle-ware, microcode, hardware description language, or other names, it should be broadly interpreted as meaning instructions, instruction sets, code, code segments, program code, program, sub-program, software modules, applications, software applications, software packages, routines, sub-routines, objects, executable files, execution threads, procedures, functions, etc.

[0477] In addition, software, instructions, information, etc. may also be sent and received via a transmission medium. For example, when software is sent from a website, server, or other remote source using at least one of wired technology (coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), etc.) and wireless technology (infrared, microwave, etc.), at least one of these wired technology and wireless technology is included in the definition of transmission medium.

[0478] The terms "system" and "network" used in the present disclosure can be used interchangeably. "Network" may also refer to a device (eg, a base station) included in the network.

[0479] In the present disclosure, terms such as "precoding", "precoder", "weight (precoding weight)", "Quasi-Co-Location (QCL)", "Transmission Configuration Indication state (TCI state)", "spatial relation", "spatial domain filter", "transmission power", "phase rotation", "antenna port", "antenna port group", "layer", "number of layers", "rank", "resource", "resource set", "resource group", "beam", "beam width", "beam angle", "antenna", "antenna element", and "panel" can be used interchangeably.

[0480] In the present disclosure, terms such as "Base Station (BS)", "wireless base station", "fixed station", "NodeB", "eNB (eNodeB)", "gNB (gNodeB)", "access point", "Transmission Point (TP)", "Reception Point (RP)", "Transmission / Reception Point (TRP)", "panel", "cell", "sector", "cell group", "carrier", and "component carrier" can be used interchangeably. There are also cases where base stations are referred to by terms such as macro cell, small cell, micro-micro cell, and micro-micro cell.

[0481] A base station can accommodate one or more (for example, three) cells. When a base station accommodates multiple cells, the overall coverage area of ​​the base station can be divided into multiple smaller areas, and each smaller area can also provide communication services through a base station subsystem (for example, a small base station for indoor use (Remote Radio Head (RRH))). Terms such as "cell" or "sector" refer to a part or the entirety of the coverage area of ​​at least one of a base station and a base station subsystem that provides communication services within the coverage area.

[0482] In the present disclosure, the base station sending information to the terminal may also be rewritten as the base station instructing the terminal to control / operate based on the information.

[0483] In the present disclosure, terms such as “mobile station (MS)”, “user terminal”, “user device (User Equipment (UE))”, and “terminal” can be used interchangeably.

[0484] There are also cases where a mobile station is referred to as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, hand set, user agent, mobile client, client or several other appropriate terms.

[0485] At least one of the base station and the mobile station may also be referred to as a transmission device, a reception device, a wireless communication device, etc. In addition, at least one of the base station and the mobile station may be a device mounted on a moving object, a moving object body, etc.

[0486] The mobile body refers to a movable object, and the moving speed is arbitrary, and of course it also includes the situation where the mobile body stops. The mobile body includes, for example, vehicles, transport vehicles, cars, automatic two-wheeled vehicles (motorcycles), bicycles, connected cars, loading shovels, bulldozers, wheel loaders, dump trucks, forklifts, trains, buses, trolleys, rickshaws, ships (ships and other watercrafts), airplanes, rockets, artificial satellites, drones, multi-rotor aircraft, quadcopters, balloons and objects carried on them, but are not limited to these. In addition, the mobile body can also be a mobile body that drives autonomously based on operating instructions.

[0487] The mobile body may be a means of transportation (e.g., a vehicle, an airplane, etc.), a mobile body that moves unmanned (e.g., a drone, an autonomous vehicle, etc.), or a robot (manned or unmanned). In addition, at least one of the base station and the mobile station also includes a device that does not necessarily move when performing communication operations. For example, at least one of the base station and the mobile station may also be an Internet of Things (IoT) device such as a sensor.

[0488] Fig. 22 1 is a diagram showing an example of a vehicle according to an embodiment. The vehicle 40 includes a drive unit 41, a steering unit 42, an accelerator pedal 43, a brake pedal 44, a shift lever 45, left and right front wheels 46, left and right rear wheels 47, an axle 48, an electronic control unit 49, various sensors (including a current sensor 50, a rotation speed sensor 51, an air pressure sensor 52, a vehicle speed sensor 53, an acceleration sensor 54, an accelerator pedal sensor 55, a brake pedal sensor 56, a shift lever sensor 57, and an object detection sensor 58), an information service unit 59, and a communication module 60.

