Method and apparatus for reporting of channel state information in wireless communication system

Abstract

A method of a UE according to one embodiment of the present disclosure may include: receiving a first message from a base station including an AI / ML CSI reporting setting, wherein the AI / ML CSI reporting setting is associated with one ML monitoring reporting setting; receiving an RS set based on instructions from the base station to activate the AI / ML CSI reporting setting; obtaining a CSI prediction value by using a measurement value of the received RS as an input to an ML model; and transmitting a second message generated based on the CSI prediction value to the base station.

Description

Technology Field

[0001] The present disclosure relates to a technology for reporting channel state information in a wireless communication system, and more specifically, to a technology for reporting channel state information based on Artificial Intelligence (AI) / Machine Learning (ML). Background Technology

[0002] As wireless communication technology advances, the 5th generation (5 th The commercialization of wireless communication systems (generation, 5G) is underway. In wireless communication technology, communication between a base station and a UE takes place over a specific channel; therefore, the base station must be able to know the channel state information (CSI) with the UE in order to perform proper communication. In particular, in wireless communication systems that use multiple antennas to form a specific beam, channel state information can be a very important factor in communication between the base station and the UE.

[0003] Meanwhile, with the recent advancement of Artificial Intelligence (AI) and Machine Learning (ML), various studies on utilizing AI / ML in wireless communication systems are actively underway. In particular, discussions are being held on methods to provide faster and higher quality services by using AI / ML to estimate (or predict) the CSI between a base station and a UE, and by using the estimated (or predicted) CSI to perform communication.

[0004] However, there is currently no proposed method for reporting predicted CSI using AI / ML in wireless communication systems. Furthermore, monitoring of the performance of AI / ML models must be performed to determine whether the performance of AI / ML models running on the UE is maintained at an appropriate level or whether configuration changes are necessary. Methods and devices for such performance monitoring are required. The problem to be solved

[0005] The objective of the present disclosure to address the above-mentioned requirements is to provide a method for monitoring the performance of an AI / ML model in a wireless communication system. means of solving the problem

[0006] A method of user equipment (UE) according to one embodiment of the disclosure for achieving the above-mentioned purpose may include the steps of: receiving a first message from a base station including an Artificial Intelligence (AI) / Machine Learning (ML) channel state information (CSI) reporting setting, wherein the AI / ML CSI reporting setting is associated with one ML monitoring reporting setting; receiving a reference signal (RS) set based on the activation of the AI / ML CSI reporting setting from the base station; obtaining a CSI prediction value by using a measurement value of the received RS as an input to an ML model; and transmitting a second message generated based on the CSI prediction value to the base station, wherein the second message further includes ML monitoring information based on the ML monitoring reporting setting associated with the AI / ML CSI reporting setting, and the ML monitoring information may be determined based on the CSI prediction value and / or the measurement value of the RS.

[0007] The above ML monitoring report settings may include one or more of the transmission type of RS for monitoring the ML model, the transmission resource of RS, or the reporting quantity.

[0008] The above reported quantity includes prediction accuracy information, and the prediction accuracy information can be determined based on the predicted value for a first time instance using the ML model and the value of the RS measured at the first time point.

[0009] The activation of the above AI / ML CSI reporting setting may be indicated by a medium access control-control element (MAC-CE) message.

[0010] The activation of the above AI / ML CSI reporting settings can be indicated by downlink control information (DCI).

[0011] The activation of the above AI / ML CSI reporting setting can be indicated by a combination of a medium access control-control element (MAC-CE) message and downlink control information (DCI).

[0012] A method of user equipment (UE) according to one embodiment of the disclosure for achieving the above-mentioned purpose may include: receiving a first message from a base station comprising ML monitoring report settings associated with an Artificial Intelligence (AI) / Machine Learning (ML) channel state information (CSI) reporting setting; receiving a reference signal (RS) from the base station based on an ML monitoring report setting indicated by the first ML monitoring report setting among the plurality of ML monitoring settings; obtaining a CSI prediction value by using a measurement value of the received RS as an input to an ML model; and transmitting a second message generated based on the CSI prediction value to the base station, wherein the second message further comprises ML monitoring information based on the ML monitoring report setting associated with the AI / ML CSI reporting setting, and the ML monitoring information may be determined based on the CSI prediction value and / or the measurement value of the RS.

[0013] Each of the above plurality of ML monitoring report settings may include one or more of the transmission type of RS for monitoring the ML model, the transmission resource of the RS, or the reporting quantity.

[0014] The above reported quantity includes prediction accuracy information, and the prediction accuracy information can be determined based on the predicted value for a first time instance using the ML model and the value of the RS measured at the first time point.

[0015] The above-mentioned first ML monitoring report setting may be an ML monitoring setting that is instructed to be activated among a plurality of ML monitoring report settings included in a medium access control-control element (MAC-CE) message.

[0016] The above-mentioned first ML monitoring report setting may be an ML monitoring setting indicated by a first indicator included in downlink control information (DCI).

[0017] The first ML monitoring report setting can be indicated by a first indicator included in downlink control information (DCI) scrambled with a first radio network temporary identifier (RNTI).

[0018] A first ML monitoring report setting is indicated by a combination of a medium access control-control element (MAC-CE) message and downlink control information (DCI), the MAC-CE message indicates the activation of one or more ML monitoring report settings among a plurality of ML monitoring report settings included in the first message, and a first field of the DCI may indicate the first ML monitoring report setting among the one or more ML monitoring report settings whose activation is indicated by the MAC-CE.

[0019] User equipment (UE) according to one embodiment of the disclosure for achieving the above-mentioned purpose comprises at least one processor, wherein the at least one processor may cause the UE to: receive a first message from a base station comprising ML monitoring report settings associated with an Artificial Intelligence (AI) / Machine Learning (ML) channel state information (CSI) reporting setting; receive a reference signal (RS) from the base station based on the indication of an ML monitoring report by the first ML monitoring report setting among the plurality of ML monitoring settings; obtain a CSI prediction value by using a measurement value of the received RS as an input to an ML model; and transmit a second message generated based on the CSI prediction value to the base station, wherein the second message may further include ML monitoring information based on the ML monitoring report setting associated with the AI / ML CSI reporting setting, and the ML monitoring information may be determined based on the CSI prediction value and / or the measurement value of the RS.

[0020] Each of the above plurality of ML monitoring report settings may include one or more of the transmission type of RS for monitoring the ML model, the transmission resource of the RS, or the reporting quantity.

[0021] The above reported quantity includes prediction accuracy information, and the prediction accuracy information can be determined based on the predicted value for a first time instance using the ML model and the value of the RS measured at the first time point.

[0022] The above-mentioned first monitoring report setting may be an ML monitoring setting that is instructed to be activated among a plurality of ML monitoring settings included in a medium access control-control element (MAC-CE) message.

[0023] The above-mentioned first ML monitoring report setting may be an ML monitoring setting indicated by a first indicator included in downlink control information (DCI).

[0024] The first ML monitoring report setting can be indicated by a first indicator included in downlink control information (DCI) scrambled with a first radio network temporary identifier (RNTI).

[0025] A first ML monitoring report setting may be indicated by a combination of a medium access control-control element (MAC-CE) message and downlink control information (DCI), wherein the MAC-CE message indicates the activation of one or more ML monitoring report settings among a plurality of ML monitoring report settings included in the first message, and the first field of the DCI may indicate the first ML monitoring report setting among the one or more ML monitoring report settings whose activation is indicated by the MAC-CE. Effects of the invention

[0026] According to one embodiment of the present disclosure, a UE can switch to an AI / ML CSI reporting operation based on the configuration and / or triggering of a base station during a legacy CSI reporting operation. Additionally, the UE can switch to a legacy CSI reporting operation based on the configuration and / or triggering of a base station during an AI / ML CSI reporting operation. This allows the UE to perform a legacy CSI reporting procedure or an AI / ML CSI reporting procedure as needed.

[0027] Additionally, according to another embodiment of the present disclosure, a procedure for monitoring the performance of an AI / ML model may be provided. In the present disclosure, the performance monitoring of an AI / ML model may be initiated by a base station or by a UE. Additionally, a procedure for monitoring the performance of an AI / ML model may be provided by defining an upper layer configuration for the performance monitoring of an AI / ML model and a triggering signal and / or message for the performance monitoring of an AI / ML model. In particular, the present disclosure has the advantage that the procedure for monitoring the performance of an AI / ML model can be performed while minimizing changes to the current 5G NR specifications. Brief explanation of the drawing

[0028] FIG. 1 is a conceptual diagram illustrating an embodiment of a communication system. FIG. 2 is a block diagram illustrating an example of a communication node constituting a communication system. Figure 3a is a flowchart of a UE performing a CSI report to a base station based on a periodic CSI reporting method in a mobile communication system. Figure 3b is a flowchart of a UE performing CSI reporting to a base station based on a semi-static or non-periodic CSI reporting method in a mobile communication system. Figure 4 is a flowchart illustrating the transition from a periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system. Figure 5 is a flowchart illustrating the transition from a semi-static or non-periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system. Figure 6 is a flowchart illustrating the transition from a periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system. Figure 7 is a flowchart illustrating the transition from a semi-static or non-periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system. Figure 8 is a flowchart illustrating the monitoring method of an AI / ML model starting from a base station. FIG. 9 is a conceptual diagram illustrating the configuration of a MAC-CE for triggering AI / ML monitoring according to the 2-1 embodiment of the present disclosure. FIG. 10 is a flowchart illustrating a first embodiment of a monitoring method for an AI / ML model starting from a UE. FIG. 11 is a flowchart illustrating a second embodiment of a monitoring method for an AI / ML model starting from a UE. Specific details for implementing the invention

[0029] The present disclosure is capable of various modifications and may have various embodiments, and specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the present disclosure to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the present disclosure.

[0030] Terms such as "first," "second," etc., may be used to describe various components, but said components should not be limited by said terms. Such terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present disclosure, the first component may be named the second component, and similarly, the second component may be named the first component. The term "and / or" includes a combination of a plurality of related described items or any of a plurality of related described items.

[0031] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0032] The terms used in this disclosure are used merely to describe specific embodiments and are not intended to limit this disclosure. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this disclosure, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0033] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which this disclosure pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this disclosure.

[0034] A communication system to which embodiments according to the present disclosure are applied will be described. The communication system to which embodiments according to the present disclosure are applied is not limited to the details described below, and embodiments according to the present disclosure may be applied to various communication systems. Here, the term "communication system" may be used interchangeably with "communication network."

[0035] Throughout the specification, a network may include, for example, wireless internet such as WiFi (wireless fidelity), mobile internet such as WiBro (wireless broadband internet) or WiMAX (world interoperability for microwave access), 2G mobile communication networks such as GSM (global system for mobile communication) or CDMA (code division multiple access), 3G mobile communication networks such as WCDMA (wideband code division multiple access) or CDMA2000, 3.5G mobile communication networks such as HSDPA (high speed downlink packet access) or HSUPA (high speed uplink packet access), 4G mobile communication networks such as LTE (long term evolution) networks or LTE-Advanced networks, and 5G mobile communication networks.

[0036] Throughout the specification, the term "terminal" may refer to a mobile station, mobile terminal, subscriber station, portable subscriber station, user equipment (UE), access terminal, etc., and may include all or part of the functions of a terminal, mobile station, mobile terminal, subscriber station, portable subscriber station, user equipment, access terminal, etc.

[0037] Here, a desktop computer, laptop computer, tablet PC, wireless phone, mobile phone, smartphone, smart watch, smart glass, e-book reader, PMP (portable multimedia player), portable game console, navigation device, digital camera, DMB (digital multimedia broadcasting) player, digital audio recorder, digital audio player, digital picture recorder, digital picture player, digital video recorder, digital video player, etc., capable of communicating with a terminal can be used.

[0038] Throughout the specification, the term "base station" may refer to an access point, a radio access station, a node B, an evolved node B, a base transceiver station, a mobile multihop relay (MMR)-BS, etc., and may include all or part of the functions of a base station, access point, radio access station, node B, eNodeB, base transceiver station, MMR-BS, etc.

[0039] Hereinafter, preferred embodiments of the present disclosure will be described in more detail with reference to the attached drawings. In order to facilitate an overall understanding of the present disclosure, the same reference numerals are used for identical components in the drawings, and redundant descriptions of identical components are omitted.

[0040] FIG. 1 is a conceptual diagram illustrating an embodiment of a communication system.

[0041] Referring to FIG. 1, the communication system (100) may include a plurality of communication nodes (110-1, 110-2, 110-3, 120-1, 120-2, 130-1, 130-2, 130-3, 130-4, 130-5, 130-6). The plurality of communication nodes may support 4G communication (e.g., LTE (long term evolution), LTE-A (advanced)), 5G communication (e.g., NR (new radio)), etc., as defined in the 3GPP (3rd generation partnership project) standard. 4G communication may be performed in a frequency band of 6 GHz or lower, and 5G communication may be performed not only in a frequency band of 6 GHz or lower but also in a frequency band of 6 GHz or higher.

[0042] For example, for 4G communication and 5G communication, multiple communication nodes can support communication protocols based on CDMA (code division multiple access), WCDMA (wideband CDMA), TDMA (time division multiple access), FDMA (frequency division multiple access), OFDM (orthogonal frequency division multiplexing), Filtered OFDM, CP (cyclic prefix)-OFDM, DFT-s-OFDM (discrete Fourier transform-spread-OFDM), OFDMA (orthogonal frequency division multiple access), SC (single carrier)-FDMA, NOMA (Non-orthogonal Multiple Access), GFDM (generalized frequency division multiplexing), FBMC (filter bank multi-carrier) based communication protocol, UFMC (universal filtered multi-carrier) based communication protocol, SDMA (Space Division Multiple Access) based communication protocol, etc.

[0043] Additionally, the communication system (100) may further include a core network. If the communication system (100) supports 4G communication, the core network may include an S-GW (serving-gateway), a P-GW (PDN (packet data network)-gateway), an MME (mobility management entity), etc. If the communication system (100) supports 5G communication, the core network may include a UPF (user plane function), an SMF (session management function), an AMF (access and mobility management function), etc.

