WO2025220811A1 - Method, base station, and user equipment for transmitting and receiving signals in wireless communication system - Google Patents

Abstract

A method and a device for transmitting and receiving signals in a wireless communication system are provided. The method and the device may comprise: training an artificial intelligence or machine learning-based model for channel estimation; transmitting, to a user equipment (UE), information associated with the training among a plurality of parameters of the model; and receiving channel estimation information based on the model from the UE, wherein the model is composed of a plurality of model structures, and each parameter included in the plurality of parameters is a parameter of each model structure included in the plurality of model structures.

Description

Method for transmitting and receiving signals in a wireless communication system, base station and user equipment

This specification relates to a wireless communication system, a method for transmitting and receiving signals, a base station, and user equipment.

As wireless communication technology advances, communication devices requiring high data rates are emerging and becoming widespread. As more communication devices demand greater capacity, the need for enhanced mobile broadband (eMBB), massive machine type communication (mMTC), and ultra-reliable and low latency communication (URLLC) is emerging.

Meanwhile, the introduction of artificial intelligence (AI) technology in wireless communication systems is expected to help support low-latency, low-complexity, and high-performance communication systems. To design communication systems that take into account latency-sensitive services and/or communication devices, AI-based communication technologies that achieve system optimization are key issues to consider in next-generation communications.

With the introduction of new wireless communication technologies, the amount of data and/or control information transmitted and received by communication devices is increasing. Since the amount of radio resources available for communication is finite, new methods are needed to enable base stations and/or user devices in wireless communication systems to efficiently utilize radio resources to transmit and receive data and/or control information.

The technical tasks that this specification aims to achieve are not limited to the technical tasks mentioned above, and other technical tasks that are not mentioned will be clearly understood by those skilled in the art related to this specification from the detailed description below.

In one aspect of the present disclosure, a method for transmitting and receiving signals by a base station in a wireless communication system is provided. The method includes: training an artificial intelligence or machine learning-based model for channel estimation, transmitting information related to the training among a plurality of parameters of the model to a user equipment (UE), and receiving channel estimation information based on the model from the UE, wherein the model is composed of a plurality of model structures, and each parameter included in the plurality of parameters is a parameter of each model structure included in the plurality of model structures.

In another aspect of the present disclosure, a base station for transmitting and receiving signals in a wireless communication system is provided. The base station includes: at least one transceiver; at least one processor; and at least one computer memory operably connected to the at least one processor and storing instructions that, when executed, cause the at least one processor to perform operations. The operations include: training an artificial intelligence or machine learning-based model for channel estimation, transmitting information related to the training among a plurality of parameters of the model to a user equipment (UE), and receiving channel estimation information based on the model from the UE, wherein the model comprises a plurality of model structures, and each parameter included in the plurality of parameters is a parameter of each model structure included in the plurality of model structures.

In another aspect of the present specification, a method for a user equipment to transmit and receive signals in a wireless communication system is provided. The method includes: determining an artificial intelligence or machine learning-based model for channel estimation between a base station and a user equipment based on a model identifier (ID), receiving information related to training corresponding to the model ID from the base station, and transmitting channel estimation information using a model reflecting the information related to the training to the base station, wherein the model is composed of a plurality of model structures, and each parameter included in the plurality of parameters is a parameter of each model structure included in the plurality of model structures.

In another aspect of the present disclosure, a user equipment (UE) for transmitting and receiving signals in a wireless communication system is provided. The UE includes: at least one transceiver; at least one processor; and at least one computer memory operably connectable to the at least one processor and storing instructions that, when executed, cause the at least one processor to perform operations. The operations include: determining an artificial intelligence or machine learning-based model for channel estimation with a base station based on a model identifier (ID), receiving information associated with training corresponding to the model ID from the base station, and transmitting channel estimation information using a model reflecting the information associated with the training to the base station, wherein the model is composed of a plurality of model structures, and each parameter included in the plurality of parameters is a parameter of each model structure included in the plurality of model structures.

In each aspect of the present specification, the information associated with the training may include index information based on a predetermined codebook for the plurality of parameters.

In each aspect of this specification, training the model may be performed for the UE or a group of UEs including the UE.

Alternatively, training the model may include receiving information from the UE or a UE other than the UE that trained the model.

In each aspect of the present specification, training the model may include determining a model identifier (ID) of one or more models supported by the UE, and performing training on a model corresponding to the determined model ID.

In each aspect of the present specification, transmitting the information associated with the training to the UE may be performed via radio resource control (RRC) signaling, and may include configuring the information associated with the training within the transmission space size of the RRC signaling.

Meanwhile, the transmission space of the RRC signaling can be set to a size of 45 kByte or more.

In addition, after transmitting information related to the training through the RRC signaling, the method may further include transmitting additional information to the UE through downlink control information (DCI).

Additionally, after transmitting information associated with the training via the RRC signaling, the model may be configured to reflect additional information acquired from the UE.

In each aspect of this specification, transmitting information related to the training to the UE may be performed via user data, but may be performed periodically or based on a request from the UE.

In each aspect of this specification, parameters of the plurality of model structures may be transmitted to the UE in different ways.

Meanwhile, the parameters of the plurality of model structures may be transmitted in different ways, i) a first parameter of a first model structure among the plurality of model structures may be transmitted via radio resource control (RRC) signaling, and a second parameter of a second model structure among the plurality of model structures may be transmitted via user data, or ii) the first parameter may be transmitted via RRC signaling, and the second parameter may be trained in the UE.

The above problem solving methods are only some of the examples of this specification, and various examples reflecting the technical features of this specification can be derived and understood by a person having ordinary knowledge in the relevant technical field based on the detailed description below.

According to an implementation(s) of this specification, a method for transmitting and receiving an artificial intelligence (AI) or machine learning (ML) based model in a wireless communication system is provided.

According to the implementation(s) of this specification, wireless communication signals can be transmitted and received efficiently. Accordingly, the overall throughput of the wireless communication system can be increased.

According to implementation(s) of this specification, delay/latency occurring during wireless communication can be reduced.

The effects obtained from the present invention may not be limited to the effects described above. Furthermore, other effects not mentioned can be clearly understood by those skilled in the art from the following description.