[0489] The driving unit 41 is composed of at least one of an engine, a motor, or a combination of an engine and a motor. The steering unit 42 is composed of at least a steering wheel (also called a handlebar), and steers at least one of the front wheels 46 and the rear wheels 47 based on the operation of the steering wheel operated by the user.

[0490] The electronic control unit 49 is composed of a microprocessor 61, a memory (ROM, RAM) 62, and a communication port (for example, an input / output (IO) port) 63. Signals from various sensors 50-58 provided in the vehicle are input to the electronic control unit 49. The electronic control unit 49 may also be referred to as an ECU (Electronic Control Unit).

[0491] Signals from various sensors 50-58 include a current signal from a current sensor 50 for sensing the current of a motor, a speed signal of the front wheels 46 / rear wheels 47 obtained by a speed sensor 51, an air pressure signal of the front wheels 46 / rear wheels 47 obtained by other sensors 52, a vehicle speed signal obtained by a vehicle speed sensor 53, an acceleration signal obtained by an acceleration sensor 54, a depression amount signal of an accelerator pedal 43 obtained by an accelerator pedal sensor 55, a depression amount signal of a brake pedal 44 obtained by a brake pedal sensor 56, an operation signal of a shift lever 45 obtained by a shift lever sensor 57, a detection signal for detecting obstacles, vehicles, pedestrians, etc. obtained by an object detection sensor 58, and the like.

[0492] The information service unit 59 is composed of various devices such as a vehicle navigation system, an audio system, a speaker, a display, a television, and a radio for providing (outputting) various information such as driving information, traffic information, and entertainment information, and one or more ECUs for controlling these devices. The information service unit 59 provides various information / services (for example, multimedia information / multimedia services) to the occupants of the vehicle 40 using information obtained from an external device via the communication module 60, etc.

[0493] The information service unit 59 may include input devices for accepting input from the outside (e.g., keyboard, mouse, microphone, switch, button, sensor, touch panel, etc.), and may also include output devices for implementing output to the outside (e.g., display, speaker, LED light, touch panel, etc.).

[0494] The driving assistance system unit 64 is composed of various devices such as millimeter wave radar, light detection and ranging (LiDAR), camera, positioning locator (e.g., Global Navigation Satellite System (GNSS)), map information (e.g., high-precision (High Definition (HD))) map, autonomous vehicle (Autonomous Vehicle (AV)) map, etc.), gyroscope system (e.g., inertial measurement unit (Inertial Measurement Unit (IMU))), inertial navigation unit (Inertial Navigation System (INS))), artificial intelligence (Artificial Intelligence (AI)) chip, AI processor, etc., for providing functions for preventing accidents before they happen or reducing the driver's driving load, and one or more ECUs for controlling these devices. In addition, the driving assistance system unit 64 sends and receives various information via the communication module 60 and realizes driving assistance function or autonomous driving function.

[0495] The communication module 60 can communicate with the microprocessor 61 and the components of the vehicle 40 via the communication port 63. For example, the communication module 60 transmits and receives data (information) between the drive unit 41, the steering unit 42, the accelerator pedal 43, the brake pedal 44, the shift lever 45, the left and right front wheels 46, the left and right rear wheels 47, the axle 48, the microprocessor 61 and the memory (ROM, RAM) 62 in the electronic control unit 49, and the various sensors 50-58 of the vehicle 40 via the communication port 63.

[0496] The communication module 60 can be controlled by the microprocessor 61 of the electronic control unit 49, and is a communication device that can communicate with an external device. For example, various information is sent and received with the external device via wireless communication. The communication module 60 can be located either inside or outside the electronic control unit 49. The external device can also be, for example, the above-mentioned base station 10, user terminal 20, etc. In addition, the communication module 60 can also be, for example, at least one of the above-mentioned base station 10 and user terminal 20 (it can also function as at least one of the base station 10 and user terminal 20).