[0044] Meanwhile, each of the plurality of communication nodes (110-1, 110-2, 110-3, 120-1, 120-2, 130-1, 130-2, 130-3, 130-4, 130-5, 130-6) constituting the communication system (100) may have the following structure.

[0045] FIG. 2 is a block diagram illustrating an example of a communication node constituting a communication system.

[0046] Referring to FIG. 2, the communication node (200) may include at least one processor (210), a memory (220), and a transceiver (230) that is connected to a network to perform communication. Additionally, the communication node (200) may further include an input interface device (240), an output interface device (250), a storage device (260), etc. Each component included in the communication node (200) may be connected by a bus (270) to communicate with one another.

[0047] However, each component included in the communication node (200) may be connected via individual interfaces or individual buses centered around the processor (210), rather than via a common bus (270). For example, the processor (210) may be connected via a dedicated interface to at least one of a memory (220), a transmission / reception device (230), an input interface device (240), an output interface device (250), and a storage device (260).

[0048] The processor (210) can execute a program command stored in at least one of the memory (220) and the storage device (260). The processor (210) may mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which the methods according to embodiments of the present disclosure are performed. Each of the memory (220) and the storage device (260) may be composed of at least one of a volatile storage medium and a non-volatile storage medium. For example, the memory (220) may be composed of at least one of read-only memory (ROM) and random access memory (RAM).

[0049] Referring again to FIG. 1, the communication system (100) may include a plurality of base stations (110-1, 110-2, 110-3, 120-1, 120-2) and a plurality of terminals (130-1, 130-2, 130-3, 130-4, 130-5, 130-6). The communication system (100) including the base stations (110-1, 110-2, 110-3, 120-1, 120-2) and terminals (130-1, 130-2, 130-3, 130-4, 130-5, 130-6) may be referred to as an "access network". Each of the first base station (110-1), the second base station (110-2), and the third base station (110-3) can form a macro cell. Each of the fourth base station (120-1) and the fifth base station (120-2) can form a small cell. The fourth base station (120-1), the third terminal (130-3), and the fourth terminal (130-4) may be located within the cell coverage of the first base station (110-1). The second terminal (130-2), the fourth terminal (130-4), and the fifth terminal (130-5) may be located within the cell coverage of the second base station (110-2). The fifth base station (120-2), the fourth terminal (130-4), the fifth terminal (130-5), and the sixth terminal (130-6) may be located within the cell coverage of the third base station (110-3). The first terminal (130-1) may be located within the cell coverage of the fourth base station (120-1). The sixth terminal (130-6) may be located within the cell coverage of the fifth base station (120-2).

[0050] Here, each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) may be referred to as Node B, evolved Node B, base transceiver station (BTS), radio base station, radio transceiver, access point, access node, road side unit (RSU), radio remote head (RRH), transmission point (TP), transmission and reception point (TRP), eNB, gNB, etc.

[0051] Each of the multiple terminals (130-1, 130-2, 130-3, 130-4, 130-5, 130-6) may be referred to as a UE (user equipment), terminal, access terminal, mobile terminal, station, subscriber station, mobile station, portable subscriber station, node, device, IoT (Internet of Thing) device, mounted module / device / terminal or on board device / terminal, etc.

[0052] Meanwhile, each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) may operate in different frequency bands or in the same frequency band. Each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) may be connected to each other via an ideal backhaul link or a non-ideal backhaul link, and may exchange information with each other via an ideal backhaul link or a non-ideal backhaul link. Each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) may be connected to a core network via an ideal backhaul link or a non-ideal backhaul link. Each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) can transmit a signal received from the core network to the corresponding terminal (130-1, 130-2, 130-3, 130-4, 130-5, 130-6), and can transmit a signal received from the corresponding terminal (130-1, 130-2, 130-3, 130-4, 130-5, 130-6) to the core network.

[0053] In addition, each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) can support MIMO transmission (e.g., SU (single user)-MIMO, MU (multi user)-MIMO, massive MIMO, etc.), CoMP (coordinated multipoint) transmission, CA (carrier aggregation) transmission, transmission in an unlicensed band, device-to-device communication (D2D) (or ProSe (proximity services)), etc. Here, each of the plurality of terminals (130-1, 130-2, 130-3, 130-4, 130-5, 130-6) can perform an operation corresponding to the base station (110-1, 110-2, 110-3, 120-1, 120-2) and an operation supported by the base station (110-1, 110-2, 110-3, 120-1, 120-2). For example, the second base station (110-2) can transmit a signal to the fourth terminal (130-4) based on the SU-MIMO method, and the fourth terminal (130-4) can receive a signal from the second base station (110-2) based on the SU-MIMO method. Alternatively, the second base station (110-2) can transmit a signal to the fourth terminal (130-4) and the fifth terminal (130-5) based on the MU-MIMO method, and each of the fourth terminal (130-4) and the fifth terminal (130-5) can receive a signal from the second base station (110-2) by the MU-MIMO method.

[0054] Each of the first base station (110-1), the second base station (110-2), and the third base station (110-3) can transmit a signal to the fourth terminal (130-4) based on the CoMP method, and the fourth terminal (130-4) can receive a signal from the first base station (110-1), the second base station (110-2), and the third base station (110-3) by the CoMP method. Each of the multiple base stations (110-1, 110-2, 110-3, 120-1, 120-2) can transmit and receive signals based on the CA method with terminals (130-1, 130-2, 130-3, 130-4, 130-5, 130-6) within its cell coverage area. Each of the first base station (110-1), the second base station (110-2), and the third base station (110-3) can control D2D between the fourth terminal (130-4) and the fifth terminal (130-5), and each of the fourth terminal (130-4) and the fifth terminal (130-5) can perform D2D by controlling each of the second base station (110-2) and the third base station (110-3).

[0055] Next, methods for configuring and managing wireless interfaces in a communication system will be described. Even when a method performed by a first communication node among the communication nodes (e.g., transmission or reception of a signal) is described, the corresponding second communication node may perform a method corresponding to the method performed by the first communication node (e.g., reception or transmission of a signal). That is, when the operation of a UE is described, the corresponding base station may perform an operation corresponding to the operation of the UE. Conversely, when the operation of a base station is described, the corresponding UE may perform an operation corresponding to the operation of the base station.

[0056] Meanwhile, in a communication system, a base station can perform all functions of the communication protocol (e.g., remote radio transmission and reception functions, baseband processing functions). Alternatively, among all functions of the communication protocol, the remote radio transmission and reception function may be performed by a TRP (transmission reception point) (e.g., f(flexible)-TRP), and among all functions of the communication protocol, the baseband processing function may be performed by a BBU (baseband unit) block. The TRP may be an RRH (remote radio head), RU (radio unit), TP (transmission point), etc. A BBU block may include at least one BBU or at least one DU (digital unit). A BBU block may be referred to as a "BBU pool," "centralized BBU," etc. A TRP may be connected to a BBU block via a wired fronthaul link or a wireless fronthaul link. A communication system composed of backhaul links and fronthaul links may be as follows. When the function split method of the communication protocol is applied, the TRP can selectively perform some functions of the BBU or some functions of MAC (medium access control) / RLC (radio link control).

[0057] In the present disclosure, a phrase containing "~ case (e.g., when ~)" may be expressed as a phrase containing "~ based on (e.g., based on ~)" or a phrase containing "~ in response to (e.g., in response to ~)". In other words, a phrase containing "~ case" may be interpreted as identical or similar to a phrase containing "~ based on" or a phrase containing "~ in response to".

[0058] Meanwhile, we will examine the procedure for channel state information (CSI) being reported from the UE to the base station in a 5G mobile communication system, also known as new radio (NR), with reference to FIGS. 3a and 3b.

[0059] Figure 3a is a flowchart of a UE performing a CSI report to a base station based on a periodic CSI reporting method in a mobile communication system.

[0060] Referring to FIG. 3a, in step S310, the base station can transmit upper layer configuration information to the UE. The upper layer configuration information may be transmitted, for example, by a radio resource control (RRC) signaling message. The RRC signaling message may include information on the transmission type and CSI reporting type of the CSI-reference signal (RS) (CSI-RS) transmitted to the UE. Since FIG. 3a describes a periodic CSI reporting method, it is assumed that the CSI-RS transmission type is periodic and the CSI reporting type is periodic.

[0061] In step S312, the base station can transmit periodic CSI-RS. Therefore, the UE can receive the periodic CSI-RS transmitted by the base station in step S312. Since base stations generally use the Multiple Input Multiple Output (MIMO) method in 5G NR, CSI-RS can be transmitted through multiple beams. Therefore, the UE can receive the CSI-RS corresponding to each beam through the multiple beams transmitted by the base station. The UE can measure the received CSI-RS. The UE can generate a CSI report message based on the received RRC signaling message and the measured values ​​of the CSI-RS.

[0062] In step S314, the UE can transmit the generated CSI report message to the base station. The timing of transmission of the CSI report message can be determined based on the RRC signaling message in step S310. Since the embodiment of FIG. 3a is a case where periodic measurement reporting is performed, it can be transmitted based on the reporting period set in the RRC signaling message.

[0063] Figure 3b is a flowchart of a UE performing CSI reporting to a base station based on a semi-static or non-periodic CSI reporting method in a mobile communication system.

[0064] Referring to FIG. 3b, in step S320, the base station can transmit upper layer configuration information to the UE. The upper layer configuration information may be transmitted, for example, by an RRC signaling message. The RRC signaling message may include information on the transmission type and CSI reporting type of CSI-RS transmitted to the UE. Since FIG. 3b illustrates a semi-persistent CSI reporting method or an aperioditic CSI reporting method, it is assumed that the CSI-RS transmission type is aperioditic or semi-persistent.

[0065] In step S322, the base station may transmit a CSI report triggering message to the UE. The CSI report triggering message may consist of only one message or two messages. It should be noted that Figure 3b is illustrated as a single form for convenience of explanation. For example, a triggering message for semi-static CSI reporting may be triggered by a medium access control-control element (MAC-CE) message and downlink control information (DCI). A triggering message for non-periodic CSI reporting may be triggered by the DCI. Such triggering is described in more detail with reference to Table 1, which is described below.

[0066] In step S324, the base station may transmit semi-static or non-periodic CSI-RS. Therefore, the UE may receive the semi-static or non-periodic CSI-RS transmitted by the base station in step S324. As previously described, since the base station generally uses the MIMO method, the CSI-RS may be transmitted through multiple beams. Therefore, the UE may receive the CSI-RS corresponding to each beam through the multiple beams transmitted by the base station. The UE may measure the received CSI-RS. The UE may generate a CSI report message based on the measured values ​​of the CSI-RS and the received RRC signaling message.

[0067] In step S326, the UE can transmit the generated CSI report message to the base station. The timing of transmission of the CSI report message can be determined based on the received DCI.

[0068] The CSI reporting settings described in the procedures of FIGS. 3a and 3b described above can be provided to the UE by an RRC signaling message. The CSI reporting settings transmitted by the RRC signaling message ( CSI-RrportConfigThe ) information element (IE) is the reporting configuration identifier ( reportConfigId ), resources for channel measurement ( resourcesForChannelMeasurement ), resources for interference measurement( resourcesForInterferenceMeasurement ), Reporting setting type( reportConfigType ), reported quantity( reportQuantity ), Report frequency setting( reportFreqConfiguration ), timeout for channel measurement( timeRestrictionForChannelMeasurements ), time limit for interference measurement( timeRestrictionForInterferenceMeasurements ), Codebook settings( codebookConfig It may include information elements such as ).

[0069] The IEs exemplified above are described as examples of only some components of an RRC signaling message to aid in understanding the present disclosure, and additional IEs may be included in addition to those exemplified above. In the CSI reporting configuration, the CSI reporting type is the reporting configuration type ( reportConfigType As described in FIGS. 3a and 3b, it can be set to any one of the aperioditic reporting method, semipersistent reporting method, or periodic reporting method by IE.

[0070] Representative CSIs used in 5G NR include channel quality information (CQI), precoding matrix indicator (PMI), CSI-RS resource indicator (CRI), synchronization signal / physical broadcast channel (SS / PBCH) resource block indicator (SSSBRI), layer indicator (LI), rank indicator (RI), and layer 1 (L1)-reference signal received power (L1-RSRP).

[0071] Among the CSI examples above, the CSI reporting settings are the reporting quantity ( reportQuantity Through IE, you can configure reporting for CSI-related quantities or L1-RSRP-related quantities. A typical example of L1-RSRP-related quantities is the reporting of beam-related information for beam management (BM). In this case, CRI, SSBRI, etc., can be reported for specific beam indications, and corresponding L1-RSRP values ​​can be reported together for quality reporting regarding that beam. CSI can be configured to report CSI-related quantities such as PMI, RI, and CQI, in addition to BM-related matters. Furthermore, various other information settings related to reporting can be configured.

[0072] The configuration for CSI reporting can be broadly divided into two parts. For example, it can be divided into configuration information for triggering states for CSI reporting, and CSI reporting-related configurations associated with each of the triggering states for CSI reporting.

[0073] The UE may receive from the base station configuration information for one or more CSI-ReportConfigs for CSI report-related report settings, one or more CSI-ResourceConfigs for resource settings, and one or two CSI report triggering state lists (TriggeringStateList) containing multiple triggering states. The CSI report triggering state lists may be, for example, a CSI aperiodic triggering state list (CSI-AperiodicTriggerStateList) and / or a semi-static CSI triggering state list on PUSCH (CSI-SemiPersistentOnPUSCH-TriggerStateList). Triggering states within the CSI report triggering state lists may be associated with the configuration information of the CSI-ReportConfigs and CSI-ResourceConfigs.

[0074] In addition, configuration information related to CSI reporting is defined, and configuration information related to CSI reporting can be configured through association.

[0075] Also, CSI report settings ( CSI-ReportConfig ) is CSI resource configuration( CSI-ResourceConfig It can be associated with ). CSI report settings ( CSI-ReportConfig ) and CSI resource settings( CSI-ResourceConfigThe values ​​may be associated with resource configuration information that includes resource configuration information of a reference signal (RS) transmitted for CSI reporting. The resource configuration information of the RS may, for example, mean resource configuration information related to CSI-RS or a synchronization signal block (SSB).