The accompanying drawings, which are included as part of the detailed description to aid in understanding implementations of this specification, provide examples of implementations of this specification and, together with the detailed description, serve to illustrate implementations of this specification.

Figure 1 illustrates the structure of a system for 5th generation (5G) communication.

FIG. 2 is a block diagram illustrating examples of communication devices capable of performing a method according to the present specification.

FIG. 3 illustrates a wireless communication system to which some implementations of the present specification are applied.

Figure 4 illustrates a channel in wireless communication.

Figure 5 illustrates a flow of a signal transmission and reception method according to some implementations of this specification.

Figure 6 illustrates another flow of a signal transmission and reception method according to some implementations of the present specification.

Figure 7 illustrates the structure of wireless communication based on deep learning (DL).

Figure 8 illustrates another structure of wireless communication based on deep learning (DL).

FIGS. 9 through 11 illustrate examples of a flow of a method for transmitting and receiving an artificial intelligence (AI) and/or machine learning (ML) based model according to some implementations of the present specification.

FIGS. 12 and 13 illustrate further examples of a flow of a method for transmitting and receiving AI and/or ML-based models according to some implementations of the present specification.

Hereinafter, implementations according to this specification will be described in detail with reference to the attached drawings. The detailed description provided below, together with the attached drawings, is intended to describe exemplary implementations of this specification and is not intended to represent the only possible implementations of this specification. The detailed description below includes specific details to provide a thorough understanding of this specification. However, one of ordinary skill in the art will appreciate that this specification may be practiced without these specific details.

In some cases, to avoid ambiguity in the concepts of this specification, known structures and devices may be omitted or illustrated in block diagram form focusing on the core functions of each structure and device. Throughout this specification, identical components are described using the same reference numerals.

The terms and words used in the following description and drawings should not be interpreted as limited to their conventional or dictionary meanings, but should be interpreted with meanings and concepts that conform to the technical idea of this specification based on the principle that the inventor can appropriately define the concept of the term to best describe his or her invention. Therefore, the examples described in this specification and the configurations illustrated in the drawings are merely the most preferred embodiments of this specification and do not represent all of the technical idea of this specification. Therefore, it should be understood that there may be various equivalents and modified examples that can replace them at the time of filing this application.

Additionally, unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and shall not be interpreted in an idealized or overly formal sense unless explicitly defined herein.

The terminology used in this specification is only used to describe specific embodiments and is not intended to limit the features, components, etc. described in the specification. The singular expression includes plural expressions unless the context clearly indicates otherwise. It should be understood that the terms "comprises" or "has" described in this specification are intended to specify the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

Additionally, in various embodiments of the present specification, "/" and "," should be interpreted as indicating "and/or". For example, "A/B" can mean "A and/or B". Furthermore, "A, B" can mean "A and/or B". Furthermore, "A/B/C" can mean "at least one of A, B, and/or C". Furthermore, "A, B, C" can mean "at least one of A, B, and/or C".

Additionally, terms that include ordinal numbers, such as "first," "second," etc., are used to describe various components and are only used to distinguish one component from another, not to limit said components. For example, without exceeding the scope of this specification, a second component could be referred to as a first component, and similarly, a first component could also be referred to as a second component.

The techniques, devices, and systems described below can be applied to various wireless multiple access systems. Examples of multiple access systems include code division multiple access (CDMA) systems, frequency division multiple access (FDMA) systems, time division multiple access (TDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, and single carrier frequency division multiple access (SC-FDMA) systems. CDMA can be implemented in radio technologies such as Universal Terrestrial Radio Access (UTRA) or CDMA2000. TDMA can be implemented in radio technologies such as Global System for Mobile communications (GSM)/General Packet Radio Service (GPRS)/Enhanced Data Rates for GSM Evolution (EDGE). OFDMA can be implemented in wireless technologies such as IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802-20, and Evolved UTRA (E-UTRA). UTRA is part of UMTS (Universal Mobile Telecommunications System), and 3GPP (3rd Generation Partnership Project) LTE (long term evolution) is part of E-UMTS (Evolved UMTS) that uses E-UTRA. 3GPP LTE adopts OFDMA in the downlink (DL) and SC-FDMA in the uplink (UL). LTE-A (Advanced) is an evolved form of 3GPP LTE, and 3GPP NR (New Radio or New Radio Access Technology) is an evolved form of 3GPP LTE/LTE-A.

For convenience of explanation, the following description assumes that this specification applies to 3GPP-based communication systems, such as LTE and NR. However, the technical features of this specification are not limited to this. For example, although the detailed description below is based on a mobile communication system corresponding to a 3GPP LTE/NR system, it can also be applied to any other mobile communication system, except for features specific to 3GPP LTE/NR.

For terms and technologies used in this specification that are not specifically explained, reference can be made to 3GPP-based standard documents.

In the examples of this specification described below, the expression "assumes" that a device "assumes" that the entity transmitting the channel transmits the channel in a manner consistent with the "assume." The entity receiving the channel may mean that, under the assumption that the channel was transmitted in a manner consistent with the "assume," the entity receiving the channel receives or decodes the channel in a manner consistent with the "assume."

In this specification, user equipment (UE) may be fixed or mobile, and includes various devices that communicate with a base station (BS) to transmit and/or receive user data and/or various control information. UE may be called Terminal Equipment (TE), Mobile Station (MS), Mobile Terminal (MT), User Terminal (UT), Subscriber Station (SS), wireless device, Personal Digital Assistant (PDA), wireless modem, handheld device, etc. In addition, in this specification, BS generally refers to a fixed station that communicates with UE and/or other BS, and exchanges various data and control information with UE and other BS. BS may be called by other terms such as Advanced Base Station (ABS), Node-B (NB), evolved-NodeB (eNB), Base Transceiver System (BTS), Access Point, and Processing Server (PS). In particular, the BS in UTRAN is called a Node-B, the BS in E-UTRAN is called an eNB, and the BS in a new radio access technology network is called a gNB. For convenience of explanation, the base station is referred to as a BS below, regardless of the type or version of communication technology.

Figure 1 illustrates the structure of a system for 5th generation (5G) communication.

