Method and apparatus for transmitting and receiving signal in wireless communication system

Abstract

A terminal according to at least one of the embodiments disclosed in the present specification may: configure an artificial intelligence / machine learning (AI / ML) model; obtaining information on channel state information (CSI) prediction performance of the AI / ML model by monitoring of the AI / ML model; and performing a lifecycle management (LCM) related process for the AI / ML model based on the obtained information on the CSI prediction performance, where the LCM related process may include at least one of: (i) transmitting information requesting for updating a data set of the AI / ML model; (ii) transmitting information requesting configuration of a time interval in which an update of the AI / ML model is to be performed; (iii) sending information requesting to switch the AI / ML model; and (iv) transmitting information requesting fallback using the non-AI / ML-based operation.

Description

Technical Field

[0001] The present disclosure relates to a wireless communication system, and more particularly, to a method and apparatus for transmitting / receiving uplink / downlink wireless signals in a wireless communication system. Background Art

[0002] Generally, wireless communication systems are developing to provide communication services such as audio and data communication services, with diverse coverage over a wide range. Wireless communication is a multiple-access system capable of supporting communication with multiple users by sharing available system resources (e.g., bandwidth, transmit power, etc.). For example, a multiple-access system may be any of a code division multiple access (CDMA) system, a frequency division multiple access (FDMA) system, a time division multiple access (TDMA) system, an orthogonal frequency division multiple access (OFDMA) system, and a single-carrier frequency division multiple access (SC-FDMA) system. Summary of the Invention

[0003] Technical issues

[0004] An object of the present disclosure is to provide a method and apparatus for efficiently performing a wireless signal transmission / reception process.

[0005] Those skilled in the art will understand that the objectives that can be achieved using the present disclosure are not limited to those specifically described above, and the above and other objectives that can be achieved by the present disclosure will be more clearly understood from the following detailed description.

[0006] Technical Solution

[0007] In one aspect, a method performed by a terminal in a wireless communication system may include: configuring an artificial intelligence / machine learning (AI / ML) model; obtaining information about performance of channel state information (CSI) prediction of the AI / ML model through monitoring of the AI / ML model; and performing lifecycle management (LCM) related processes for the AI / ML model based on the obtained information about the performance of the CSI prediction.

[0008] The LCM-related process may include at least one of: (i) sending information for requesting a data set for updating of the AI / ML model, (ii) sending information for requesting setting of a period in which the updating of the AI / ML model is to be performed, (iii) sending information for requesting switching of the AI / ML model, or (iv) sending information for requesting fallback to non-AI / ML based operation.

[0009] The LCM-related process may further include at least one of: (v) transmitting information notifying completion of update or switching of the AI / ML model, or (vi) transmitting information for requesting activation of the updated or switched AI / ML model.

[0010] The information about the performance of the CSI prediction may be determined based on at least one of a change in a channel quality indicator (CQI) or a signal to interference and noise ratio (SINR) during a specific period, accuracy of the CSI prediction, or decoding performance of a downlink channel.

[0011] The specific period may be a CSI prediction window including the CSI prediction time point.

[0012] The terminal may transmit information for requesting a change in the size of a CSI prediction window for monitoring an AI / ML model.

[0013] Datasets can be used to train or retrain AI / ML models.

[0014] During a period in which updating of the AI / ML model is to be performed, transmission or reception of at least one signal may be suspended.

[0015] The AI / ML model can be a unilateral AI / ML model configured for the terminal through network signaling.

[0016] In another aspect, a computer-readable recording medium having recorded thereon a program for executing the above-described method may be provided.

[0017] In another aspect, a terminal for executing the above method may be provided.

[0018] In another aspect, a processing device for controlling a terminal for executing the above method may be provided.

[0019] In another aspect, a method performed by a base station in a wireless communication system may include: transmitting information for configuring an artificial intelligence / machine learning (AI / ML) model to a terminal; and performing a lifecycle management (LCM)-related procedure for the AI / ML model based on performance of channel state information (CSI) prediction of the AI / ML model configured for the terminal. The LCM-related procedure may include at least one of the following: (i) transmitting a data set for updating the AI / ML model, (ii) transmitting information for setting a period in which the AI / ML model is to be updated, (iii) transmitting information indicating switching of the AI / ML model, or (iv) transmitting information indicating fallback to non-AI / ML-based operation.

[0020] In another aspect, a base station for performing the above method may be provided.