[0076] CSI reporting can be performed based on the configuration information described above. At this time, the report configuration available according to the CSI-RS configuration can be configured as shown in Table 1 below. Table 1 below is the content of Table 5.2.1.4-1 of 3GPP TS 38.214, a standard specification for mobile communication systems.

[0077] CSI-RS configuration Periodic CSI Report Semi-static CSI Report Non-periodic CSI reports Periodic CSI-RS No dynamic triggering / activation For PUCCH reporting, the UE receives an enable command as described in Section 6.1.3.16 of [10, TS 38.321]. For PUSCH reporting, the UE receives a triggering from the DCI. Triggered by DCI; also, the sub-selection indication described in Section 6.1.3.13 of [10, TS 38.321] is possible as defined in Section 5.2.1.5.1. Semi-static CSI-RS Not supported For PUCCH reporting, the UE receives an enable command as described in Section 6.1.3.16 of [10, TS 38.321]. For PUSCH reporting, the UE receives a triggering from the DCI. Triggered by DCI; also, the sub-selection indication described in Section 6.1.3.13 of [10, TS 38.321] is possible as defined in Section 5.2.1.5.1. Aperiodic CSI-RS Not supported Not supported Triggered by DCI; also, the sub-selection indication described in Section 6.1.3.13 of [10, TS 38.321] is possible as defined in Section 5.2.1.5.1.

[0078] As exemplified in Table 1 above, in an environment where periodic CSI-RS is configured (as in Fig. 3a), the UE can perform periodic CSI reporting after RRC configuration without triggering / activation for CSI reporting by DCI or MAC-CE. On the other hand, in the case of semi-static CSI reporting or non-periodic CSI reporting (as in Fig. 3b), CSI reporting operations can be performed by triggering / activation for CSI reporting by DCI or a combination of MAC-CE and DCI based on configured information.

[0079] Based on Table 1 above, it can be seen that in an environment where semi-static CSI-RS is configured, it is impossible for the UE to periodically report CSI, and the UE can report CSI semi-permanently or non-periodically through triggering / activation by DCI or MAC-CE.

[0080] In addition, based on Table 1 above, in an environment where non-periodic CSI-RS is configured, the UE cannot report CSI periodically and semi-statically, and the UE can only report CSI non-periodically through triggering / activation by DCI or MAC-CE.

[0081] Meanwhile, artificial intelligence (AI) / machine learning (ML)-based beam management technology was introduced as one of the consensus points at the 3GPP standardization body. Additionally, CSI prediction and CSI compression technologies are also highly likely to be introduced at the 3GPP standardization body. If AI / ML-based beam management and CSI prediction and / or CSI compression are introduced, the UE must be supported with the ability to report at least the results of measuring a specific RS transmitted by the base station and / or the output values ​​from the UE's AI / ML model—in other words, the prediction results. However, a method for the UE to report the results of measuring a specific RS received from the base station and / or the prediction results from the UE's AI / ML model has not yet been presented.

[0082] The present disclosure described below describes a method for a UE to report the result of measuring a specific RS received from a base station and / or the result of a prediction from the UE's AI / ML model. To perform such a method, the UE may have an AI / ML model in an AI / ML-based air interface technology. In the following description, an AI / ML model that is loaded on or runs on the UE will be referred to as a UE-side model. The AI / ML model may also be loaded on or run on a network (e.g., a base station). In the following description, an AI / ML model that is loaded on or runs on a network will be referred to as a Network-side model (NW-side model). Additionally, for convenience of explanation, it is assumed that the network is a base station.

[0083] When an AI / ML model is deployed on both the UE and the network, it may be referred to as a two-sided model, and when an AI / ML model is deployed on only one side, either the UE or the network, it may be referred to as a one-sided model.

[0084] In the case of the UE-side model and the two-side model, the UE may be equipped with an AI / ML model (the UE possesses an AI / ML model). The UE receives a reference signal (RS) transmitted by the base station and can input a measured value of the received RS (e.g., a measured value of reference signal received power (RSRP), or a measured value for beam management (BM), etc.) into the UE-side model. The UE-side model can output a specific prediction value based on the measured RS value. The UE can report the prediction value of the UE-side model to the base station.

[0085] For example, when a UE-side model makes predictions related to a beam, the UE may report identification (ID) information and quality information for the beam to the base station. The ID for the beam may include one of the Channel State Information-Reference Signal Resource Indicator (CRI) or the Synchronization Signal (SS) / Physical Broadcast Channel (PBCH) Block Resource Indicator (SSBRI). Additionally, the quality information for the beam may include Layer 1-RSRP (L1-RSRP) and Signal-to-Interference Noise Ratio (SINR) information.

[0086] In this case, since the information reported by the UE to the base station is the output of the UE-side model, the beam ID can be the CRI or SSBRI predicted by the UE-side model, and the beam quality information can be the L1-RSRP, SINR, etc. predicted by the UE-side model.

[0087] The UE may report all beams predicted by the UE-side model—that is, all predicted values—to the base station, or may report predicted value(s) for some beams to the base station. If the information included in the report message reported by the UE to the base station contains predicted values ​​for only a portion of the CSI information, the remaining information may contain actual measured values ​​of the CSI information.

[0088] If the UE-side model predicts CSI information for the channel in addition to the BM, the UE may generate a report message containing predicted information related to the channel quality indicator (CQI), precoding matrix indicator (PMI), CRI, SSBRI, layer indicator (LI), rank indicator (RI), L1-RSRP, and SINR. The UE may report (or transmit) the generated report message to the base station. Even when the UE-side model predicts CSI information for the channel, the CSI information included in the report message may consist entirely of predicted values ​​or only partially of predicted value(s). If only partially of the CSI information included in the report message consists of predicted value(s), the remainder of the CSI information may include actual measured value(s).

[0089] Meanwhile, in the case of a network-side model or a two-side model, the base station may carry (or include) the network-side model. The base station transmits RS to the UE and may receive measurement information and / or prediction information of the RS from the UE via a report message. The base station may input the measurement values ​​and / or prediction values ​​included in the received report message into the network-side model. The network-side model may output a specific prediction value using the measurement values ​​and / or prediction values.

[0090] The case where the input to the network-side model consists only of measurement values ​​may be any one of the following: the UE does not have a UE-side model; even if a UE-side model exists, it is not used; or even if a UE-side model exists and is used, the base station inputs only the measurement values ​​included in the report message into the network-side model.

[0091] If the input of the network-side model is a predicted value of the UE-side model, the input of the network-side model may be one or more of the output values ​​of the UE-side model described above. In other words, the input of the network-side model may be one or more of the predicted CQI, predicted PMI, predicted CRI, predicted SSBRI, predicted LI, predicted RI, predicted L1-RSRP, or predicted SINR. As another example, the input of the network-side model may be one or more values ​​obtained by compressing the channel values ​​estimated by RS, or values ​​obtained by modifying the estimated channel values ​​by a specific method.

[0092] For example, when using two-sided models, the UE can transmit a report message containing reporting information to the base station. The reporting information may be channel values ​​(or information) estimated by the UE using RS received from the base station. Alternatively, the reporting information may be the data in its original form (output source) output from the UE-side model based on CSI information. The base station may use the reporting information included in the report message (e.g., channel values ​​estimated by the UE or original output data output from the UE-side model) as input to the network-side model. The network-side model may output specific information based on the input information. In this case, if the input information of the network-side model is the output information of the UE-side model, the output of the network-side model may be the reconstructed information of that specific information. In such cases, where the UE-side model compresses specific information and the network-side model reconstructs it, the UE-side model and the network-side model can be understood as auto-encoders. Therefore, when the UE-side model and the network-side model operate as auto-encoders, the value reported by the UE may simply be the output value generated by the UE-side model.

[0093] Meanwhile, in the disclosure described below, a CSI reporting method in which an AI / ML wireless interface is not used is referred to as legacy CSI reporting. Additionally, in the disclosure, a CSI reporting method in which an AI / ML wireless interface is used is referred to as AI / ML CSI reporting. Cases where an AI / ML wireless interface is used may include both the UE-side model, the network-side model, and the two-side model, which are the one-side models described above.

[0094] Legacy CSI reports may be reports containing both CSI-related quantities and L1-RSRP-related quantities. Additionally, AI / ML CSI reports may include both CSI-related quantities and L1-RSRP-related quantities from legacy CSI reports, and may also include newly defined quantities. For example, AI / ML CSI reports may be channel values ​​estimated based on RS, and / or values ​​that have been compressed or modified.

[0095] [First Embodiment: A method for enabling only one of the two CSI reporting methods for a UE that supports both legacy CSI reporting and AI / ML CSI reporting]

[0096] The first embodiment of the present disclosure described below describes the case where the UE has a UE-side model. As previously explained, cases where the UE has a UE-side model may include a one-side model where only the UE has an AI / ML model, and a two-side model where both the base station and the UE have AI / ML models. It should be noted that the present disclosure described below may be applicable not only to the case of a one-side model but also to the case of a two-side model.

[0097] The first embodiment described below can be broadly classified into two methods. First, the CSI reporting method may be changed after being (re)configured by a higher-layer signaling (e.g., RRC signaling) message. Second, the CSI reporting method may be changed by a lower-layer triggering signal without being (re)configured by a higher-layer signaling (e.g., RRC signaling) message. In the following description, these two methods are described separately.

[0098] <Method 1>

[0099] In the first embodiment of the present disclosure, the first method may be a case where a UE that enables only one of legacy CSI reporting and AI / ML CSI reporting switches to the AI / ML CSI reporting method while the legacy CSI reporting method is applied. In such a case, the first method according to the first embodiment of the present disclosure may be a case where the legacy CSI reporting setting and the AI / ML CSI reporting setting are respectively performed by an upper layer signaling (e.g., RRC signaling) message. The first method of the first embodiment is described in more detail with reference to the attached drawings.

[0100] Figure 4 is a flowchart illustrating the transition from a periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system.

[0101] Referring to FIG. 4, in step S410, the base station can transmit upper-layer configuration information for legacy CSI reporting to the UE. The upper-layer configuration information for legacy CSI reporting can be transmitted, for example, by a radio resource control (RRC) signaling message. The RRC signaling message for legacy CSI reporting may include the transmission type and CSI reporting type information of the CSI-reference signal (RS) (CSI-RS) transmitted to the UE. Since FIG. 4 describes a periodic CSI reporting method, it is assumed that the CSI-RS transmission type is periodic and the CSI reporting type is periodic.

[0102] In step S412, the base station can transmit periodic CSI-RS based on an RRC signaling message regarding legacy CSI reporting. In step S412, the UE can receive the periodic CSI-RS transmitted by the base station based on an RRC signaling message regarding legacy CSI reporting. Since base stations generally use the Multiple Input Multiple Output (MIMO) method in 5G NR, CSI-RS can be transmitted through multiple beams. Therefore, the UE can receive CSI-RS corresponding to each beam through multiple beams transmitted by the base station. The UE can measure the received CSI-RS. The UE can generate a CSI report message based on the measured values ​​of the CSI-RS and the received RRC signaling message regarding legacy CSI reporting.

[0103] In step S414, the UE can transmit the generated CSI report message to the base station. The timing of transmission of the CSI report message can be determined based on the RRC signaling message for the legacy CSI report in step S410. Since the embodiment of FIG. 4 is a case where periodic legacy CSI measurement reporting is performed, the UE can transmit the CSI report message to the base station based on the legacy CSI report cycle set in the RRC signaling message. Steps S412 through S414 described above can be repeated every CSI report cycle set by the RRC signaling message for the legacy CSI report set by the base station to the UE in step S400.

[0104] If an AI / ML CSI report is required, the base station may transmit upper-layer (re)configuration information regarding the AI / ML CSI report to the UE at step S420. The upper-layer configuration information regarding the AI / ML CSI report may be, for example, RRC configuration information or RRC reconfiguration information. For the sake of convenience of explanation, the following description assumes that the RRC signaling message transmitted at step S420 is RRC configuration information. However, the same applies even if it is RRC reconfiguration information.

[0105] In step S420, the RRC configuration information for the AI / ML CSI report transmitted from the base station to the UE may include the transmission type of the AI / ML CSI-RS and the AI / ML CSI report type information. The RRC configuration information may set the transmission type of the AI / ML CSI-RS to one of periodic, semi-static, or non-periodic transmission types. Additionally, the RRC configuration information may set the AI / ML CSI report type to one of periodic, semi-static, or non-periodic report types.

[0106] Additionally, RRC configuration information for AI / ML CSI reporting may include release or disable information for legacy CSI reporting settings. In specific cases, RRC configuration information for AI / ML CSI reporting may also include instruction information directing the triggering or activation of part or all of the AI / ML CSI reporting information.

[0107] If the transmission type of the AI / ML CSI-RS is set to the periodic AI / ML CSI reporting type, the UE may perform step S424 without performing step S422. On the other hand, if the transmission type of the AI / ML CSI-RS is set to the semi-static AI / ML CSI reporting type or the non-periodic AI / ML CSI reporting type, step S424 may be performed. It should be noted that in Figure 4, step S422 is illustrated with a dotted line to explain all cases where the transmission type of the AI / ML CSI-RS is the periodic AI / ML CSI reporting type, the semi-static AI / ML CSI reporting type, or the non-periodic AI / ML CSI reporting type.

[0108] If the transmission type of the AI / ML CSI-RS is a semi-static AI / ML CSI reporting type or a non-periodic AI / ML CSI reporting type, the base station may instruct (or set) the UE to enable or trigger the AI / ML CSI reporting using a low-layer triggering message (or signal). In other words, at step S422, the base station may transmit a low-layer triggering message (or signal) to the UE.