Specifically, it illustrates the structure of a 5G communication system to which the method(s) described in FIGS. 5 and 6 are applied.

Referring to FIG. 1, a next generation radio access network (NG-RAN) may include a BS (20) that provides user plane and control plane protocol termination to a UE (10).

The example of Fig. 1 illustrates a case including only gNB. BSs (20) can be connected to each other via Xn interfaces. BSs (20) can be connected to a 5th generation core network (5GC) via an NG interface. Specifically, BSs (20) can be connected to an access and mobility management function (AMF) (30) via an NG-C interface, and can be connected to a user plane function (UPF) (30) via an NG-U interface.

According to some implementations of this specification, the functions, procedures, and/or methods described in this specification may be performed via the AMF/UPF (30). For example, according to some implementations of this specification, the operation(s) performed by the BS (20) may be processed by the AMF/UPF (30) (rather than the BS (20)).

Meanwhile, the UE (10) and/or BS (20) of FIG. 1 may correspond to the wireless device illustrated in FIG. 2.

In a wireless communication system, a UE (10) performs beam management by transmitting feedback information through uplink control information (UCI) transmitted through a physical uplink control channel (PUCCH) and/or a physical uplink shared channel (PUSCH) for a beam transmitted by a BS (20). The UCI information transmitted through the uplink may include a rank indicator (RI), a layer indicator, a channel quality indicator, a channel state information reference signal (CSI-RS) resource indicator (CRI), etc.

Meanwhile, in the next generation communication system of 5G communication, the necessity of beam management and methods for efficiently performing beam management are being discussed.

This specification proposes a method for transmitting and receiving signals by a UE (10) and/or a BS (20) using an artificial intelligence (AI) and/or machine learning (ML)-based model (hereinafter, referred to as an AI/ML model). For example, a method for transmitting and receiving an AI/ML model for channel estimation is proposed.

FIG. 2 is a block diagram illustrating examples of communication devices capable of performing a method according to the present specification.

Specifically, FIG. 2 illustrates an example of a communication device(s) that transmits and receives signals according to the method(s) described in FIGS. 5 to 6.

Referring to FIG. 2, the first wireless device (100) and the second wireless device (200) can transmit and/or receive wireless signals via various wireless access technologies (e.g., LTE, NR). Here, {the first wireless device (100), the second wireless device (200)} can correspond to {UE, BS}.

A first wireless device (100) includes one or more processors (102) and one or more memories (104), and may further include one or more transceivers (106) and/or one or more antennas (108). The processor(s) (102) controls the memories (104) and/or the transceivers (106), and may be configured to implement the functions, procedures, and/or methods described/proposed above. For example, the processor(s) (102) may process information in the memories (104) to generate first information/signals, and then transmit a wireless signal including the first information/signals via the transceivers (106). In addition, the processor(s) (102) may receive a wireless signal including second information/signal through the transceiver(s) (106), and then store information obtained from signal processing of the second information/signal in the memory(s) (104). The memory(s) (104) may be connected to the processor(s) (102) and may store various information related to the operation of the processor(s) (102). For example, the memory(s) (104) may perform some or all of the processes controlled by the processor(s) (102), or store software code including instructions for performing the procedures and/or methods described/proposed above. Here, the processor(s) (102) and the memory(s) (104) may be part of a communication modem/circuit/chip designed to implement a wireless communication technology (e.g., LTE, NR). The transceiver(s) (106) may be connected to the processor(s) (102) and may transmit and/or receive wireless signals via one or more antennas (108). The transceiver(s) (106) may include a transmitter and/or a receiver. The transceiver(s) (106) may be used interchangeably with an RF (Radio Frequency) unit. In this specification, a wireless device may also mean a communication modem/circuit/chip.

The second wireless device (200) includes one or more processors (202), one or more memories (204), and may further include one or more transceivers (206) and/or one or more antennas (208). The processor(s) (202) controls the memories (204) and/or the transceivers (206), and may be configured to implement the functions, procedures, and/or methods described/proposed above. For example, the processor(s) (202) may process information in the memories (204) to generate third information/signals, and then transmit a wireless signal including the third information/signals via the transceivers (206). In addition, the processor(s) (202) may receive a wireless signal including the fourth information/signal through the transceiver(s) (206), and then store information obtained from signal processing of the fourth information/signal in the memory(s) (204). The memory(s) (204) may be connected to the processor(s) (202) and may store various information related to the operation of the processor(s) (202). For example, the memory(s) (204) may perform some or all of the processes controlled by the processor(s) (202), or store software code including instructions for performing the procedures and/or methods described/proposed above. Here, the processor(s) (202) and the memory(s) (204) may be part of a communication modem/circuit/chip designed to implement a wireless communication technology (e.g., LTE, NR). The transceiver(s) (206) may be connected to the processor(s) (202) and may transmit and/or receive wireless signals via one or more antennas (208). The transceiver(s) (206) may include a transmitter and/or a receiver. The transceiver(s) (206) may be used interchangeably with an RF unit. In this specification, a wireless device may also mean a communication modem/circuit/chip.

Hereinafter, the hardware elements of the device (100, 200) will be described in more detail. Although not limited thereto, one or more protocol layers may be implemented by one or more processors (102, 202). For example, one or more processors (102, 202) may implement one or more layers (e.g., functional layers such as a physical (PHY) layer, a medium access control (MAC) layer, a radio link control (RLC) layer, a packet data convergence protocol (PDCP) layer, a radio resource control (RRC) layer, and a service data adaptation protocol (SDAP) layer). One or more processors (102, 202) may generate one or more protocol data units (PDUs) and/or one or more service data units (SDUs) according to the functions, procedures, proposals, and/or methods disclosed in this specification. One or more processors (102, 202) may generate messages, control information, data or information according to the functions, procedures, proposals and/or methods disclosed in this specification. One or more processors (102, 202) may generate signals (e.g., baseband signals) including PDUs, SDUs, messages, control information, data or information according to the functions, procedures, proposals and/or methods disclosed in this specification, and provide the signals to one or more transceivers (106, 206). One or more processors (102, 202) may receive signals (e.g., baseband signals) from one or more transceivers (106, 206) and obtain PDUs, SDUs, messages, control information, data or information according to the functions, procedures, proposals and/or methods disclosed in this specification.