[0497] The communication module 60 may also transmit at least one of the signals from the various sensors 50-58 input to the electronic control unit 49, information obtained based on the signals, and information based on input from the outside (user) obtained via the information service unit 59 to an external device via wireless communication. The electronic control unit 49, the various sensors 50-58, the information service unit 59, etc. may also be referred to as an input unit that receives input. For example, the PUSCH transmitted through the communication module 60 may also include information based on the above input.

[0498] The communication module 60 receives various information (traffic information, signal information, vehicle-to-vehicle information, etc.) transmitted from an external device, and displays it to the information service unit 59 provided in the vehicle. The information service unit 59 may also be referred to as an output unit for outputting information (for example, outputting information to a display, a speaker, etc. based on the PDSCH received by the communication module 60 (or data / information decoded from the PDSCH)).

[0499] In addition, the communication module 60 stores various information received from the external device in the memory 62 that can be used by the microprocessor 61. The microprocessor 61 can also control the drive unit 41, the steering unit 42, the accelerator pedal 43, the brake pedal 44, the shift lever 45, the left and right front wheels 46, the left and right rear wheels 47, the axle 48, and various sensors 50-58 of the vehicle 40 based on the information stored in the memory 62.

[0500] In addition, the base station in the present disclosure may also be rewritten as a user terminal. For example, the various methods / implementations of the present disclosure may also be applied to a structure in which the communication between a base station and a user terminal is replaced by the communication between multiple user terminals (for example, it may also be referred to as device-to-device (D2D)), vehicle-to-everything (V2X), etc.). In this case, it may also be set as a structure in which the user terminal 20 has the functions possessed by the above-mentioned base station 10. In addition, terms such as "uplink", "downlink", etc. may also be rewritten as terms corresponding to communication between terminals (for example, "sidelink"). For example, uplink channels, downlink channels, etc. may also be rewritten as sidelink channels.

[0501] Likewise, the user terminal in the present disclosure may be rewritten as a base station. In this case, the base station 10 may have the functions of the user terminal 20 described above.

[0502] In the present disclosure, the actions are assumed to be performed by the base station, and sometimes by its upper node (uppernode) depending on the situation. Obviously, in a network including one or more network nodes having a base station, various operations performed for communication with a terminal can be performed by the base station, one or more network nodes other than the base station (for example, considering the Mobility Management Entity (MME)), the Serving-Gateway (S-GW), etc., but not limited to these) or a combination thereof.

[0503] The various methods / implementations described in this disclosure may be used alone or in combination, and may be used in a switched manner as the method is executed. In addition, the processing procedures, timings, flow charts, etc. of the various methods / implementations described in this disclosure may be reversed in order as long as they are not contradictory. For example, for the method described in this disclosure, the elements of various steps are presented in an illustrative order, but are not limited to the specific order presented.

[0504] The various modes / implementations described in the present disclosure may also be applied to Long Term Evolution (LTE), LTE-Advanced (LTE-A), LTE-Beyond (LTE-B), SUPER 3G, IMT-Advanced, 4th generation mobile communication system (4G), 5th generation mobile communication system (5G), 6th generation mobile communication system (6G), xth generation mobile communication system (xG (x is, for example, an integer or a decimal)), Future Radio Access (FRA), New-Radio Access Technology (RAT), New Radio (NR), New Radio Access (NX), Future generation radio access (FX), Global System for Mobile communications (GSM (registered trademark)), CDMA2000, Ultra Mobile Broadband (UMB), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE 802.20, Ultra-WideBand (UWB), Bluetooth (registered trademark), systems using other appropriate wireless communication methods, and next-generation systems based on these enhancements, corrections, production or regulations, etc. In addition, multiple systems can also be combined (for example, LTE or LTE-A, combination with 5G, etc.) for application.

[0505] The phrase “based on” used in the present disclosure does not mean “based only on” unless otherwise specified. In other words, the phrase “based on” means both “based only on” and “based at least on”.

[0506] Any reference to an element using the terms "first", "second", etc. used in this disclosure does not fully define the amount or order of these elements. These terms can be used in this disclosure as a convenient method to distinguish between two or more elements. Therefore, reference to the first and second elements does not mean that only two elements can be used or that the first element must take precedence over the second element in some form.

[0507] The term "determining" used in the present disclosure may include a variety of actions. For example, "determining" may also refer to situations where judging, calculating, computing, processing, deriving, investigating, looking up (e.g., searching in a table, database or other data structure), ascertaining, etc. are considered to be "determining".