[0109] When step S422 is performed, the lower-level triggering messages may be, for example, MAC-CE messages and / or DCI messages. For instance, similar to legacy CSI reporting, MAC-CE messages may be used to enable semi-static AI / ML CSI reporting, and DCIs may be used to trigger acyclic AI / ML CSI reporting. As another example, both semi-static AI / ML CSI reporting and acyclic AI / ML CSI reporting may be enabled or triggered by a combination of MAC-CE messages and DCIs. As yet another example, only acyclic AI / ML CSI reporting may be triggered by a combination of MAC-CE messages and DCIs.

[0110] If semi-static AI / ML CSI reporting and non-periodic AI / ML CSI reporting are performed in the same manner as legacy CSI reporting methods, they can be configured (or directed) and operated as follows.

[0111] Semi-static AI / ML CSI reporting can be enabled by a MAC-CE message. If multiple semi-static AI / ML CSI reports are configured by the RRC, the MAC-CE for enabling the semi-static AI / ML CSI reporting may include activation instruction information for one or more of the configured semi-static AI / ML CSI reports. Such instruction information may be provided by a bitmap (instruction field) for enabling semi-static CSI reporting within the MAC-CE, which is mapped to an identifier for each semi-static CSI report configuration.

[0112] Non-periodic or semi-static AI / ML CSI reporting can be triggered or activated by DCI messages. For non-periodic reporting, specific reporting settings and resources are indicated through the Trigger State pointed to by the DCI's CSI request field; for PUSCH-based semi-static reporting, a specific SP CSI trigger state can be activated through the DCI's CSI request field scrambled as SP-CSI-RNTI. If the RRC configuration information includes multiple AI / ML CSI-RS resource sets associated with AI / ML CSI reporting, the DCI may selectively indicate a specific combination of AI / ML CSI-RS resources (Trigger State) among the multiple resource sets to be used for reporting through the CSI request field indication.

[0113] In step S424, the base station can transmit CSI-RS for AI / ML to the UE. The CSI-RS for AI / ML may be transmitted via the same beams as those transmitted in step S412, or via different beams. In step S424, the UE can receive the CSI-RS for AI / ML transmitted by the base station. The UE can measure the received CSI-RS for AI / ML and use the measured value as input to the UE-side model. The UE can obtain AI / ML CSI (predicted CSI or inferred CSI) through the UE-side model. The UE can generate an AI / ML CSI report message containing the obtained AI / ML CSI. At this time, depending on the AI / ML CSI report settings, the AI / ML CSI report message may include not only the AI / ML CSI (predicted CSI or inferred CSI) obtained through the UE-side model but also the measured CSI values.

[0114] In step S426, the UE can transmit the generated AI / ML CSI report message to the base station. If the generated AI / ML CSI report message is a message generated based on periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting settings, the UL channel may be a UL channel configured by the RRC configuration information. On the other hand, if the message is generated based on non-periodic reporting settings or PUSCH-based semi-static reporting settings, the UL channel may be a UL channel granted by the DCI. Therefore, the base station can receive the AI / ML CSI report message from the UE through the UL channel configured by the RRC configuration information or the DCI. The base station can monitor the prediction performance of the UE-side model based on the received AI / ML CSI report message.

[0115] Figure 5 is a flowchart illustrating the transition from a semi-static or non-periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system.

[0116] Referring to FIG. 5, in step S510, the base station can transmit upper-layer configuration information for legacy CSI reporting to the UE. The upper-layer configuration information for legacy CSI reporting can be transmitted, for example, by an RRC signaling message. The RRC signaling message for legacy CSI reporting may include information on the transmission type and CSI reporting type of CSI-RS transmitted to the UE. Since FIG. 5 describes a semi-static or non-periodic CSI reporting method, it is assumed that the CSI-RS transmission type is semi-static or non-periodic, and the CSI reporting type is also semi-static or non-periodic.

[0117] In step S512, if a CSI report is required from the UE, the base station can trigger a semi-static or non-periodic CSI report using a lower-layer triggering message. As previously described, the semi-static or non-periodic CSI report triggering message may consist of only one message (e.g., MAC-CE or DCI) or two messages (e.g., MAC-CE and DCI). It should be noted that Figure 5 illustrates a form in which a single message is transmitted for the convenience of explanation. For example, if two messages are used, the base station may transmit the MAC-CE message first and then the DCI.

[0118] In step S514, the base station may transmit semi-static or non-periodic CSI-RS to the UE based on RRC signaling messages and lower-layer triggering messages regarding legacy CSI reports. In step S514, the UE may receive semi-static CSI-RS or non-periodic CSI-RS based on RRC signaling messages and lower-layer triggering messages regarding legacy CSI reports. As previously described, since the base station uses the MIMO method in 5G NR, CSI-RS may be transmitted through multiple beams. Therefore, the UE may receive CSI-RS corresponding to each beam through multiple beams transmitted by the base station. The UE may measure the received CSI-RS. The UE may generate a CSI report message based on the measured values ​​of the CSI-RS and the received RRC signaling messages and lower-layer triggering messages regarding legacy CSI reports.

[0119] In step S516, the UE can transmit the generated CSI report message to the base station. Since the embodiment of FIG. 5 is a case where semi-static or non-periodic legacy CSI measurement reporting is performed, the UE can transmit the CSI report message to the base station based on the legacy CSI reporting period set in the RRC signaling message and / or the lower layer triggering message. In this case, the CSI report message transmitted over the uplink can be transmitted through the UL channel set by the RRC signaling or the UL channel set by the lower layer triggering message (e.g., DCI).

[0120] If an AI / ML CSI report is required, the base station may transmit upper-layer (re)configuration information regarding the AI / ML CSI report to the UE at step S520. The upper-layer configuration information regarding the AI / ML CSI report may be, for example, RRC configuration information or RRC reconfiguration information. For the sake of convenience of explanation, the following description assumes that the RRC signaling message transmitted at step S520 is RRC configuration information. However, the same applies even if it is RRC reconfiguration information.

[0121] In step S520, the RRC configuration information for the AI / ML CSI report transmitted from the base station to the UE may include the transmission type of the AI / ML CSI-RS and the AI / ML CSI report type information. The RRC configuration information may set the transmission type of the AI / ML CSI-RS to one of periodic, semi-static, or non-periodic transmission types. Additionally, the RRC configuration information may set the AI / ML CSI report type to one of periodic, semi-static, or non-periodic report types.

[0122] As previously described, RRC configuration information for AI / ML CSI reporting may include release or disable information for legacy CSI reporting settings. In specific cases, RRC configuration information for AI / ML CSI reporting may also include instruction information directing the activation of part or all of the AI / ML CSI reporting information.

[0123] If the transmission type of the AI / ML CSI-RS is set to the periodic AI / ML CSI reporting type, the UE may perform step S524 without performing step S522. On the other hand, if the transmission type of the AI / ML CSI-RS is set to the semi-static AI / ML CSI reporting type or the non-periodic AI / ML CSI reporting type, step S524 may be performed. It should be noted that in Figure 5, step S522 is illustrated with a dotted line to explain all cases where the transmission type of the AI / ML CSI-RS is the periodic AI / ML CSI reporting type, the semi-static AI / ML CSI reporting type, or the non-periodic AI / ML CSI reporting type.

[0124] If the transmission type of the AI / ML CSI-RS is a semi-static AI / ML CSI reporting type or a non-periodic AI / ML CSI reporting type, the base station may instruct (or set) the UE to enable or trigger the AI / ML CSI reporting using a low-layer triggering message (or signal). In other words, at step S522, the base station may transmit a low-layer triggering message (or signal) to the UE.

[0125] When step S522 is performed, the lower-level triggering messages may be, for example, MAC-CE messages and / or DCI messages. For instance, similar to legacy CSI reporting, MAC-CE messages may be used to enable semi-static AI / ML CSI reporting, and DCIs may be used to trigger acyclic AI / ML CSI reporting. As another example, both semi-static AI / ML CSI reporting and acyclic AI / ML CSI reporting may be enabled or triggered by a combination of MAC-CE messages and DCIs. As yet another example, only acyclic AI / ML CSI reporting may be triggered by a combination of MAC-CE messages and DCIs.

[0126] Since the cases where semi-static AI / ML CSI reporting and non-periodic AI / ML CSI reporting use the same method as legacy CSI reporting methods have already been explained in Figure 4, a redundant explanation is omitted.

[0127] In step S524, the base station can transmit CSI-RS for AI / ML to the UE. The CSI-RS for AI / ML may be transmitted via the same beams as those transmitted in step S514, or via different beams. In step S524, the UE can receive the CSI-RS for AI / ML transmitted by the base station. The UE can measure the received CSI-RS for AI / ML and use the measured value as input to the UE-side model. The UE can obtain AI / ML CSI (predicted CSI or inferred CSI) through the UE-side model. The UE can generate an AI / ML CSI report message containing the obtained AI / ML CSI. At this time, depending on the AI / ML CSI report settings, the AI / ML CSI report message may include not only the AI / ML CSI (predicted CSI or inferred CSI) obtained through the UE-side model but also the measured CSI values.

[0128] In step S526, the UE can transmit the generated AI / ML CSI report message to the base station. If the generated AI / ML CSI report message is a message generated based on periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting settings, the UL channel may be a UL channel configured by the RRC configuration information. On the other hand, if the message is generated based on non-periodic reporting settings or PUSCH-based semi-static reporting settings, the UL channel may be a UL channel granted by the DCI. Therefore, the base station can receive the AI / ML CSI report message from the UE through the UL channel configured by the RRC configuration information or the DCI. The base station can monitor the prediction performance of the UE-side model based on the received AI / ML CSI report message.

[0129] In FIGS. 4 and FIG. 5 described above, there may be cases where AI / ML CSI reporting is directed (or set) via a lower-layer triggering message after RRC setting information for AI / ML CSI reporting is transmitted from the base station to the UE. If AI / ML CSI reporting is initiated based on a lower-layer triggering message, such as a semi-static AI / ML CSI reporting type or a non-periodic AI / ML CSI reporting type, and the base station does not transmit a triggering message by the lower layer to the UE, the base station may not transmit CSI-RS for AI / ML. In this case, the UE may not only not receive CSI-RS for AI / ML but also not transmit an AI / ML CSI reporting message to the base station.

[0130] In FIGS. 4 and FIG. 5 described above, AI / ML CSI reporting messages may be transmitted via a physical uplink control channel (PUCCH) or a physical uplink shared channel (PUSCH). When AI / ML CSI reporting is triggered by triggering or activation by a lower-level triggering method, it may be non-periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting as previously described, and when AI / ML CSI reporting is activated by an activation indicator of upper-level configuration information, it may be periodic AI / ML CSI reporting. A specific AI / ML CSI reporting method may be determined by one or more combinations of the type of CSI reporting method, the uplink (UL) channel used for reporting, or the type of CSI-RS measured for CSI reporting.

[0131] Meanwhile, Figures 4 and 5 described above illustrate a case where the execution of legacy CSI reporting is changed to AI / ML CSI reporting. However, the opposite case is also possible. In other words, the execution of legacy CSI reporting can be changed while the execution of AI / ML CSI reporting is in progress. When the execution of legacy CSI reporting is changed while the execution of AI / ML CSI reporting is in progress, the base station may transmit (re)configuration information related to CSI reporting to the UE via an upper-layer signaling message. Subsequently, the legacy CSI reporting method may be performed based on an RRC signaling message and / or a lower-layer triggering message.

[0132] For example, in the example of FIG. 4, while steps S424 and S426 are being repeated, the upper layer configuration information described in step S410 (e.g., periodic CSI report) may be transmitted to the UE by the base station. Subsequently, the UE may repeat steps S412 and S414. In the case of FIG. 5, if steps S524 and S526 are performed once (e.g., in the case of a non-periodic AI / ML CSI report), or if steps S524 and S526 are repeated over a certain period (e.g., in the case of a semi-static AI / ML CSI report), the base station may change from an AI / ML CSI report to a legacy CSI report (e.g., a semi-static CSI report or a non-periodic CSI report) by performing steps S510 through S512. Accordingly, the UE and the base station may perform steps S514 and S516.

[0133] When changing from AI / ML CSI reporting to legacy CSI reporting, the RRC configuration information may additionally include disable instructions for AI / ML CSI reporting or deactivate instructions for AI / ML CSI reporting. Additionally, the RRC configuration information may include some or all of the instruction information that triggers or enables specific legacy CSI reporting methods among the legacy CSI reporting information.

[0134] The first method described above may be applied when the UE performs only legacy CSI reporting or only AI / ML CSI reporting by means of the RRC settings related to CSI reporting. Here, the meaning that only legacy CSI reporting operation is possible may mean that legacy CSI reporting can be performed as one of various methods determined by a combination of the type of legacy CSI reporting method, the UL channel used for legacy CSI reporting, and the type of CSI-RS measured for legacy CSI reporting. Likewise, the meaning that only AI / ML CSI reporting operation is possible may mean that AI / ML CSI reporting can be performed as one of various methods determined by a combination of the type of AI / ML CSI reporting method, the UL channel used for AI / ML CSI reporting, and the type of AI / ML CSI-RS measured for AI / ML CSI reporting.

[0135] The first method of the present disclosure described above may include an (re)configuration message by RRC, or a lower-level triggering signal such as a MAC-CE message and / or DCI, which may include explicit indicator information regarding the legacy CSI reporting or AI / ML-based CSI reporting, which is the CSI reporting method subject to change. In other words, in FIGS. 4 and 5, the RRC signaling message or the lower-level triggering message (e.g., MAC-CE message and / or DCI) may include activation / triggering indicator information for the AI / ML CSI reporting method. Additionally, the RRC signaling message or the lower-level triggering message (e.g., MAC-CE message and / or DCI) may include deactivation indicator information for the legacy CSI reporting method.

[0136] Meanwhile, in FIGS. 4 and FIG. 5 described above, if the legacy CSI reporting method is a periodic CSI reporting method, the AI / ML CSI reporting method may be performed in any one of the following ways: periodic AI / ML CSI reporting, semi-static AI / ML CSI reporting, or non-periodic AI / ML CSI reporting method. Additionally, even if the legacy CSI reporting method is a non-periodic CSI reporting method or a semi-static CSI reporting method, the AI / ML CSI reporting method may be performed in any one of the following ways: periodic AI / ML CSI reporting, semi-static AI / ML CSI reporting, or non-periodic AI / ML CSI reporting method.