One or more processors (102, 202) may be referred to as a controller, a microcontroller, a microprocessor, or a microcomputer. One or more processors (102, 202) may be implemented by hardware, firmware, software, or a combination thereof. For example, one or more Application Specific Integrated Circuits (ASICs), one or more Digital Signal Processors (DSPs), one or more Digital Signal Processing Devices (DSPDs), one or more Programmable Logic Devices (PLDs), or one or more Field Programmable Gate Arrays (FPGAs) may be included in one or more processors (102, 202). The functions, procedures, proposals, and/or methods disclosed in this specification may be implemented using firmware or software, and the firmware or software may be implemented to include modules, procedures, functions, etc. Firmware or software configured to perform the functions, procedures, suggestions and/or methods disclosed in this specification may be included in one or more processors (102, 202) or stored in one or more memories (104, 204) and executed by one or more processors (102, 202). The functions, procedures, suggestions and/or methods disclosed in this specification may be implemented using firmware or software in the form of codes, instructions and/or sets of instructions.

The processor(s) (102, 202) may perform the method(s) and/or procedure(s) according to the present specification. For example, the processor(s) (102, 202) may train and/or retrain an AI/ML model. In another example, the processor(s) (102, 202) may transmit and receive an AI/ML model via the transceiver(s) (106, 206).

One or more memories (104, 204) may be coupled to one or more processors (102, 202) and may store various forms of data, signals, messages, information, programs, codes, instructions, and/or commands. The one or more memories (104, 204) may be configured as ROM, RAM, EPROM, flash memory, hard drives, registers, cache memory, computer-readable storage media, and/or combinations thereof. The one or more memories (104, 204) may be located internally and/or externally to the one or more processors (102, 202). Additionally, the one or more memories (104, 204) may be coupled to the one or more processors (102, 202) via various technologies, such as wired or wireless connections.

One or more transceivers (106, 206) may transmit user data, control information, wireless signals/channels, etc., as described in the methods and/or flowcharts according to some implementations of this specification to one or more other devices. One or more transceivers (106, 206) may receive user data, control information, wireless signals/channels, etc., as described in the functions, procedures, proposals, methods and/or flowcharts disclosed in this specification from one or more other devices. For example, one or more transceivers (106, 206) may be coupled to one or more processors (102, 202) and may transmit and/or receive wireless signals. For example, one or more processors (102, 202) may control one or more transceivers (106, 206) to transmit user data, control information, or wireless signals to one or more other devices. Additionally, one or more processors (102, 202) may control one or more transceivers (106, 206) to receive user data, control information, or wireless signals from one or more other devices. Additionally, one or more transceivers (106, 206) may be coupled to one or more antennas (108, 208), and one or more transceivers (106, 206) may be configured to transmit and/or receive user data, control information, wireless signals/channels, or the like, as referred to in the functions, procedures, proposals, methods, and/or operational flowcharts disclosed in this specification, via one or more antennas (108, 208). In this specification, one or more antennas may be multiple physical antennas or multiple logical antennas (e.g., antenna ports). One or more transceivers (106, 206) may convert received user data, control information, wireless signals/channels, etc. from RF band signals to baseband signals in order to process the received user data, control information, wireless signals/channels, etc. using one or more processors (102, 202). One or more transceivers (106, 206) may convert processed user data, control information, wireless signals/channels, etc. from baseband signals to RF band signals using one or more processors (102, 202). For this purpose, one or more transceivers (106, 206) may include an (analog) oscillator and/or a filter.

FIG. 3 illustrates a wireless communication system to which some implementations of the present specification are applied.

Referring to FIG. 3, a wireless communication system may include one or more BSs and one or more UEs. With the introduction of new wireless communication technologies, the amount of data and control information transmitted and received by the BSs and/or UEs is increasing. The introduction of AI technology in wireless communication systems provides enhanced wireless communication systems by enabling system optimization. AI technology overcomes problems arising from modeling issues, implementation complexity, and/or signal distortion/interference of wireless communication system configurations, thereby supporting low-latency, low-complexity, and high-performance communication systems. For example, AI technology can determine appropriate representations for problems that are difficult to model in wireless communication system configurations. For another example, an improved AI-based wireless communication system can reduce implementation complexity by finding an ideal and computationally feasible solution, or can optimize modem parameters. As another example, improved wireless communication systems based on AI can solve the non-linearity problem of existing wireless communication through non-linear function modeling.

In particular, this specification proposes a method for transmitting and receiving wireless communication signals based on such AI. Specifically, it proposes a method for transmitting and receiving an AI/ML model and transmitting and receiving wireless communication signals using the AI/ML model. For example, according to some implementations of this specification, a BS and/or UE can transmit and receive an AI/ML model for channel estimation (CE).

Figure 4 illustrates a channel in wireless communication.

Specifically, the devices/devices shown in FIGS. 1 and 2 illustrate channels that are transmitted and received and/or estimated according to some implementations of the present specification.

In wireless communications, a channel can be represented as an image in the subcarrier and time domains. Referring to Figure 4, channel elements in wireless communications are highly correlated in the spatial, temporal, and/or frequency domains. Deep learning (DL), which is powerful for image-related tasks, can efficiently process these channel elements. For example, convolutional neural networks (CNNs) have the potential to exploit correlations between adjacent channel elements in the spatial, temporal, and/or frequency domains. In particular, DL-based channel estimation methods can improve channel estimation performance in extreme environments. For example, in cases where pilots are scarce and/or nonlinear distortions exist, DL can be utilized to implement nonlinear filters for channel estimation.