[0021] Beneficial effects

[0022] According to the present disclosure, wireless signal transmission and reception can be efficiently performed in a wireless communication system.

[0023] Those skilled in the art will understand that the effects that can be achieved using the present disclosure are not limited to those specifically described above, and other advantages of the present disclosure will be more clearly understood from the following detailed description in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Illustrations illustrate physical channels used in a 3rd Generation Partnership Project (3GPP) system as an exemplary wireless communication system and a general signal transmission method using the same.

[0025] Figure 2 Figure 1 shows the radio frame structure.

[0026] Figure 3 A resource grid illustrating time slots is shown.

[0027] Figure 4 An exemplary mapping of physical channels in time slots is illustrated.

[0028] Figure 5 An exemplary physical downlink shared channel (PDSCH) and ACK / NACK transmission and reception process is illustrated.

[0029] Figure 6 Figure 1 illustrates an exemplary physical uplink shared channel (PUSCH) transmission process.

[0030] Figure 7 An example of a channel state information (CSI) related process is shown.

[0031] Figure 8 This diagram illustrates the concepts of artificial intelligence, machine learning (AI / ML), and deep learning.

[0032] Figures 9 to 12 Shows various AI / ML models for deep learning.

[0033] Figure 13 This is a diagram illustrating the splitting of AI reasoning.

[0034] Figure 14 is a diagram illustrating the framework of 3GPP Radio Access Network (RAN) intelligence.

[0035] Figures 15 to 17 Diagram of the AI model training and inference environment.

[0036] Figure 18 is a diagram illustrating predictions in the time domain based on a unilateral AI / ML model.

[0037] Figure 19 is a diagram illustrating a CSI prediction report according to an embodiment.

[0038] Figure 20is a diagram illustrating management of an AI / ML model for CSI prediction in a wireless communication system according to an embodiment.

[0039] Figure 21 is a diagram illustrating an operation of a terminal according to an embodiment.

[0040] Figure 22 is a diagram illustrating an operation of a base station according to an embodiment.

[0041] Figures 23 to 26 An example of the communication system 1 and wireless device applied to the present disclosure is illustrated.

[0042] Figure 27 An exemplary discontinuous reception (DRX) operation suitable for use with the present disclosure is illustrated. DETAILED DESCRIPTION

[0043] Embodiments of the present disclosure are applicable to various wireless access technologies, such as code division multiple access (CDMA), frequency division multiple access (FDMA), time division multiple access (TDMA), orthogonal frequency division multiple access (OFDMA), and single-carrier frequency division multiple access (SC-FDMA). CDMA can be implemented as a radio technology such as Universal Terrestrial Radio Access (UTRA) or CDMA2000. TDMA can be implemented as a radio technology such as Global System for Mobile Communications (GSM) / General Packet Radio Service (GPRS) / Enhanced Data Rates for GSM Evolution (EDGE). OFDMA can be implemented as a radio technology such as Institute of Electrical and Electronics Engineers (IEEE) 802.11 (Wireless Fidelity (Wi-Fi)), IEEE 802.16 (Worldwide Interoperability for Microwave Access (WiMAX)), IEEE 802.20, and Evolved UTRA (E-UTRA). UTRA is part of the Universal Mobile Telecommunications System (UMTS). 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) is part of Evolved UMTS (E-UMTS) using E-UTRA. LTE-Advanced (A) is an evolution of 3GPP LTE. 3GPP NR (New Radio or New Radio Access Technology) is an evolution of 3GPP LTE / LTE-A.

[0044] As more and more communication devices demand greater communication capacity, there is a need for enhanced mobile broadband communications compared to traditional radio access technologies (RATs). Furthermore, massive machine-type communications (MTC), which can connect multiple devices and objects to provide a variety of services anytime and anywhere, is another key consideration for next-generation communications. Discussions are also underway to design communication systems that take into account reliability- and latency-sensitive services and users. Consequently, discussions are underway to introduce new radio access technologies that take into account enhanced mobile broadband communications (eMBB), massive MTC, and ultra-reliable low-latency communications (URLLC). In this disclosure, for simplicity, this technology will be referred to as NR (New Radio or New RAT).

[0045] For the sake of brevity, 3GPP NR is mainly described, but the technical concept of the present disclosure is not limited thereto.