[0137] <Method 2>

[0138] In the first embodiment of the present disclosure, the second method may also be a case where a UE that enables only one of legacy CSI reporting and AI / ML CSI reporting switches to the AI / ML CSI reporting method while the legacy CSI reporting method is applied. Unlike the first method, the second method according to the first embodiment of the present disclosure may be a case where the CSI reporting method switches without transmitting additional upper-layer signaling (e.g., RRC signaling) messages. The second method of the first embodiment is described in more detail with reference to the attached drawings.

[0139] Figure 6 is a flowchart illustrating the transition from a periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system.

[0140] Referring to FIG. 6, in step S610, the base station can transmit upper layer configuration information for CSI reporting to the UE. The upper layer configuration information for CSI reporting can be transmitted, for example, by an RRC signaling message. The RRC signaling message for CSI reporting may include information on the transmission type and CSI reporting type of CSI-RS transmitted to the UE. Since FIG. 6 illustrates a periodic CSI reporting method, it is assumed that the CSI-RS transmission type is periodic and the CSI reporting type is periodic.

[0141] In the second method of the present disclosure, the RRC signaling message may include both configuration information for legacy CSI reporting and configuration information for AI / ML CSI reporting. In this case, at least some of the configuration information for legacy CSI reporting may be identical to or related to the configuration information for AI / ML CSI reporting. As another example, the configuration information for legacy CSI reporting and the configuration information for AI / ML CSI reporting may be completely different configuration information. As yet another example, the configuration information for AI / ML CSI reporting may be configured separately from the configuration information for legacy CSI reporting. Step S610 should be understood as a form that includes all such configurations. However, for the convenience of explanation, the following description assumes that the RRC signaling message includes both configuration information for legacy CSI reporting and configuration information for AI / ML CSI reporting.

[0142] In Fig. 6, step S610 may be an example where the base station includes activation instruction information for a specific legacy CSI report among the configuration information for legacy CSI reports.

[0143] In step S612, the base station may transmit periodic CSI-RS based on legacy CSI reports that have been instructed to be activated. In step S612, the UE may receive the CSI-RS transmitted periodically by the base station based on an RRC signaling message regarding legacy CSI reports that have been instructed to be activated. Since base stations generally use MIMO in 5G NR, CSI-RS may be transmitted through multiple beams. Therefore, the UE may receive CSI-RS corresponding to each beam through multiple beams transmitted by the base station. The UE may measure the received CSI-RS. The UE may generate a CSI report message based on the measured values ​​of the CSI-RS and the RRC signaling message regarding the activated legacy CSI reports.

[0144] In step S614, the UE can transmit the generated CSI report message to the base station. The timing of transmission of the CSI report message and the channel through which the CSI report message is transmitted can be determined based on the RRC signaling message for legacy CSI reporting activated in step S610. Since the embodiment of FIG. 6 is a case where periodic legacy CSI measurement reporting is performed, the UE can transmit the CSI report message to the base station based on the CSI reporting period set in the RRC signaling message. Steps S612 through S614 described above can be repeated by the RRC signaling message for legacy CSI reporting activated by the base station to the UE in step S600.

[0145] If AI / ML CSI reporting is required, in step S620, the base station may transmit to the UE a triggering or enabling instruction (or setting) for one of the AI / ML CSI reporting method(s) set by the RRC signaling message in step S610. For lower-layer triggering messages for AI / ML CSI reporting, MAC-CE and / or DCI may be used, for example.

[0146] Since the first embodiment of the present disclosure is a case where only one of legacy CSI reporting or AI / ML CSI reporting is performed, when one of the AI / ML CSI reporting method(s) is triggered (or activated), the lower layer signaling message may include instructions to release or disable the legacy CSI reporting related settings.

[0147] As another example, since the first embodiment of the present disclosure performs only one of legacy CSI reporting or AI / ML CSI reporting, when one of the AI / ML CSI reporting method(s) is triggered (or enabled) in a lower-level signaling message, it may be interpreted as implicitly released or disabled even if there is no instruction information for the release or disable of the legacy CSI reporting related settings.

[0148] As another example, if the most recently activated (or triggered) CSI reporting action is AI / ML CSI reporting and the previously performed CSI reporting action was legacy CSI reporting, the UE may automatically interpret legacy CSI reporting as disallowed.

[0149] A lower-level signaling message according to one of the methods described above may include a part of indicator information or the whole of indicator information that triggers a specific AI / ML CSI reporting method (e.g., any one of periodic AI / ML CSI reporting, semi-static AI / ML CSI reporting, or non-periodic AI / ML CSI reporting) or enables a specific AI / ML CSI reporting method among the AI / ML CSI reporting information.

[0150] In step S622, the base station may transmit CSI-RS for AI / ML to the UE. The CSI-RS for AI / ML may be transmitted via transmission beams based on the CSI reporting settings for AI / ML activated in step S620. In step S622, the UE may receive the CSI-RS for AI / ML transmitted by the base station. The UE may measure the received CSI-RS for AI / ML and use the measured value as input to a UE-side model. The UE may obtain AI / ML CSI through the UE-side model. The UE may generate an AI / ML CSI reporting message containing the obtained AI / ML CSI. If the upper layer setting information or lower layer triggering message is configured to include at least some of the measured CSI values, the AI / ML CSI reporting message may include the measured CSI values ​​in addition to the AI / ML CSI.

[0151] In step S624, the UE can transmit the generated AI / ML CSI report message to the base station. If the generated AI / ML CSI report message is a message generated based on periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting settings, the UL channel may be a UL channel configured by the RRC configuration information. On the other hand, if the message is generated based on non-periodic reporting settings or PUSCH-based semi-static reporting settings, the UL channel may be a UL channel granted by the DCI. Therefore, the base station can receive the AI / ML CSI report message from the UE through the UL channel configured by the RRC configuration information or the DCI. The base station can monitor the prediction performance of the UE-side model based on the received AI / ML CSI report message.

[0152] Figure 7 is a flowchart illustrating the transition from a semi-static or non-periodic legacy CSI reporting method to an AI / ML CSI reporting method in a mobile communication system.

[0153] Referring to FIG. 7, in step S710, the base station can transmit upper-layer configuration information for CSI reporting to the UE. The upper-layer configuration information for CSI reporting can be transmitted, for example, via an RRC signaling message. The RRC signaling message for legacy CSI reporting may include information on the transmission type and CSI reporting type of CSI-RS transmitted to the UE. Since FIG. 7 describes a semi-static or non-periodic CSI reporting method, it is assumed that the CSI-RS transmission type is semi-static or non-periodic, and the CSI reporting type is also semi-static or non-periodic.

[0154] In the second method of the present disclosure, the RRC signaling message may include both configuration information for legacy CSI reporting and configuration information for AI / ML CSI reporting. In this case, at least some of the configuration information for legacy CSI reporting may be identical to or related to the configuration information for AI / ML CSI reporting. As another example, the configuration information for legacy CSI reporting and the configuration information for AI / ML CSI reporting may be completely different configuration information. In FIG. 7, step S710 may be an example where the base station includes activation instruction information for a specific configuration for legacy CSI reporting among the configuration information for legacy CSI reporting. As yet another example, the configuration information for AI / ML CSI reporting may be configured separately from the configuration information for legacy CSI reporting. Step S710 should be understood as a form that includes all such configurations. However, for the convenience of explanation, the following description assumes that the RRC signaling message includes both configuration information for legacy CSI reporting and configuration information for AI / ML CSI reporting.

[0155] In step S712, if a CSI report is required from the UE, the base station may use a lower-layer triggering message to enable semi-static CSI reporting or trigger a non-periodic CSI reporting. As previously described, the triggering message for enabling semi-static CSI reporting or triggering non-periodic CSI reporting may consist of only one message (e.g., MAC-CE or DCI) or two messages (e.g., MAC-CE and DCI). It should be noted that Figure 7 illustrates a form in which a single message is transmitted for the convenience of explanation. For example, if two messages are used, the base station may transmit the MAC-CE message first and then the DCI.

[0156] In step S714, the base station may transmit semi-static or non-periodic CSI-RS to the UE based on RRC signaling messages and lower-layer triggering messages regarding legacy CSI reports. In step S714, the UE may receive semi-static CSI-RS or non-periodic CSI-RS based on RRC signaling messages and lower-layer triggering messages regarding legacy CSI reports. As previously described, since the base station uses the MIMO method in 5G NR, CSI-RS may be transmitted through multiple beams. Therefore, the UE may receive CSI-RS corresponding to each beam through multiple beams transmitted by the base station. The UE may measure the received CSI-RS. The UE may generate a CSI report message based on the measured values ​​of the CSI-RS and the received RRC signaling messages and lower-layer triggering messages regarding legacy CSI reports.

[0157] In step S716, the UE can transmit the generated CSI report message to the base station. Since the embodiment of FIG. 7 is a case where semi-static or non-periodic legacy CSI measurement reporting is performed, the UE can transmit the CSI report message to the base station based on the legacy CSI reporting period set in the RRC signaling message and / or the lower layer triggering message. In this case, the CSI report message transmitted over the uplink can be transmitted through a channel (e.g., PUCCH or PUSCH) set by the lower layer triggering message (e.g., MAC-CE message or DCI) transmitted from the base station to the UE in step S712.

[0158] Subsequently, if AI / ML CSI reporting is required, the base station may transmit a lower-layer triggering message (e.g., a MAC-CE message or DCI) to the UE in step S720. The lower-layer triggering message transmitted in step S720 may include instruction information for activating (or triggering) one of the AI / ML CSI reporting settings (e.g., any one of the periodic AI / ML CSI reporting setting, semi-static AI / ML CSI reporting setting, or non-periodic AI / ML CSI reporting setting) set in the RRC message transmitted from the base station to the UE in step S710. Additionally, the lower-layer triggering message transmitted in step S720 may include instruction information for deactivating legacy CSI reporting settings. If the lower-level triggering message transmitted in step S720 does not include instruction information for disabling legacy CSI reporting settings, the UE may interpret the legacy CSI reporting settings as disabled based on instruction information for enabling (or triggering) one AI / ML CSI reporting setting, as described above.

[0159] In step S722, the base station may transmit CSI-RS for AI / ML to the UE. The CSI-RS for AI / ML may be transmitted via beams based on the AI / ML CSI reporting settings triggered in step S720. In step S722, the UE may receive the CSI-RS for AI / ML transmitted by the base station via multiple beams. The UE may measure the CSI-RS for AI / ML received via multiple beams and use the measured values ​​as input to the UE-side model. The UE may obtain AI / ML CSI (predicted CSI or inferred CSI) through the UE-side model. The UE may generate an AI / ML CSI reporting message containing the obtained AI / ML CSI. At this time, based on the settings of the upper layer or the settings (or instructions) of the lower layer, the AI / ML CSI reporting message may further include measured CSI values ​​in addition to the AI / ML CSI (predicted CSI or inferred CSI) obtained through the UE-side model.

[0160] In step S724, the UE can transmit the generated AI / ML CSI report message to the base station. If the generated AI / ML CSI report message is a message generated based on periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting settings, the UL channel may be a UL channel configured by the RRC configuration information. On the other hand, if the message is generated based on non-periodic reporting settings or PUSCH-based semi-static reporting settings, the UL channel may be a UL channel granted by the DCI. Therefore, the base station can receive the AI / ML CSI report message from the UE through the UL channel configured by the RRC configuration information or the DCI. The base station can monitor the prediction performance of the UE-side model based on the received AI / ML CSI report message.

[0161] The second method of the present disclosure described above may perform legacy CSI reporting or AI / ML CSI reporting based on information that is activated or triggered by a lower-layer triggering message (e.g., MAC-CE message, and / or DCI) within the scope of a setting message by an RRC signaling message. In an environment where a method is used in which a lower-layer triggering message must be transmitted, as shown in FIGS. 6 and 7, if the base station does not transmit the lower-layer triggering message, the base station may not transmit CSI-RS for AI / ML. In such cases where a new triggering message is not transmitted, the switching of the CSI reporting method may not be performed.

[0162] In Figures 6 and 7, the behavior of switching to AI / ML CSI reporting while legacy CSI reporting is being performed is illustrated. However, it should be noted that the opposite is also possible. In other words, while AI / ML CSI reporting is being performed, it may be switched to legacy CSI reporting based on a lower-level triggering message.

[0163] For example, in the example of FIG. 6, when the UE performs steps S622 and S624 (repeatedly if periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting is set, and once if non-periodic AI / ML CSI reporting is set), if a lower layer triggering message as exemplified in step S620 is received instructing (or setting) to disable AI / ML CSI reporting and enable legacy CSI reporting, the base station and the UE may repeat steps S612 and S614.

[0164] In the case of FIG. 7, while the UE performs steps S722 and S724 (repeatedly if periodic AI / ML CSI reporting or semi-static AI / ML CSI reporting is set, and once if non-periodic AI / ML CSI reporting is set), the base station can switch from the AI / ML CSI reporting method to the legacy CSI reporting method by sending a lower-layer triggering message such as S712 to the UE. Accordingly, the UE and the base station can perform steps S714 through S716.

[0165] The second method described above may be a method in which the RRC signaling message is transmitted only once (if the legacy CSI report and the AI / ML CSI report are each set as RRC signaling messages, each RRC signaling message is transmitted only once), and the transition from the legacy CSI report to the AI / ML CSI report or the transition from the AI / ML CSI report to the legacy CSI report is made by the lower-level signaling message.

[0166] The second method described above may be applied when the UE performs only legacy CSI reporting or only AI / ML CSI reporting by means of the RRC settings related to CSI reporting. Here, the meaning that only legacy CSI reporting operation is possible may mean that legacy CSI reporting can be performed as one of various methods determined by a combination of the type of legacy CSI reporting method, the UL channel used for legacy CSI reporting, and the type of CSI-RS measured for CSI reporting. Likewise, the meaning that only AI / ML CSI reporting operation is possible may mean that AI / ML CSI reporting can be performed as one of various methods determined by a combination of the type of AI / ML CSI reporting method, the UL channel used for AI / ML CSI reporting, and the type of CSI-RS measured for AI / ML CSI reporting.