Channel estimation according to some implementations of this specification may differ from CSI feedback enhancement, beam management, and/or positioning accuracy enhancement in the following respects: In time division duplexing (TDD) mode, the UE-side AI/ML model and the network-side AI/ML model may be the same due to channel reciprocity. When the UE-side AI/ML model and the network-side AI/ML model are the same, there is an advantage in that there is no need to train the AI/ML model of either the UE or the network while training the AI/ML model of the other. For example, the BS can train the AI/ML model, transmit/forward the trained AI/ML model to the UE, and the UE can use the received/forwarded AI/ML model for channel estimation. The opposite is also possible. For example, the UE can train the AI/ML model, transmit/forward the trained AI/ML model to the BS, and the BS can use the received/forwarded AI/ML model for channel estimation. In this way, when one side trains an AI/ML model and the other side does not train an AI/ML model, it is called offline learning. However, offline learning is not limited to the fact that the UE-side AI/ML model and the network-side AI/ML model are the same. For example, the UE can receive an AI/ML model trained from the BS and additionally train the AI/ML model based on the AI/ML model received from the BS. In this way, (re)training the AI/ML model on the other side that received the AI/ML model is called online learning. In this specification, offline/online learning can also be referred to as offline/online training.

According to some implementations of this specification, the BS can train AI/ML model(s) and transmit the AI/ML model(s) to the UE, and the UE can transmit and receive wireless communication signals with the BS based on the AI/ML model received from the BS. Furthermore, according to some implementations of this specification, the UE can (re)train the AI/ML model received from the BS and transmit feedback information including information related to the (re)training to the BS. That is, according to some implementations of this specification, both offline learning and/or online learning can be performed.

In this specification, transmitting and receiving an AI/ML model may mean transmitting and receiving parameter(s) of the AI/ML model, information about the parameter(s), and/or information related to (re)training the AI/ML model. For example, an AI/ML model may be determined by a BS and a UE, wherein the AI/ML model has a structure known to the BS and the UE, and one of the BS and the UE may train the AI/ML model and transmit parameter(s) of the AI/ML model and/or information related to training among the parameters to the other.

Below, we specifically describe how to transmit and receive signals according to some implementations of this specification.

Figure 5 illustrates a flow of a signal transmission and reception method according to some implementations of this specification.

Specifically, FIG. 5 illustrates a flow of a method in which the devices/devices shown in FIGS. 1 and 2 transmit and receive AI/ML models.

Referring to FIG. 5, the BS can train AI/ML model(s) (S510), transmit information related to the training to the UE (S520), and receive feedback information based on the AI/ML model(s) in which the information related to the training is reflected from the UE (S530). Here, the information related to the training may refer to information about the parameter(s) of the AI/ML model. For example, the BS can train an AI/ML model for channel estimation with the UE, transmit the parameter(s) of the AI/ML model and/or codebook information about the parameter(s) of the AI/ML model to the UE, and receive channel estimation information based on the AI/ML model from the UE.

The UE can determine AI/ML model(s) based on the model ID, receive training-related information corresponding to the model ID from the BS (S520), and transmit feedback information to the BS based on the model reflecting the training-related information (S530). For example, the UE can determine an AI/ML model for channel estimation with the BS based on the model ID, receive parameters of the AI/ML model corresponding to the model ID from the BS, and transmit channel estimation information using the AI/ML model reflecting the parameters to the BS.

Additionally, BS can determine AI/ML model(s) based on model identifier (ID).

In S510, BS may train AI/ML model(s) by determining the AI/ML model based on the model ID and training the model corresponding to the determined model ID.

Additionally, in S510, AI/ML model training may be performed by the BS for a UE or a group of UEs including UE(s).

However, in the case of channel estimation, there may be an advantage in that the AI/ML model is trained at the UE level due to the reciprocity of the channel and information about it is transmitted to the BS without affecting performance. As in the embodiment described above, it is possible for the BS to train the AL/ML models of all UEs individually, but in another embodiment of the present specification, in order to reduce the training burden on the BS, the AL/ML model may be trained by training the AI/ML model at some or all UE(s) and receiving information about it from the BS.

In S520, information related to training may be transmitted via RRC signaling and/or downlink control information (DCI).

Additionally, after S520, the AI/ML model can be configured to reflect additional information acquired from the UE. A UE that has received training-related information can estimate additional information (e.g., codebook index, model parameters, etc.) and reflect it in the AI/ML model.

Figure 6 illustrates another flow of a signal transmission and reception method according to some implementations of the present specification.

Specifically, FIG. 6 illustrates a flow of a method in which the devices/devices illustrated in FIGS. 1 and 2 transmit and receive an AI/ML model composed of one or more model structures.

Referring to FIG. 6, the BS can train AI/ML model(s) composed of one or more model structures (Model #1, Model #2, ...) (S610), transmit information related to the training to the UE (S620a), and receive feedback information based on the AI/ML model(s) in which the information related to the training is reflected from the UE (S630). Here, the information related to the training may mean information about parameter(s) of the AI/ML model, and the parameter(s) may mean parameter(s) for a certain model structure.

The UE can determine AI/ML model(s) composed of one or more model structures based on a model ID, receive information related to training corresponding to the model ID(s) from the BS (S620a), and transmit feedback information to the BS based on a model reflecting the information related to training (S630).

Additionally, the BS can determine AI/ML model(s) based on the model ID. Furthermore, AI/ML model(s) composed of one or more model structures can be determined based on one or more model IDs. That is, one or more model structures of an AI/ML model can be determined based on one or more model IDs. Here, the model ID corresponding to the model structure can be distinguished from the model ID corresponding to the AI/ML model, and can also be referred to as a sequential ID. Hereinafter, the model ID may be interpreted as a model ID that serves as the basis for determining an AI/ML model or a model ID that serves as the basis for determining a model structure, depending on some implementations of this specification.

In S610, BS may train AI/ML model(s) by determining an AI/ML model based on one model ID and one or more model IDs, and training a model corresponding to the determined model ID.

In S620a, training-related information may be transmitted via RRC signaling and/or downlink control information (DCI). Furthermore, parameters for each model structure may be transmitted in different ways. For example, the BS may transmit parameter(s) for one model structure to the UE via RRC signaling, and parameter(s) for another model structure to the UE via DCI signaling.

Additionally, after S620a, the AI/ML model can be configured to reflect additional information acquired from the UE. A UE that has received training-related information can estimate additional information (e.g., codebook index, model parameters, etc.) and reflect it in the AI/ML model.

S610 to S630 of FIG. 6 may correspond to S510 to S530 of FIG. 5.