[0046] In this disclosure, the term "configure" can be replaced with "set," and the two can be used interchangeably. Furthermore, conditional expressions (e.g., "if," "in the case of," or "when") can be replaced with "based on" or "under the condition of." Furthermore, the operation or software / hardware (SW / HW) configuration of a user equipment (UE) / base station (BS) can be inferred / understood based on the satisfaction of the corresponding condition. When the processing on the receiving (or transmitting) side can be inferred / understood from the processing on the transmitting (or receiving) side during signal transmission / reception between a wireless communication device (e.g., a BS and a UE), the description of the processing can be omitted. For example, signal determination / generation / encoding / transmission on the transmitting side can be understood as signal monitoring, reception, decoding, and determination on the receiving side. Furthermore, when it is said that the UE performs (or does not perform) a specific operation, this can also be interpreted as the BS expecting / assuming (or not expecting / assuming) that the UE performs that specific operation. When it is said that the BS performs (or does not perform) a specific operation, this can also be interpreted as the UE expecting / assuming (or not expecting / assuming) that the BS performs that specific operation. In the following description, for the sake of convenience, sections, embodiments, examples, options, methods, solutions, etc. are distinguished from each other and indexed, which does not mean that each of them necessarily constitutes an independent invention or that each of them should only be implemented separately. Unless clearly contradicted, it can be inferred / understood that at least some of the sections, embodiments, examples, options, methods, solutions, etc. can be implemented in combination or can be omitted.

[0047] In a wireless communication system, a user equipment (UE) receives information from a base station (BS) via a downlink (DL) and transmits information to the BS via an uplink (UL). The information transmitted and received by the BS and the UE includes data and various control information, and various physical channels are used depending on the type and purpose of the information transmitted and received by the UE and the BS.

[0048] Figure 1This section describes physical channels used in the 3GPP NR system and a general signal transmission method using the channels.

[0049] When a UE is powered on again after powering off or enters a new cell, in step S101, the UE performs an initial cell search procedure (e.g., establishing synchronization with the base station). To this end, the UE receives a synchronization signal block (SSB) from the base station. The SSB includes the primary synchronization signal (PSS), the secondary synchronization signal (SSS), and the physical broadcast channel (PBCH). The UE establishes synchronization with the base station based on the PSS / SSS and obtains information such as the cell identity (ID). The UE can obtain broadcast information in the cell based on the PBCH. During the initial cell search, the UE can receive a downlink reference signal (RS) to monitor downlink channel status.

[0050] After the initial cell search, the UE may acquire more specific system information by receiving a physical downlink control channel (PDCCH) and a physical downlink shared channel (PDSCH) based on information of the PDCCH in step S102 .

[0051] In steps S103 to S106, the UE may perform a random access procedure to access the base station. For random access, the UE may transmit a preamble to the base station on a physical random access channel (PRACH) (S103) and receive a response message to the preamble on a PDCCH and a corresponding PDSCH (S104). In the case of contention-based random access, the UE may further perform a contention resolution procedure by transmitting a PRACH (S105) and receiving a PDCCH and a corresponding PDSCH (S106).

[0052] After the aforementioned process, the UE may receive the PDCCH / PDSCH (S107) and transmit the Physical Uplink Shared Channel (PUSCH) / Physical Uplink Control Channel (PUCCH) (S108) as a standard downlink / uplink signaling process. The control information sent from the UE to the base station is called uplink control information (UCI). UCI includes hybrid automatic repeat request acknowledgement / negative acknowledgement (HARQ-ACK / NACK), scheduling request (SR), channel state information (CSI), etc. CSI includes channel quality indicator (CQI), precoding matrix indicator (PMI), rank indicator (RI), etc. Although UCI is typically transmitted on the PUCCH, it can be transmitted on the PUSCH when control information and traffic data need to be transmitted simultaneously. In addition, UCI can be transmitted aperiodically via the PUSCH based on a request / command from the network.

[0053] Figure 2The radio frame structure is shown. In NR, uplink and downlink transmissions are configured in frames. Each radio frame has a length of 10ms and is divided into two 5ms half-frames (HF). Each half-frame is divided into five 1ms subframes (SF). A subframe is divided into one or more slots, and the number of slots in a subframe depends on the subcarrier spacing (SCS). Depending on the cyclic prefix (CP), each slot includes 12 or 14 orthogonal frequency division multiplexing (OFDM) symbols. When using a normal CP, each slot includes 14 OFDM symbols. When using an extended CP, each slot includes 12 OFDM symbols.