[0167] As previously described, MAC-CE messages or DCIs may explicitly disable or implicitly instruct other CSI reporting methods when a specific CSI reporting method is configured. For example, if legacy CSI reporting is enabled or triggered by a MAC-CE message or DCI, AI / ML CSI reporting may include an indicator explicitly instructing it to be disabled, or it may cause AI / ML CSI reporting to be interpreted as implicitly disabled based on the activation of legacy CSI reporting.

[0168] The two methods of the first embodiment described above describe cases where only one of the legacy CSI reporting method or the AI / ML CSI reporting method is supported. However, it should be noted that even when both the legacy CSI reporting method and the AI / ML CSI reporting method are supported, they can be extended or combined with other methods by indicating active / inactive or enable / disable based on the description above.

[0169] For example, if legacy CSI reporting and AI / ML CSI reporting can be enabled simultaneously, the base station can resolve operational ambiguity of the UE by explicitly instructing the enable / disable for each CSI report. Additionally, if the base station can simultaneously instruct the enable / disable for one or more CSI reporting methods in the signaling procedure, there is an advantage of reducing not only the complexity of signaling but also latency. Accordingly, the first embodiment may include enable / disable indicator information for two or more CSI reports together in an RRC signaling message, MAC-CE message, or DCI containing execution instruction information for CSI reporting.

[0170] As another example, RRC signaling messages, MAC-CE messages, or DCIs for AI / ML CSI reporting separate from RRC signaling messages, MAC-CE messages, or DCIs may be used for legacy CSI reporting. In this case, each signaling message may include instruction information regarding permission or denial for each CSI report.

[0171] As another example, for some signaling, legacy CSI reporting and AI / ML CSI reporting may be operated in common, while for others, legacy CSI reporting and AI / ML CSI reporting may be operated separately. For instance, in the case of RRC signaling messages, configuration information for CSI reporting may be operated as a common signal, while signaling regarding the permission or denial of legacy CSI reporting or AI / ML CSI reporting may be operated separately (e.g., as a lower-level triggering signal). In this case, each lower-level triggering signal may contain only indicator information regarding the permission or denial of a single CSI report.

[0172] [Second Embodiment: Method for monitoring an AI / ML model starting from a base station]

[0173] In wireless communication systems, a two-sided or one-sided model may be used as the AI / ML radio (air) interface technology. As previously described, a two-sided model may involve the combined use of a base station-side model and a UE-side model, while a one-sided model may involve the use of only the base station-side model or only the UE-side model. To maintain the performance of each AI / ML model(s) used in these two-sided or one-sided models, a performance evaluation procedure regarding the accuracy of the output values ​​of each AI / ML model(s) may be required. To evaluate the performance of each AI / ML model(s), a procedure for monitoring the output of each AI / ML model(s) must be provided.

[0174] In the second embodiment described below, a method for evaluating the performance of each AI / ML model(s) is described. In the second embodiment of the present disclosure, a monitoring procedure is described for cases where the output result of each AI / ML model(s) is a CSI report result, and in particular, in the second embodiment of the present disclosure, a method for monitoring an AI / ML model initiated at a base station is described.

[0175] Figure 8 is a flowchart illustrating the monitoring method of an AI / ML model starting from a base station.

[0176] In step S810, the base station may configure configuration information for the monitoring operation of the AI / ML model as upper-layer configuration information and transmit it to the UE. The upper-layer configuration information may be transmitted, for example, by being included in an RRC signaling message. Since the monitoring procedure of the AI / ML model is initiated by the base station transmitting configuration information for the monitoring operation to the UE as in step S810, the procedure of FIG. 8 may be a procedure in which the monitoring of the AI / ML model is initiated by the base station.

[0177] In embodiments of the present disclosure, monitoring configuration information for an AI / ML model may be configured in various forms. For example, monitoring configuration information for an AI / ML model may be configured to be included within the AI / ML CSI reporting configuration. As another example, monitoring configuration information for an AI / ML model may be configured as separate configuration information and configured to be associated with the AI / ML CSI reporting configuration information. It may also be configured as monitoring configuration information for an AI / ML model in other forms.

[0178] In addition, the configuration of the AI / ML CSI Report Configuration (AI / ML CSI-ReportConfig) information can be configured in various forms. For example, the configuration of the AI / ML CSI Report Configuration (AI / ML CSI-ReportConfig) information may be identical to the configuration of the legacy CSI Report Configuration. As another example, the configuration of the AI / ML CSI Report Configuration (AI / ML CSI-ReportConfig) information may consist of some identical information and some different information compared to the configuration of the legacy CSI Report Configuration. As yet another example, the AI / ML CSI Report Configuration (AI / ML CSI-ReportConfig) information may consist of information that is completely different from the configuration of the legacy CSI Report Configuration. However, for the convenience of explanation, the present disclosure described below assumes that the configuration of the AI / ML CSI Report Configuration (AI / ML CSI-ReportConfig) information is configured in the same form as the CSI Report Configuration (CSI-ReportConfig) information used in the legacy CSI Report Configuration.

[0179] AI / ML CSI reporting configuration information may include one or more AI / ML CSI reporting configuration identifiers (AI / ML CSI-ReportConfigID). Each of the AI / ML CSI reporting configuration identifiers may be associated with the monitoring configuration information of an AI / ML model. Additionally, in the following description, the monitoring configuration information of an AI / ML model will be referred to as "ML monitoring configuration (ML-MonitorConfig)".

[0180] ML monitoring settings may include the following information:

[0181] - ML monitoring settings may include configuration information regarding transmission types, such as periodic, semi-static, or non-periodic transmission, for RS transmitted by a base station to evaluate the performance of an AI / ML model. Additionally, ML monitoring settings may include some or all of the resource information used for RS transmission. Resources used for RS transmission may include, for example, time resources and frequency resources. If resource information used for RS transmission is not directly included in the ML monitoring settings, information associated with resource-related information included in other settings may be included in the ML monitoring settings. For example, if the resource-related configuration information included in the ML monitoring settings is referred to as ML resource configuration (ML-ResourceConfig) information, the ML resource-related configuration information may directly include the resource-related configuration information of the RS transmitted by the base station or be configured to be associated with such information. In this case, the RS transmitted by the base station may refer to CSI-RS or SSB, etc.

[0182] - The ML monitoring settings may include quantity information that must be reported by measuring the RS transmitted by the base station, or setting information regarding the reported quantity of output values ​​obtained by using the measured RS as input to the UE-side model. In this case, the reported quantity information may include the type of information to be reported. For example, if the type of reported quantity included in the ML monitoring settings is a reported quantity related to legacy CSI reporting, it may include one or more types among CQI, PMI, CRI, SSBRI, LI, RI, and L1-RSRP. Furthermore, the reported quantity information included in the ML monitoring settings may additionally define various forms of modified values ​​that can be obtained through the UE-side model using the measured RS value. For example, the reported quantity information included in the ML monitoring settings may include one or more of predicted CQI, predicted PMI, predicted CRI, predicted SSBRI, predicted LI, predicted RI, and predicted L1-RSRP. As another example, the reported quantity information included in the ML monitoring setup may include one or more variant values ​​among the predicted CQI, predicted PMI, predicted CRI, predicted SSBRI, predicted LI, predicted RI, and predicted L1-RSRP. Here, the variant value may include the difference value (or the degree of difference, or the prediction accuracy, etc.) between the predicted value based on the previous measurement and the measurement at the time of prediction.

[0183] - The ML monitoring configuration may include UL channel resource information for reporting one or more quantity information obtained through a UE-side model to a base station. Here, one or more quantity information may be CSI reporting information (CSI prediction information or variation information based on CSI prediction) obtained through an ML model. The CSI reporting information obtained through the ML model described below may include variation information based on CSI prediction.

[0184] The information included in the ML monitoring settings mentioned above may be basic information for the monitoring operation of the AI / ML model. Therefore, in addition to the information mentioned above, the ML monitoring settings may include additional information necessary for monitoring the AI / ML model. For example, the ML monitoring settings may further include one or more of the following: configuration information required for the base station to transmit RS, configuration information required for the UE to receive and process RS, configuration information required for the UE to measure RS, configuration information required to be obtained through the UE-side model based on the value of RS measured by the UE, configuration information regarding the value of RS measured by the UE, configuration information regarding the value of RS modified by the UE, or configuration information required for the UE to report to the base station. Here, the value of RS measured by the UE or the value of RS modified by the UE may refer to the quantities reported by the UE to the base station based on RS.

[0185] The upper layer configuration information transmitted by the base station to the UE in step S810 may include one or more AI / ML CSI reporting configuration information and / or one or more ML monitoring reporting configurations associated therewith. The upper layer configuration information transmitted by the base station to the UE in step S810 may be the ML monitoring reporting configuration described above. Thus, the UE may receive the ML monitoring reporting configuration from the base station via an RRC signaling message in step S810. If the AI / ML CSI reporting configuration and the ML monitoring reporting configuration are associated, the upper layer configuration information transmitted by the base station to the UE in step S810 may be the AI / ML CSI reporting configuration information. Thus, the ML monitoring reporting configuration may be identified from the AI / ML CSI reporting configuration information.

[0186] If monitoring of the AI / ML model is required, the base station can trigger the monitoring operation of the AI / ML model by transmitting a lower-layer triggering signal to the UE, as in step S812. The lower-layer triggering signal may be activated or triggered by, for example, either MAC-CE or DCI, or triggered using both MAC-CE and DCI. Although step S812 in Fig. 8 is illustrated as a single signaling, this is for the sake of simplification of the drawing, and it should be noted that if both MAC-CE and DCI are used, two signalings are performed.

[0187] When a UE triggers an ML monitoring action based on a lower-level triggering signal, such as in step S812, the triggering may vary depending on the configuration of the ML monitoring settings. For example, if there is only one ML monitoring setting associated with AI / ML CSI reporting setting information configured for CSI reporting, the activation / deactivation of the ML monitoring setting may be indicated by the instruction (or setting) of activation / deactivation of the AI / ML CSI reporting setting information. On the other hand, if there are multiple ML monitoring settings associated with AI / ML CSI reporting setting information configured for CSI reporting, the lower-level triggering signal may trigger one of the multiple ML monitoring settings.

[0188] In step S814, the base station may transmit a CSI-RS to the UE for monitoring the AI / ML model. As previously described, if the ML monitoring settings are associated with AI / ML CSI reporting setting information configured for CSI reporting, the CSI-RS for monitoring the AI / ML model may be a CSI-RS based on the AI / ML CSI reporting setting information. Since Figure 8 describes the procedure for monitoring the AI / ML model, it should be noted that the CSI-RS transmitted by the base station to the UE in step S814 is referred to as the CSI-RS for monitoring the AI / ML model.

[0189] In step S814, the UE can receive CSI-RS from the base station and measure the received CSI-RS. The UE can generate a CSI report message based on the ML monitoring settings. In this case, the generated CSI report message may be an AI / ML CSI report message. The AI / ML CSI report message may include CSI predicted (or inferred) by an AI / ML model. Additionally, the CSI report message may further include ML monitoring information based on the ML monitoring reporting settings.

[0190] In step S816, the UE can transmit a generated CSI report message to the base station. Therefore, in step S816, the base station can receive the CSI report message transmitted by the UE. Subsequently, the base station can evaluate the performance of the AI / ML model based on the received CSI report message. The present disclosure does not impose specific restrictions on the method for evaluating the performance of the AI / ML model. In other words, the present disclosure provides a monitoring procedure for evaluating the performance of an AI / ML model, and it should be noted that various methods may be used for evaluating the performance of the AI / ML model.

[0191] <2-1 Example>

[0192] According to the second embodiment described above, the base station can transmit ML monitoring settings to the UE (step S810) and then trigger AI / ML model monitoring using a lower layer triggering signal (step S812). At this time, AI / ML model monitoring may be triggered using either MAC-CE or DCI as described above, or through a combination of MAC-CE and DCI. The second-1 embodiment described below may be a case where AI / ML monitoring is triggered using only MAC-CE.

[0193] FIG. 9 is a conceptual diagram illustrating the configuration of a MAC-CE for triggering AI / ML model monitoring according to the second-1 embodiment of the present disclosure.

[0194] Referring to FIG. 9, the MAC-CE for triggering AI / ML model monitoring can be composed of three octets. Each octet can be composed of 8 bits. The most significant bit (MSB) of Octet 1 can be a reserved (R) bit, the 5 consecutive bits below the MBS of Octet 1 can be a serving cell ID, and the remaining 2 bits of Octet 1 can be a bandwidth part (BOP) ID. The serving cell ID, consisting of 5 bits, can represent the identifier of the serving cell to which the MAC CE is applied. The BWP ID, consisting of 2 bits, can represent the uplink (UL) BWP to which the MAC CE is applied as a code point in the DCI bandwidth part indicator field. Additionally, the reserved (R) bits exemplified in FIG. 9 can all be set to a zero value.

[0195] In octet 2, the two consecutive bits from the MSB are reserved (R) bits, and the remaining six consecutive bits (901) may consist of AI / ML CSI reporting configuration identifier information. The AI / ML CSI reporting configuration identifier information included in octet 2 may, for example, be an AI / ML CSI reporting configuration identifier corresponding to configuration information for a currently running CSI report. As another example, the AI / ML CSI reporting configuration identifier information included in octet 2 may be an AI / ML CSI reporting configuration identifier corresponding to an ML model that requires monitoring for a specific AI / ML CSI reporting among the pre-configured CSI reporting configurations, regardless of the currently running CSI reporting configuration.

[0196] The N field (910), consisting of 8 bits of octet 3, can be composed of N0, N1, N2, N3, N4, N5, N6, N7 sequentially from the least significant bit (LSB) to the most significant bit (MSB). The N field (910) of octet 3 can be used to indicate one of a plurality of ML monitoring settings corresponding to a specific CSI reporting setting identifier.