According to some implementations of this specification, a method for transmitting and receiving an AI/ML-based model comprising one or more model structures in a wireless communication system is provided. This allows a BS and/or UE to efficiently transmit and receive wireless communication signals. For example, according to some implementations of this specification, a BS and/or UE transmitting and receiving an AI/ML model can reduce implementation complexity, computational load, and time delay when performing channel estimation based on the AI/ML model.

Therefore, the overall throughput of a wireless communication system can be increased and the delay/latency occurring during wireless communication can be reduced through the AI/ML model transmission/reception method according to this specification.

Figure 7 illustrates the structure of wireless communication based on deep learning (DL).

Specifically, FIG. 7 is an example of the structure of a DL-based wireless communication system to which the method(s) described in FIGS. 5 to 6 are applied.

Referring to FIG. 7, a communication system can be designed block-by-block. A DL-based block-structured communication system can be composed of multiple blocks to partition signal processing. For example, a DL-based block-structured communication system can be composed of a channel encoding/decoding block (702), a modulation/demodulation block (704), a radio frequency (RF) transceiver (706), a channel estimation block (708), and a signal detection block (710). Here, each block(s) can be a processing block(s) individually/independently optimized based on DL for stable communication according to some implementations of the present specification.

In particular, the channel estimation block (708) may perform channel state information (CSI) estimation and/or direction of arrival (DOA) estimation according to some implementations of the present specification to improve the wireless communication environment. For example, according to some implementations of the present specification, the channel estimation block (708) may perform AI/ML model training and/or AI/ML model transmission/reception (having multiple model structures), and perform CSI estimation, DOA estimation, etc. based on the AI/ML model, thereby improving the implementation complexity and the performance of the wireless communication system for a small number of pilot symbols.

Figure 8 illustrates another structure of wireless communication based on deep learning (DL).

Specifically, FIG. 8 is an example of the structure of a DL-based wireless communication system to which the method(s) described in FIGS. 5 to 6 are applied.

Block-structured communication systems allow for individual/independent optimization of blocks, but do not guarantee overall system performance. As an alternative to such block-structured communication systems, the DL-based end-to-end communication system illustrated in Figure 8 is trained based on DL, enabling optimization of the entire system.

Some implementations of this specification are applicable to DL-based end-to-end communication systems, depending on their embodiment.

Below, the operation of transmitting and receiving an AI/ML model among the method(s) according to the implementation of this specification is specifically described.

FIGS. 9 through 11 illustrate examples of a flow of a method for transmitting and receiving an artificial intelligence (AI) and/or machine learning (ML) based model according to some implementations of the present specification.

Specifically, FIGS. 9 to 11 are examples of a method for transmitting and receiving an AI/ML model based on the method(s) described in FIGS. 5 to 6.

According to some implementations of this specification, the BS and/or UE can transmit and receive AI/ML models. In a wireless communication system, having the UE receive a trained model and/or model parameter(s) from the BS is advantageous in terms of learning overhead, storage space, etc., compared to having the UE train a model and transmit it to the BS.

This specification describes, but is not limited to, a method for transmitting an AI/ML model relative to a BS. According to some implementations of this specification, the operations of the BS may be performed by the UE, and vice versa. For example, according to some implementations of this specification, model(s)/model parameter(s) trained at the BS and transmitted/transferred from the BS to the UE may be trained at the UE and transmitted/transferred from the UE to the BS.

In some scenarios for channel estimation, the BS may train an AI/ML model for a UE and/or a group of UEs for channel reciprocity, and transmit/forward the AI/ML model to the UE via RRC signaling.

Here, based on the existing number of RRC segments, a problem may arise where model sizes exceeding a certain size are not supported. For example, models larger than 45 kBytes cannot be transmitted/transmitted via RRC signaling based on the existing number of RRC segments. However, this problem can be resolved using the proposed method(s) described below.

1. Proposal 1

(1) Proposal 1-(1)

Referring to Figure 9, the BS and UE can determine an AI/ML model through a model ID (S910). If the BS-side model structure and the UE-side model structure differ, signaling overhead may increase when transmitting and receiving the model. Therefore, the BS and/or UE can determine an AI/ML model through a model ID to transmit and receive a model composed of a mutually known model structure.

Additionally, the BS can transmit/transmit parameter(s) of the AI/ML model corresponding to the model ID to the UE via RRC signaling (S920).

(2) Proposal 1-(2)

Referring to FIG. 10, an AI/ML model can be determined through a model ID in BS and UE (S1010).

Additionally, the BS may transmit/deliver a codebook for the parameter(s) of the AI/ML model corresponding to the model ID to the UE via RRC signaling (S1020). Here, the codebook for the parameter(s) may be dynamically changed. For example, the codebook may be dynamically changed based on the channel environment as the channel environment changes. The size of the codebook may be determined based on the model ID for the AI/ML model.

Additionally, the BS may transmit/deliver the codebook index to the UE via DCI (S1030a), or the UE may estimate the codebook index (S1030b). For example, the UE may estimate the codebook index based on the received signal-to-noise ratio (SNR).

Additionally, the UE can reflect the estimated parameter(s) based on the estimated codebook index into the AI/ML model.

(3) Proposal 1-(3)

Referring to FIG. 11, the AI/ML model can be determined through the model ID in the BS and UE (S1110).

In addition, the BS may transmit/deliver a predefined codebook index for parameter(s) of the AI/ML model corresponding to the model ID to the UE via RRC signaling and/or DCI (S1120a), or the UE may estimate the predefined codebook index (S1120b). Here, the codebook for the parameter(s) of the AI/ML model corresponding to the model ID is predefined in the form of a table, and the BS and/or the UE may transmit/deliver or estimate the index of the predefined codebook. Here, the predefined codebook may need to be semi-statically changed according to changes in the channel environment.

Additionally, the UE can reflect the estimated parameter(s) based on the estimated codebook index into the AI/ML model.

According to Proposal 1-(1) to Proposal 1-(3), transmitting/transmitting information about the parameter(s) of an AI/ML model (e.g., parameter(s), codebook and/or codebook index, etc.) via RRC signaling may be done by configuring the information about the parameter(s) within the transmission space size of the RRC signaling and transmitting/transmitting it. In addition, the transmission space of the RRC signaling may be set to a size of 45 kBytes or more.