[0508] In addition, "judgment (decision)" may also be a situation where receiving (for example, receiving information), sending (for example, sending information), input (input), output (output), accessing (for example, accessing data in a memory), etc. are regarded as making a "judgment (decision)".

[0509] In addition, "judgment (decision)" can also be regarded as a situation where resolving, selecting, choosing, establishing, comparing, etc. are regarded as a situation where "judgment (decision)" is performed. That is, "judgment (decision)" can also be regarded as a situation where some actions are regarded as a situation where "judgment (decision)" is performed.

[0510] In addition, "judge (decide)" can also be rewritten as "assuming (assuming)", "expecting (expecting)", "considering (considering)" and so on.

[0511] The terms "connected", "coupled", or all their variations used in this disclosure refer to all direct or indirect connections or combinations between two or more elements, and may include the situation where one or more intermediate elements exist between two elements that are "connected" or "coupled" to each other. The combination or connection between elements may be physical, logical, or a combination thereof. For example, "connection" may also be rewritten as "access".

[0512] In the present disclosure, when two elements are connected, it is possible to consider being "connected" or "combined" to each other using one or more wires, cables, printed electrical connections, etc., as well as being "connected" or "combined" to each other using electromagnetic energy with wavelengths in the wireless frequency domain, microwave region, light (both visible and invisible) region, etc. as several non-limiting and non-inclusive examples.

[0513] In the present disclosure, the term "A is different from B" may also mean "A and B are different from each other". In addition, the term may also mean "A and B are different from C, respectively". The terms "separate" and "combined" may also be interpreted in the same manner as "different".

[0514] When the terms “include,” “including,” and variations thereof are used in the present disclosure, these terms have an inclusive meaning, similar to the term “comprising.” Furthermore, the term “or” used in the present disclosure does not have an exclusive OR meaning.

[0515] In the present disclosure, when an article is added by translation like a, an, and the in English, for example, the present disclosure may also include a case where the noun following the article is in plural form.

[0516] In the present disclosure, "below", "less than", "above", "more than", "equal to", etc. can also be replaced by each other. In addition, in the present disclosure, sentences meaning "good", "bad", "big", "small", "high", "low", "early", "late", etc. can also be replaced by each other (without limiting the original degree, comparative degree, and superlative degree). In addition, in the present disclosure, sentences meaning "good", "bad", "big", "small", "high", "low", "early", "late", etc. can also be replaced by each other as expressions with "ith" attached (without limiting the original degree, comparative degree, and superlative degree) (for example, "highest" can also be replaced by "ith highest").

[0517] In the present disclosure, “of,” “for,” “regarding,” “related to,” “associated with,” etc. may be rephrased.

[0518] The invention involved in the present disclosure is described in detail above, but it is obvious to those skilled in the art that the invention involved in the present disclosure is not limited to the embodiments described in the present disclosure. The invention involved in the present disclosure can be implemented as a modified and altered mode without departing from the gist and scope of the invention determined based on the description of the claims. Therefore, the description of the present disclosure is for the purpose of illustrative description and does not have any limiting meaning on the invention involved in the present disclosure.

Claims

1. A terminal having: A receiving unit: receiving NW training model information related to a model (NW training model) trained by a network (Network (NW)), or reference model information related to a reference model that is a basis for the NW training model; A control unit, further training a model determined based on the NW training model information or the reference model information; as well as A sending unit sends information related to the trained model.

2. The terminal according to claim 1, wherein: The information about the trained model includes information about which weights or layers are updated.

3. The terminal according to claim 1, wherein: The information related to the trained model includes update information of the global model in federated learning.

4. The terminal according to claim 1, wherein: The sending unit sends information about resources used for model training in a current iteration in federated learning.

5. A wireless communication method of a terminal, comprising: A step of receiving NW training model information related to a model (NW training model) trained by a network (Network (NW)) or reference model information related to a reference model that is a basis for the NW training model; A step of further training a model determined based on the NW training model information or the reference model information; as well as The step of sending information related to the trained model.

6. A base station, comprising: a sending unit, reference model information related to the reference model or information related to a model trained based on the reference model; and A receiving unit receives information related to a model further trained based on the reference model or the model.

Images