[0197] The MAC-CE exemplified in FIG. 9 may be an example where the serving cell identifier, BWP identifier, and AI / ML CSI reporting setting identifier included in Octet 1 and Octet 2, respectively, are all included. However, FIG. 9 is intended to aid in understanding the embodiments of the present disclosure and may be configured to include only some of the identifiers above. If a specific identifier among the identifiers exemplified in FIG. 9 is not included, the bits of that identifier may be set as reserved bits.

[0198] Additionally, in the example of FIG. 9, the N field (910) is illustrated as an example where the entire octet is used, but only a portion of the octet may be used to indicate ML monitoring settings. As another example, two or more octets may be used to indicate ML monitoring settings. The N field (910) exemplified in FIG. 9 may be used in different ways as follows.

[0199] <N 필드의 사용의 제1 실시예>

[0200] The N field (910) exemplified in FIG. 9 can be mapped to a specific ML monitoring configuration (ML-monitoringConfig). If multiple ML monitoring configurations exist, each ML monitoring configuration can be distinguished by an ML monitoring configuration identifier (ML-monitoringConfigID) assigned an identifier (ID). In this way, when multiple ML monitoring configurations are distinguished by an ML monitoring configuration identifier (ML-monitoringConfigID), each bit of the N field (910) can be mapped 1:1 to a single ML monitoring configuration identifier (ML-monitoringConfigID).

[0201] For example, as illustrated in FIG. 9, if the N field (910) is composed of 8 bits, 8 ML monitoring setting identifiers can be mapped 1:1 to each bit of the N field (910). Each of the 8 ML monitoring settings distinguished by the ML monitoring setting identifier may be an ML monitoring setting associated with the AI / ML CSI reporting setting of octet 2.

[0202] If, while performing AI / ML CSI reporting, a MAC-CE is received that does not include the AI / ML CSI reporting configuration identifier in octet 2 exemplified in FIG. 9, the N field (910) may be mapped 1:1 with the ML monitoring configuration identifiers (ML-monitoringConfigID) of the ML monitoring configurations associated with the AI / ML CSI reporting configuration identifier corresponding to the AI / ML CSI reporting currently being performed.

[0203] The mapping between the N field (910) and the ML monitoring setting identifier can be mapped in various forms. For example, it can be mapped sequentially from the lowest identifier in order from LSB to MSB. In this case, if a specific bit of the N field (910) has a first value (e.g., zero value), it may indicate deactivation, and if it has a second value (e.g., 1 value), it may indicate activation.

[0204] As a specific example based on the above example, if the value of N0 is 1 and the values ​​of N1 through N7 are 0, the ML monitoring setting with the lowest identifier corresponding to N0 can be enabled, and the remaining ML monitoring settings can be disabled. When used in this way through mapping, it is possible to instruct the enable / disable of each of the eight ML monitoring settings. In addition, in this case, it is also possible to enable or disable two or more ML monitoring settings.

[0205] <N 필드의 사용의 제2 실시예>

[0206] The N field (910) exemplified in FIG. 9 consists of 8 bits, so it can be composed of 256 code points. In this case, each of the 256 code points can be mapped to multiple ML monitoring configuration identifiers (ML-monitoringConfigID) associated with a specific AI / ML CSI reporting configuration. In this case, since the N field (910) indicates only one code point, only monitoring corresponding to the ML monitoring configuration corresponding to the corresponding code point can be enabled.

[0207] In the above example, one of the following methods may be used to instruct MAC-CE to enable or disable monitoring corresponding to a specific ML monitoring setting.

[0208] In the first method, 128 of the 256 code points may instruct the activation of monitoring actions corresponding to 128 ML monitoring settings, and the remaining 128 code points may instruct the deactivation of monitoring actions corresponding to the same ML monitoring settings.

[0209] In a second method, one of the bits reserved in the MAC-CE can be used to identify whether the MAC-CE is a MAC-CE intended to instruct the activation or deactivation of an ML monitoring setting. In this case, an ML monitoring setting having a specific ML monitoring setting identifier among 256 code points can be indicated according to the setting values ​​of the 8 bits of the N field (910), and the MAC-CE can be indicated to instruct the activation or deactivation of an ML monitoring operation by using the reserved R bit of the MAC-CE. The R bit used here may be, for example, the R bit which is the MSB of octet 1.

[0210] <2-2 Example>

[0211] According to the second embodiment described above, the base station can transmit ML monitoring settings to the UE (step S810) and then trigger ML monitoring using a lower layer triggering signal (step S812). At this time, AI / ML monitoring may be triggered using either MAC-CE or DCI as described above, or through a combination of MAC-CE and DCI. The second-2 embodiment described below may be a case where ML monitoring is triggered using only DCI.

[0212] When AI / ML monitoring is triggered using only a DCI, the DCI may indicate one of the ML monitoring settings associated with the currently configured or AI / ML CSI reporting setting information. In order for the DCI to indicate an ML monitoring setting, the present disclosure may define and use a new field within the DCI. A DCI having such a new field may also be defined as a new DCI format.

[0213] As another example, some of the DCI formats currently defined in 5G NR may be used. For example, specific monitoring actions can be triggered using DCIs that have a CSI request field, such as DCI format 0_1, DCI format 0_2, and DCI format 0_3. In this case, when the base station transmits the DCI to the UE, it may scramble the DCI with a specific radio network temporary identifier (RNTI) to indicate that the CSI request field included in the DCI is an instruction for an ML monitoring action.

[0214] The procedure for scrambling DCI into a specific RNTI may refer to a procedure for transforming the cyclic redundancy check (CRC) bits included in the DCI for error detection through an exclusive OR operation using specific RNTI values. Therefore, the UE can receive the DCI scrambled into a specific RNTI.

[0215] When a UE receives a DCI scrambled with a specific RNTI from a base station, the UE can descramble the DCI using the specific RNTI to obtain the DCI. Here, obtaining the DCI means performing an exclusive OR operation on the CRC bits with the specific RNTI and checking the CRC of the received DCI, and confirming that there are no errors. When the DCI is obtained by descrambling with a specific RNTI in this way, the UE can identify that the CSI request field included in the DCI is an instruction for an AI / ML monitoring operation.

[0216] In this case, the UE can confirm that one of the specific ML monitoring settings is indicated by the CSI request field of the DCI. At this time, the ML monitoring setting indicated by the CSI request field of the DCI may be one of the ML monitoring settings associated with the currently configured or running AI / ML CSI reporting setting information.

[0217] <2-3rd Example>

[0218] According to the second embodiment described above, the base station can transmit ML monitoring settings to the UE (step S810) and then trigger AI / ML model monitoring using a lower layer triggering signal (step S812). At this time, AI / ML model monitoring can be triggered using either MAC-CE or DCI as described above, or through a combination of MAC-CE and DCI. The second-third embodiment to be described below may be a case where AI / ML model monitoring is triggered through a combination of MAC-CE and DCI.

[0219] Embodiment 2-3 may be a combination of Embodiments 2-1 and 2-2 described above. In Embodiment 2-3, the base station may configure MAC-CE according to one of the methods described in Embodiment 2-1 and transmit it to the UE. Additionally, the base station may configure DCI according to one of the methods described in Embodiment 2-2 and transmit it to the UE. The UE receives MAC-CE according to one of the methods described in Embodiment 2-1 and receives DCI according to one of the methods described in Embodiment 2-2, thereby triggering an AI / ML monitoring procedure.

[0220] The base station can select some ML monitoring settings from among multiple ML monitoring settings associated with specific AI / ML CSI reporting settings, which are specific AI / ML CSI reporting setting information, using MAC-CE. The base station can then configure the selected ML monitoring settings to be included in MAC-CE. The base station can direct one of the ML monitoring settings included in MAC-CE via DCI. Through this, the base station can direct the ML monitoring setting to the UE.

[0221] The method of including multiple ML monitoring settings in MAC-CE can map each bit of the N field (910) to a specific ML monitoring setting in a 1:1 manner, as previously described in FIG. 9. In this case, multiple ML monitoring settings can be enabled by setting two or more bits in the N field (910) of MAC-CE to a second value (e.g., a value of 1). Then, DCI can indicate one of the ML monitoring settings enabled in MAC-CE. In other words, ML monitoring setting(s) set to a first value (e.g., a value of 0) in MAC-CE can be prevented from being indicated by the DCI field. This method may be such that the code points of the CSI request field of DCI indicate the ML monitoring setting selected by MAC-CE.

[0222] [Third Embodiment: Method for monitoring AI / ML models starting in the UE]

[0223] As previously explained, when AI / ML radio (air) interface technology is used in a wireless communication system, a performance evaluation procedure regarding the accuracy of the output values ​​of each AI / ML model(s) is required to maintain the performance of the AI / ML model(s), and to evaluate the performance of each AI / ML model(s), a procedure for monitoring the output of each AI / ML model(s) must be provided. Based on this, a monitoring method initiated by a base station was described in the second embodiment described above. In the third embodiment to be described below, a monitoring method for an AI / ML model initiated by a UE is described.

[0224] FIG. 10 is a flowchart illustrating a first embodiment of a monitoring method for an AI / ML model starting from a UE.

[0225] In step S1010, the base station may transmit upper-layer configuration information containing conditions for AI / ML monitoring triggering to the UE. The upper-layer configuration information containing conditions for AI / ML monitoring triggering may be transmitted, for example, by being included in an RRC signaling message. The conditions for AI / ML monitoring triggering may include one or more of specific events, for example, whether the measured value of RS falls within a specific range, whether the value calculated (or modified) by a preset calculation method of the measured value of RS falls within a specific range, whether the output value of the UE-side model falls within a specific range, whether the value calculated (or modified) by a preset calculation method of the output value of the UE-side model falls within a specific range, or specific conditions. Here, specific conditions may refer to a threshold value set by the base station, a target value to be compared with the threshold value, and a condition for triggering (for example, a condition where the target value is greater than or less than the threshold value).

[0226] Additionally, in step S1010, the base station may further transmit configuration information for the monitoring operation of the AI / ML model to the UE. The configuration information for the monitoring operation of the AI / ML model may be the same configuration information described in the second embodiment above. Therefore, it should be noted that redundant descriptions are omitted.

[0227] In step S1010, the UE can receive upper-layer configuration information from the base station that includes conditions for AI / ML monitoring triggering.

[0228] The UE may immediately execute the condition for AI / ML monitoring triggering received in step S1010, or be triggered by a specific signal. Such triggering may be triggered by lower-level signaling message(s) as previously described.

[0229] When the UE receives a condition for AI / ML monitoring triggering (or when a condition for AI / ML monitoring triggering is triggered), it can check whether the condition for AI / ML monitoring triggering is satisfied. It should be noted that in FIG. 10, the procedure for the UE to check whether the condition for AI / ML monitoring triggering is satisfied is not illustrated.

[0230] When the conditions for triggering AI / ML monitoring are met, in step S1012, the UE may transmit an AI / ML model monitoring request signal to the base station. The AI / ML monitoring request signal transmitted by the UE to the base station in step S1012 may be transmitted, for example, via a PUCCH. The PUCCH to which the AI / ML monitoring request signal is transmitted may be a UL resource configured to be periodic or semi-persistent by an RRC signaling message in step S1010. The AI / ML monitoring request signal transmitted via the PUCCH may indicate that monitoring of the AI / ML model currently being used between the UE and the base station is required. In other words, it may be a request for monitoring of the UE-side model corresponding to the currently configured or currently operating AI / ML CSI reporting configuration.

[0231] If the PUCCH to which the AI / ML monitoring request signal is transmitted is a UL resource configured to be periodic or semi-static, the request indicator for monitoring of the UE-side model transmitted through the PUCCH may be composed of 1 bit. In this case, if the request indicator for monitoring of the UE-side model has a first value (e.g., zero), it may indicate that AI / ML monitoring of the UE-side model is not required, and if it has a second value (e.g., 1), it may indicate that AI / ML monitoring of the UE-side model is required. Additionally, if the PUCCH to which the AI / ML monitoring request signal is transmitted is a UL resource configured to be periodic or semi-static, step S1012 exemplified in FIG. 10 may be an example of the case where the request indicator for monitoring of the UE-side model has a second value (e.g., 1). Alternatively, it may be operated as 'Positive' (monitoring request) if the terminal transmits a signal from the PUCCH resource, and 'Negative' (no monitoring request) if it does not transmit.

[0232] If the PUCCH to which the AI / ML monitoring request signal is transmitted is a UL resource configured to be periodic or semi-static, the monitoring request indicator for the UE-side model transmitted through the PUCCH may indicate whether monitoring is required using a preset sequence. In this case, the monitoring request indicator for the UE-side model may consist of 2 bits or more of information.

[0233] Meanwhile, if additional information needs to be provided in addition to the AI / ML monitoring request signal, the UE may request the base station to allocate additional UL resources. In response to receiving the request for additional UL resource allocation from the UE, the base station may allocate additional UL resources to the UE. In this case, the request information for additional UL resources may be implicitly or explicitly indicated by the PUCCH transmitted by the UE in step S1012. The PUCCH may be configured and operated as a PUCCH capable of transmitting information of 2 bits or more. When additional UL resources are allocated from the base station, the UE may further transmit information to the base station regarding specific AI / ML models corresponding to the AI / ML CSI reporting settings requiring monitoring. In this case, the additional UL resources allocated to the UE by the base station may be PUSCH. The UE may transmit information regarding specific AI / ML models from the UE to the base station via the payload of the PUSCH or via MAC-CE.

[0234] Additionally, the AI / ML monitoring request signal transmitted by the UE to the base station in step S1012 may be transmitted piggy-back on other UL channels. For example, the AI / ML monitoring request may be transmitted along with the PUCCH or PUSCH used for CSI reporting. As another example, it may be transmitted along with the scheduling request (RS) transmitted by the UE to the base station. Furthermore, UL channels transmitted through additional UL resources may be transmitted piggy-back through channels where other PUSCHs used for CSI reporting or PUSCHs for data transmission exist.

[0235] In step S1012, the base station may receive an AI / ML monitoring request signal from the UE. Since step S1012 involves the UE transmitting a request for an AI / ML monitoring operation to the base station, the embodiment of FIG. 10 may be an AI / ML monitoring method initiated by the UE.