2. Proposal 2

In some scenarios for channel estimation, the BS may train an AI/ML model for a UE and/or a group of UEs for channel reciprocity, and transmit/forward the AI/ML model to the UE via user-plane data.

(1) Proposal 2-(1)

AI/ML models can be determined through model ID in BS and UE.

Additionally, the BS may transmit parameter(s) of the AI/ML model corresponding to the model ID to the UE via user plane data. Here, the BS may periodically transmit parameter(s) of the AI/ML model to the UE via user plane data. Alternatively, the BS may transmit parameter(s) of the AI/ML model to the UE via user plane data based on a request from the UE.

(2) Proposal 2-(2)

AI/ML models can be determined through model ID in BS and UE.

Additionally, the BS can transmit/transmit a codebook for the parameter(s) of the AI/ML model corresponding to the model ID to the UE via user plane data. Here, the codebook for the parameter(s) can be dynamically changed. For example, the codebook can be dynamically changed based on channel conditions as the channel environment changes. The size of the codebook can be determined based on the model ID for the AI/ML model.

Additionally, the BS can transmit/transmit the codebook index to the UE via DCI, or the UE can estimate the codebook index. For example, the UE can estimate the codebook index based on the received SNR, etc.

Additionally, the UE can reflect the estimated parameter(s) based on the estimated codebook index into the AI/ML model.

FIGS. 12 and 13 illustrate further examples of a flow of a method for transmitting and receiving AI and/or ML-based models according to some implementations of the present specification.

Specifically, FIGS. 12 and 13 are examples of a method for transmitting and receiving an AI/ML model composed of one or more model structures based on the method(s) described in FIGS. 5 and 6.

3. Proposal 3

AI/ML models can be configured with dual-model and/or multiple-model architectures. Models with dual-model/multi-model architectures can effectively handle various channel characteristics (e.g., CSI reporting, beamforming, positioning, etc.).

The structure of an AI/ML model can be distinguished/determined based on the features of the input/output data. For example, an AI/ML model can be composed of a model structure (θ ₁ ) for stable features (e.g., channel features in the angle/spatial domain) and a model structure (θ ₂ ) for unstable/dynamic features (e.g., channel features in the frequency/time domain).

(1) Proposal 3-(1)

Referring to FIG. 12, the BS and UE can determine an AI/ML model through multiple model IDs and/or a single model ID (S1210). Here, determining an AI/ML model may mean determining a dual/multiple AI/ML model structure through multiple model IDs and/or a single model ID.

Additionally, the BS may transmit/transmit parameters for a dual/multiple model structure corresponding to the model ID to the UE via RRC signaling and/or user-plane data (S1220a). The parameters for the dual/multiple model structures may be transmitted in different ways (S1220a). For example, parameter θ ₁ for one model structure may be transmitted/transmitted via RRC signaling, and parameter θ ₂ for another model structure may be transmitted/transmitted via DCI.

(2) Proposal 3-(2)

Referring to FIG. 13, the BS and UE can determine an AI/ML model through multiple model IDs and/or a single model ID (S1310). Here, determining an AI/ML model may mean determining a dual/multiple AI/ML model structure through multiple model IDs and/or a single model ID.

In addition, the BS transmits/transmits parameter(s) for a dual/multiple model structure corresponding to a model ID to the UE via RRC signaling and/or user plane data (S1320a), and the UE can train other parameter(s) that are not transmitted/transmitted (S1330). For example, parameter θ ₁ for one model structure may be transmitted/transmitted by the BS to the UE via RRC signaling, and parameter θ ₂ for another model structure may be trained in the UE. In another example, parameter(s) for one model structure may be transmitted/transmitted by the BS to the UE via RRC signaling, parameter(s) for another model structure may be transmitted/transmitted by the BS to the UE via DCI, and parameter(s) for the remaining other model structure(s) may be trained in the UE.

Training the parameter(s) in the UE may mean that the UE performs online learning to (additionally) train an AI/ML model based on the parameter(s). In other words, the UE may receive parameter(s) for a dual/multi-model structure from the BS, and (additionally) (re)train an AI/ML model that reflects the received parameter(s) based on the parameter(s) for another dual/multi-model structure that has not received the received parameter(s).

Here, the parameter(s) transmitted/delivered to the UE may be parameter(s) for a model structure for stable features, and the parameter(s) trained in the UE may be parameter(s) for a model structure for unstable/dynamic features. For example, the BS may transmit/delivery parameter(s) for a model structure for stable features to the UE, and the UE may perform online training of an AI/ML model for unstable/dynamic features.

In addition, the BS may (additionally) receive feedback from the UE regarding the parameter(s) corresponding to the model ID. For example, in S1330, the UE may train an AI/ML model online based on the parameter (θ ₂ ) of the AI/ML model structure corresponding to the model ID, and then transmit/forward feedback related to the corresponding parameter (θ ₂ ) (including training-related information of the online-trained AI/ML model) to the BS.

Additionally, parameter(s) for other model structure(s) that are not transmitted/delivered to the UE may be trained at the BS.

The methods according to Proposals 1 to 3 are examples for transmitting and receiving an AI/ML model according to some implementations of the present specification, and are not limited thereto, and the AI/ML model may be transmitted and received by a method combining each of the methods included in Proposals 1 to 3. For example, the AI/ML model may be configured with a dual/multi-model structure, and the BS and/or the UE may transmit/transmit parameters of each of the dual/multi-model structures to the UE and/or the BS according to a predetermined cycle. As another example, the AI/ML model may be configured with a dual/multi-model structure, and the BS and/or the UE may transmit/transmit parameters of each of the dual/multi-model structures based on a request from the UE and/or the BS.

In this specification, a computer-readable storage medium can store at least one instruction or computer program, which when executed by at least one processor can cause the at least one processor to perform operations according to some embodiments or implementations of this specification.

In this specification, a computer program or computer program product may be recorded on at least one computer-readable (non-volatile) storage medium and may contain instructions that, when executed, cause (at least one processor) to perform operations according to some embodiments or implementations of this specification.