[0054] Table 1 exemplarily shows that the number of symbols per slot, the number of slots per frame, and the number of slots per subframe vary according to the SCS when a normal CP is used.

[0055] [Table 1]

[0056]

[0057] *N slot symb : The number of symbols in a time slot

[0058] *N frame,u slot : Number of time slots in a frame

[0059] *N subframe,u slot : Number of time slots in a subframe

[0060] Table 2 shows that the number of symbols per slot, the number of slots per frame, and the number of slots per subframe vary according to SCS when the extended CP is used.

[0061] [Table 2]

[0062]

[0063] The structure of the frame is only an example. The number of subframes, the number of time slots, and the number of symbols in a frame may vary.

[0064] In NR systems, OFDM parameter sets (e.g., SCSs) can be configured differently for multiple cells aggregated for a single UE. Consequently, the (absolute time) duration of a time resource (e.g., SF, time slot, or TTI) consisting of the same number of symbols (referred to as a time unit (TU) for simplicity) can be configured differently between the aggregated cells. Symbols can include OFDM symbols (or CP-OFDM symbols) and SC-FDMA symbols (or Discrete Fourier Transform-Spread-OFDM (DFT-s-OFDM) symbols).

[0065] Figure 3A resource grid showing a time slot. A time slot consists of multiple symbols in the time domain. For example, when using a normal CP, a time slot consists of 14 symbols. However, when using an extended CP, a time slot consists of 12 symbols. A carrier consists of multiple subcarriers in the frequency domain. A resource block (RB) is defined as multiple consecutive subcarriers in the frequency domain (e.g., 12 consecutive subcarriers). A bandwidth part (BWP) can be defined as multiple consecutive physical RBs (PRBs) in the frequency domain and corresponds to a single parameter set (e.g., SCS, CP length, etc.). A carrier can include up to N (e.g., five) BWPs. Data communication can be performed using enabled BWPs, and only one BWP can be enabled for a UE. In the resource grid, each element is called a resource element (RE), and one complex symbol can be mapped to each RE.

[0066] Figure 4 This figure shows an exemplary mapping of physical channels within a time slot. The PDCCH can be transmitted in the DL control region, and the PDSCH can be transmitted in the DL data region. The PUCCH can be transmitted in the UL control region, and the PUSCH can be transmitted in the UL data region. The guard period (GP) provides a time gap for switching from transmit mode to receive mode, or vice versa, at the base station and the user equipment terminal. Some symbols in a subframe during DL-to-UL switching can be configured as GPs.

[0067] The individual physical channels are described in more detail below.

[0068] The PDCCH transmits DCI. For example, the PDCCH (i.e., DCI) can carry information about the transport format and resource allocation of the DL Shared Channel (DL-SCH), resource allocation information for the Uplink Shared Channel (UL-SCH), paging information for the Paging Channel (PCH), system information about the DL-SCH, resource allocation for higher-layer control messages (e.g., the RAR sent on the PDSCH), transmit power control commands, and information about the activation / release of configured scheduling. The DCI includes a cyclic redundancy check (CRC). The CRC is masked with various identifiers (IDs) (e.g., the Radio Network Temporary Identifier (RNTI)) depending on the owner or purpose of the PDCCH. For example, if the PDCCH is for a specific UE, the CRC is masked with the UE ID (e.g., the Cell-RNTI (C-RNTI)). If the PDCCH is used for a paging message, the CRC is masked with the Paging-RNTI (P-RNTI). If the PDCCH is for system information (e.g., system information block (SIB)), the CRC is masked by the system information RNTI (SI-RNTI). When the PDCCH is for RAR, the CRC is masked by the random access RNTI (RA-RNTI).

[0069] The PDCCH consists of 1, 2, 4, 8, or 16 control channel elements (CCEs), depending on its aggregation level (AL). A CCE is a logical allocation unit used to provide a specific code rate to the PDCCH based on radio channel conditions. A CCE consists of 6 resource element groups (REGs), each defined by one OFDM symbol and one (P)RB. The PDCCH is transmitted in a control resource set (CORESET). A CORESET is defined as a collection of REGs with a given parameter set (e.g., SCS, CP length, etc.). Multiple CORESETs for a single UE can overlap in the time / frequency domain. A CORESET can be configured using system information (e.g., Master Information Block (MIB)) or UE-specific higher-layer signaling (e.g., Radio Resource Control (RRC) signaling). Specifically, the number of RBs and the number of symbols (up to 3) in a CORESET can be configured through higher-layer signaling.