[0236] In step S1014, the base station may transmit a lower-layer triggering signal to the UE to instruct it to perform monitoring of the AI / ML model. The lower-layer triggering signal transmitted by the base station in step S1014 may be the lower-layer triggering signal described in the preceding second embodiment. In other words, it may be the MAC-CE described in the second-1 embodiment, the DCI described in the second-2 embodiment, or any one of the combination of MAC-CE and DCI described in the second-3 embodiment.

[0237] In step S1016, the UE and the base station may perform an AI / ML monitoring procedure. The AI / ML monitoring procedure may include the CSI-RS transmission for AI / ML and the transmission of AI / ML CSI report messages described in FIG. 6 and / or FIG. 7 of the first embodiment. Step S1016 may be, for example, the procedure described in steps S622 through S624 of FIG. 6 or the procedure described in steps S722 through S724 of FIG. 7.

[0238] FIG. 11 is a flowchart illustrating a second embodiment of a monitoring method for an AI / ML model starting from a UE.

[0239] In step S1110, the base station may transmit upper-layer configuration information containing conditions for triggering AI / ML model monitoring to the UE. The upper-layer configuration information containing conditions for triggering AI / ML model monitoring may be transmitted, for example, by being included in an RRC signaling message. The conditions for triggering AI / ML model monitoring may include one or more of specific events, for example, whether the measured value of RS falls within a specific range, whether the value calculated (or modified) by a preset calculation method of the measured value of RS falls within a specific range, whether the output value of the UE-side model falls within a specific range, whether the value calculated (or modified) by a preset calculation method of the output value of the UE-side model falls within a specific range, or specific conditions. Here, specific conditions may refer to a threshold value set by the base station, a target value to compare with the threshold value, and a condition for triggering (for example, a condition where the target value is greater than or less than the threshold value).

[0240] In step S1110, the UE can receive upper-layer configuration information from the base station that includes conditions for triggering AI / ML model monitoring.

[0241] The UE may immediately execute the condition for triggering AI / ML model monitoring received in step S1110, or be triggered by a specific signal. Such triggering may be triggered by lower-level signaling message(s) as previously described.

[0242] When the UE receives a condition for triggering AI / ML model monitoring (or when a condition for triggering AI / ML model monitoring is triggered), it can check whether the condition for triggering AI / ML monitoring is satisfied. It should be noted that in FIG. 11, the procedure for the UE to check whether the condition for triggering AI / ML model monitoring is satisfied is not illustrated.

[0243] When the conditions for triggering AI / ML model monitoring are met, in step S1112, the UE can transmit an AI / ML model monitoring request signal to the base station through the first UL channel. Then, in step S1114, the UE can transmit AI / ML model identification information to the base station through the second UL channel.

[0244] The first UL channel may be a channel for transmitting a request for AI / ML model monitoring for a UE-side model corresponding to an AI / ML CSI reporting setting that is currently set or is currently running. In other words, the first UL channel may be transmitted in the same manner as the transmission of the AI / ML model monitoring request signal described above in FIG. 10. Therefore, it should be noted that the settings of the first UL channel and the information transmitted, etc., may be the same as those described in FIG. 10, so a redundant explanation is omitted.

[0245] The second UL channel can transmit information regarding a specific AI / ML model corresponding to an AI / ML CSI report setting requiring monitoring. In other words, it may be information transmitted via the additional UL resource described earlier in FIG. 10. Therefore, the information transmitted via the second UL channel may be the same as described earlier in FIG. 10. However, there is a difference in that the second UL channel may be a channel pre-allocated by the upper layer setting information of step S1110. The second UL channel may be a PUSCH and may be configured periodically or semi-statically by the upper layer setting information. In this way, an embodiment such as FIG. 11 has the advantage of reducing the signaling of the additional UL resource allocation request and the allocation procedure of the additional UL resource described earlier in FIG. 10. Additionally, since the UE can transmit AI / ML model information to the base station without such procedures, the triggering of AI / ML model monitoring can be performed faster than the procedure exemplified in FIG. 10.

[0246] Meanwhile, the first UL channel and the second UL channel may be transmitted together with other UL channels (piggy-back), as previously described in FIG. 10. In this case, the first UL channel may be a PUCCH used for other CSI reports or a channel through which SRs are transmitted, and the second UL channel may be a PUSCH used for CSI reports or data transmission, or transmitted via MAC-CE.

[0247] In step S1112, the base station may receive an AI / ML model monitoring request signal from the UE, and in step S1114, may receive UE-side model information requiring ML monitoring. Since step S1112 involves the UE transmitting a request for an AI / ML model monitoring operation to the base station, the embodiment of FIG. 11 may also be an AI / ML model monitoring method initiated by the UE.

[0248] In step S1116, the base station may transmit a lower-layer triggering signal to the UE to instruct it to perform monitoring of the AI / ML model. The lower-layer triggering signal transmitted by the base station in step S1116 may be the lower-layer triggering signal described in the preceding second embodiment. In other words, the lower-layer triggering signal may be the MAC-CE described in the second-1 embodiment, the DCI described in the second-2 embodiment, or any one of the combination of MAC-CE and DCI described in the third embodiment.

[0249] In step S1118, the UE and the base station may perform an AI / ML model monitoring procedure. The AI / ML model monitoring procedure may include the CSI-RS transmission for AI / ML and the transmission of AI / ML CSI report messages described in FIG. 6 and / or FIG. 7 of the first embodiment. Step S1118 may be, for example, the procedure described in steps S622 through S624 of FIG. 6 or the procedure described in steps S722 through S724 of FIG. 7.

[0250] Meanwhile, a base station that receives an AI / ML model monitoring request signal described in FIG. 10 or an ML model monitoring request signal through the first UL channel and the second UL channel described in FIG. 11 may operate differently from as described in FIG. 10 and FIG. 11. For example, a base station that receives an AI / ML model monitoring request signal may not trigger AI / ML model monitoring through a lower-layer triggering signal, but may change the CSI reporting method or change the settings for CSI reporting. Changing the settings for CSI reporting may involve updating the CSI reporting setting information through an RRC signaling message or changing the CSI reporting settings currently set in the RRC signaling message to another CSI reporting setting through lower-layer signaling.

[0251] On the other hand, if the UE does not receive instructions for AI / ML model monitoring from the base station within a preset time window, it may retransmit the AI / ML model monitoring request signal to the base station. If the UE does not receive instructions for AI / ML model monitoring from the base station within the preset time window but receives instructions related to CSI reporting, the UE may stop retransmitting the AI / ML model monitoring request signal. In other words, the retransmission of the AI / ML model monitoring request signal may be performed if the UE does not receive instructions for AI / ML model monitoring from the base station within the preset time window, and also does not receive instructions related to CSI reporting within that time window.

[0252] The embodiments described above may be implemented in various modified forms through simple combinations of each embodiment or combinations of modified forms. For example, a second or third embodiment may be performed in combination with a first embodiment. As another example, the second and third embodiments may be configured together. The second embodiment may be performed when the base station determines that monitoring of the AI / ML model is necessary, and the third embodiment may be performed by conditions pre-set by the base station to the UE as described above. Therefore, when the second and third embodiments are combined, the upper layer configuration information may be provided to the UE as the configuration information described in the second embodiment and the configuration information described in the third embodiment, respectively. It should be noted that the embodiments exemplified in this disclosure are capable of various combinations.

[0253] The operation of the method according to an embodiment of the present disclosure can be implemented as a computer-readable program or code on a computer-readable recording medium. A computer-readable recording medium includes all types of recording devices in which information that can be read by a computer system is stored. Additionally, the computer-readable recording medium may be distributed across networked computer systems, allowing the computer-readable program or code to be stored and executed in a distributed manner.

[0254] In addition, computer-readable recording media may include hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Program instructions may include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0255] Some aspects of the present disclosure have been described in the context of a device, but may also be described according to a corresponding method, wherein a block or device corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method may also be described according to a corresponding block or item or a feature of a corresponding device. Some or all of the method steps may be performed by (or using) a hardware device, such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, at least one of the most important method steps may be performed by such a device.

[0256] In the embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the methods described herein. In the embodiments, the field-programmable gate array may operate with a microprocessor to perform one of the methods described herein. Generally, it is preferable that the methods be performed by some hardware device.

[0257] Although the present disclosure has been described with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the present disclosure without departing from the spirit and scope of the present disclosure as set forth in the following claims.

Claims

Claim 1 A method of user equipment (UE), comprising the steps of: receiving a first message from a base station including an Artificial Intelligence (AI) / Machine Learning (ML) channel state information (CSI) reporting setting, wherein the AI / ML CSI reporting setting is associated with one ML monitoring reporting setting; receiving a reference signal (RS) set based on instructions from the base station to activate the AI / ML CSI reporting setting; obtaining a CSI prediction value by using a measurement value of the received RS as input to an ML model; and transmitting a second message generated based on the CSI prediction value to the base station, wherein the second message further comprises ML monitoring information based on the ML monitoring reporting setting associated with the AI / ML CSI reporting setting, and the ML monitoring information is determined based on the CSI prediction value and / or the measurement value of the RS. Claim 2 A method of the UE according to claim 1, wherein the ML monitoring report setting comprises one or more of a transmission type of RS for monitoring the ML model, a transmission resource of the RS, or a reporting quantity. Claim 3 A method of the UE according to claim 2, wherein the reported quantity includes prediction accuracy information, and the prediction accuracy information is determined based on a predicted value for a first time instance using the ML model and a value of measuring the RS at the first time point. Claim 4 A method of the UE according to claim 1, wherein the activation of the AI / ML CSI reporting setting is indicated by a medium access control-control element (MAC-CE) message. Claim 5 A method of the UE according to claim 1, wherein the activation of the AI / ML CSI reporting setting is indicated by downlink control information (DCI). Claim 6 A method of the UE according to claim 1, wherein the activation of the AI / ML CSI reporting setting is indicated by a combination of a medium access control-control element (MAC-CE) message and downlink control information (DCI). Claim 7 A method of user equipment (UE), comprising: receiving a first message from a base station including ML monitoring report settings associated with an Artificial Intelligence (AI) / Machine Learning (ML) channel state information (CSI) reporting setting; receiving a reference signal (RS) from the base station based on an ML monitoring report setting indicated by the first ML monitoring report setting among the plurality of ML monitoring settings; obtaining a CSI prediction value by using a measurement value of the received RS as an input to an ML model; and transmitting a second message generated based on the CSI prediction value to the base station, wherein the second message further includes ML monitoring information based on the ML monitoring report setting associated with the AI / ML CSI reporting setting, and the ML monitoring information is determined based on the CSI prediction value and / or the measurement value of the RS. Claim 8 A method of the UE according to claim 7, wherein each of the plurality of ML monitoring reporting settings comprises one or more of a transmission type of RS for monitoring the ML model, a transmission resource of the RS, or a reporting quantity. Claim 9 A method of the UE according to claim 8, wherein the reported quantity includes prediction accuracy information, and the prediction accuracy information is determined based on a predicted value for a first time instance using the ML model and a value of measuring the RS at the first time point. Claim 10 A method of the UE according to claim 7, wherein the first ML monitoring report setting is an ML monitoring setting in which activation is indicated among a plurality of ML monitoring report settings included in a medium access control-control element (MAC-CE) message. Claim 11 The method of the UE according to claim 7, wherein the first ML monitoring report setting is an ML monitoring setting indicated by a first indicator included in downlink control information (DCI). Claim 12 A method of the UE according to claim 7, wherein the first ML monitoring report setting is indicated by a first indicator included in downlink control information (DCI) scrambled with a first radio network temporary identifier (RNTI). Claim 13 A method of the UE according to claim 7, wherein the first ML monitoring report setting is indicated by a combination of a medium access control-control element (MAC-CE) message and downlink control information (DCI), the MAC-CE message indicates the activation of one or more ML monitoring report settings among a plurality of ML monitoring report settings included in the first message, and the first field of the DCI indicates the first ML monitoring report setting among the one or more ML monitoring report settings whose activation is indicated by the MAC-CE. Claim 14 User equipment (UE) comprises at least one processor, wherein the at least one processor causes the UE to: receive a first message from a base station comprising ML monitoring report settings associated with an Artificial Intelligence (AI) / Machine Learning (ML) channel state information (CSI) reporting setting; receive a reference signal (RS) from the base station configured based on an ML monitoring report by the first ML monitoring report setting among the plurality of ML monitoring settings; obtain a CSI prediction value by using a measurement value of the received RS as input to an ML model; and cause the UE to transmit a second message generated based on the CSI prediction value to the base station, wherein the second message further comprises ML monitoring information based on the ML monitoring report setting associated with the AI / ML CSI reporting setting, and the ML monitoring information is determined based on the CSI prediction value and / or the measurement value of the RS. Claim 15 In claim 14, each of the plurality of ML monitoring report settings comprises one or more of a transmission type of RS for monitoring the ML model, a transmission resource of the RS, or a reporting quantity, UE. Claim 16 In claim 15, the reported quantity includes prediction accuracy information, and the prediction accuracy information is determined based on a predicted value for a first time instance using the ML model and a value of the RS measured at the first time point, UE. Claim 17 In claim 14, the first monitoring report setting is a UE that is an ML monitoring setting instructed to be activated among a plurality of ML monitoring settings included in a medium access control-control element (MAC-CE) message. Claim 18 In claim 14, the first ML monitoring report setting is an ML monitoring setting indicated by a first indicator included in downlink control information (DCI), UE. Claim 19 In claim 14, the first ML monitoring report setting is indicated by a first indicator included in downlink control information (DCI) scrambled with a first radio network temporary identifier (RNTI), UE. Claim 20 In claim 14, the first ML monitoring report setting is indicated by a combination of a medium access control-control element (MAC-CE) message and downlink control information (DCI), the MAC-CE message indicates the activation of one or more ML monitoring report settings among a plurality of ML monitoring report settings included in the first message, and the first field of the DCI indicates the first ML monitoring report setting among the one or more ML monitoring report settings whose activation is indicated by the MAC-CE, UE.

Images