In this specification, a processing device or apparatus may include at least one processor and at least one computer memory connectable to the at least one processor. The at least one computer memory may store instructions or programs, which, when executed, cause at least one processor operably connected to the at least one memory to perform operations according to some embodiments or implementations of the present specification.

As described above, the examples disclosed in this specification are provided to enable those skilled in the relevant technical fields to implement and practice this specification. While the examples of this specification have been described above with reference to the examples, those skilled in the relevant technical fields can modify and adapt the examples of this specification in various ways. For example, those skilled in the art can utilize the individual components described in the examples of this specification in combination with each other.

Claims (26)

In a method for a base station to transmit and receive signals in a wireless communication system, Train an artificial intelligence or machine learning-based model for channel estimation, Transmitting information related to the training among the plurality of parameters of the above model to a user equipment (UE), and Including receiving channel estimation information based on the model from the UE, The above model is composed of multiple model structures, Each parameter included in the above plurality of parameters is a parameter of each model structure included in the above plurality of model structures. Method of transmitting and receiving signals. In the first paragraph, The information associated with the above training includes index information based on a predetermined codebook for the plurality of parameters. Method of transmitting and receiving signals. In the first paragraph, Training the above model is: For the UE or a UE group including the UE, Method of transmitting and receiving signals. In the first paragraph, Training the above model is: Including receiving information on training the model from the UE or a UE other than the UE, Method of transmitting and receiving signals. In the first paragraph, Training the above model is: Determine the model identifier (ID) of one or more models supported by the UE, Including performing training on a model corresponding to a determined model ID, Method of transmitting and receiving signals. In the first paragraph, Transmitting information related to the above training to the UE, Performed through radio resource control (RRC) signaling, including configuring information associated with the training within the transmission space size of the RRC signaling. Method of transmitting and receiving signals. In paragraph 6, The transmission space of the above RRC signaling is set to a size of 45 kByte or more. Method of transmitting and receiving signals. In paragraph 6, After transmitting information related to the training through the RRC signaling, further comprising transmitting additional information to the UE through downlink control information (DCI). Method of transmitting and receiving signals. In paragraph 6, After transmitting the information related to the training through the RRC signaling, the model is configured to reflect additional information obtained from the UE. Method of transmitting and receiving signals. In the first paragraph, Transmitting information related to the above training to the UE, Performed through user data, but performed periodically or based on a request from the UE. Method of transmitting and receiving signals. In the first paragraph, The parameters of the above plurality of model structures are transmitted to the UE in different ways. Method of transmitting and receiving signals. In paragraph 11, The parameters of the above multiple model structures are transmitted in different ways, i) a first parameter of a first model structure among the plurality of model structures is transmitted via radio resource control (RRC) signaling, and a second parameter of a second model structure among the plurality of model structures is transmitted via user data, or ii) the first parameter is transmitted via RRC signaling, and the second parameter is trained in the UE. Method of transmitting and receiving signals. In a base station that transmits and receives signals in a wireless communication system, At least one transmitter/receiver; at least one processor; and At least one computer memory operably connected to said at least one processor and storing instructions that, when executed, cause said at least one processor to perform operations, said operations comprising: Train an artificial intelligence or machine learning-based model for channel estimation, Transmitting information related to the training among the plurality of parameters of the above model to a user equipment (UE), and Including receiving channel estimation information based on the model from the UE, The above model is composed of multiple model structures, Each parameter included in the above plurality of parameters is a parameter of each model structure included in the above plurality of model structures. Base station. In paragraph 13, The information associated with the above training includes index information based on a predetermined codebook for the plurality of parameters. Base station. In paragraph 13, Training the above model is: For the UE or a UE group including the UE, Base station. In paragraph 15, Training the above model is: Including receiving information on training the model from the UE or a UE other than the UE, Base station. In paragraph 13, Training the above model is: Determine the model identifier (ID) of one or more models supported by the UE, Including performing training on a model corresponding to a determined model ID, Base station. In paragraph 13, Transmitting information related to the above training to the UE, Performed through radio resource control (RRC) signaling, including configuring information associated with the training within the transmission space size of the RRC signaling. Base station. In paragraph 18, The transmission space of the above RRC signaling is set to a size of 45 kByte or more. Base station. In paragraph 18, The above actions are: After transmitting information related to the training through the RRC signaling, further comprising transmitting additional information to the UE through downlink control information (DCI). Base station. In paragraph 18, After transmitting the information related to the training through the RRC signaling, the model is configured to reflect additional information obtained from the UE. Base station. In paragraph 13, Transmitting information related to the above training to the UE, Performed through user data, but performed periodically or based on a request from the UE. Base station. In paragraph 13, The parameters of the above plurality of model structures are transmitted to the UE in different ways. Base station. In paragraph 23, The parameters of the above multiple model structures are transmitted in different ways, i) a first parameter of a first model structure among the plurality of model structures is transmitted via radio resource control (RRC) signaling, and a second parameter of a second model structure among the plurality of model structures is transmitted via user data, or ii) the first parameter is transmitted via RRC signaling, and the second parameter is trained in the UE. Base station. In a method for a user device to transmit and receive signals in a wireless communication system, Determine the artificial intelligence or machine learning-based model for base station and channel estimation based on the model identifier (ID), Receive information related to training corresponding to the above model ID from the base station, and Including transmitting channel estimation information to the base station through a model reflecting information related to the above training, The above model is composed of multiple model structures, Each parameter included in the above plurality of parameters is a parameter of each model structure included in the above plurality of model structures. Method of transmitting and receiving signals. In a user device that transmits and receives signals in a wireless communication system, At least one transmitter/receiver; at least one processor; and At least one computer memory operably connectable to said at least one processor and storing instructions that, when executed, cause said at least one processor to perform operations, said operations comprising: Determine the artificial intelligence or machine learning-based model for base station and channel estimation based on the model identifier (ID), Receive information related to training corresponding to the above model ID from the base station, and Including transmitting channel estimation information to the base station through a model reflecting information related to the above training, The above model is composed of multiple model structures, Each parameter included in the above plurality of parameters is a parameter of each model structure included in the above plurality of model structures. User device.