[0070] For PDCCH reception / detection, the UE monitors PDCCH candidates. PDCCH candidates are CCEs that the UE should monitor to detect PDCCHs. Each PDCCH candidate is defined as 1, 2, 4, 8, or 16 CCEs, depending on the AL. Monitoring involves (blind) decoding of the PDCCH candidates. The set of PDCCH candidates decoded by the UE is defined as a PDCCH search space (SS). The SS can be a common search space (CSS) or a UE-specific search space (USS). The UE can obtain DCI by monitoring PDCCH candidates in one or more SSs configured by the MIB or higher-layer signaling. Each CORESET is associated with one or more SSs, and each SS is associated with one CORESET. The SS can be defined based on the following parameters.

[0071] - controlResourceSetId: CORESET related to SS.

[0072] - monitoringSlotPeriodicityAndOffset: PDCCH monitoring periodicity (time slot) and PDCCH monitoring offset (time slot).

[0073] - monitoringSymbolsWithinSlot: PDCCH monitoring symbols within a slot (e.g., the first symbol of a CORESET).

[0074] - nrofCandidates: the number of PDCCH candidates (one of 0, 1, 2, 3, 4, 5, 6 and 8) for each AL={1, 2, 4, 8, 16}.

[0075] * The timing (e.g., time / frequency resources) at which the UE monitors PDCCH candidates is defined as a PDCCH (monitoring) opportunity. One or more PDCCH (monitoring) opportunities may be configured in a time slot.

[0076] Table 3 shows the characteristics of each SS.

[0077] [Table 3]

[0078]

[0079] Table 4 shows the DCI format transmitted on the PDCCH.

[0080] [Table 4]

[0081]

[0082] DCI format 0_0 can be used to schedule TB-based (or TB-level) PUSCH, while DCI format 0_1 can be used to schedule TB-based (or TB-level) PUSCH or code block group (CBG)-based (or CBG-level) PUSCH. DCI format 1_0 can be used to schedule TB-based (or TB-level) PDSCH, while DCI format 1_1 can be used to schedule TB-based (or TB-level) PDSCH or CBG-based (or CBG-level) PDSCH (or DL grant DCI). DCI formats 0_0 / 0_1 are referred to as UL grant DCI or UL scheduling information, while DCI formats 1_0 / 1_1 are referred to as DL grant DCI or DL scheduling information. DCI format 2_0 is used to convey dynamic slot format information (e.g., dynamic slot format indicator (SFI)) to UEs, while DCI format 2_1 is used to convey DL preemption information to UEs. DCI format 2_0 and / or DCI format 2_1 can be transmitted on a group-common PDCCH (a PDCCH directed to a group of UEs) to a corresponding group of UEs.

[0083] DCI format 0_0 and DCI format 1_0 may be referred to as fallback DCI formats, while DCI format 0_1 and DCI format 1_1 may be referred to as non-fallback DCI formats. In the fallback DCI format, the DCI size / field configuration remains the same regardless of the UE configuration. In contrast, in the non-fallback DCI format, the DCI size / field configuration varies depending on the UE configuration.

[0084] The PDSCH transmits downlink data (e.g., downlink shared channel transport blocks (DL-SCH TBs)) and uses modulation schemes such as quadrature phase-shift keying (QPSK), 16-ary quadrature amplitude modulation (16QAM), 64QAM, or 256QAM. TBs are encoded as codewords. The PDSCH can transmit up to two codewords. Scrambling and modulation mapping are performed on a codeword basis, and the modulation symbols generated from each codeword can be mapped to one or more layers. Each layer is mapped to a resource along with a demodulation reference signal (DMRS). OFDM symbol signals are generated from the layers to which the DMRS is mapped and transmitted through the corresponding antenna ports.

[0085] PUCCH transmits uplink control information (UCI). UCI includes the following information.

[0086] - SR (Scheduling Request): Information used to request UL-SCH resources.

[0087] - HARQ (Hybrid Automatic Repeat Request)-ACK (Acknowledgement): A response to a DL data packet (e.g., a codeword) on the PDSCH. An HARQ-ACK indicates whether the DL data packet was successfully received. A 1-bit HARQ-ACK is sent in response to a single codeword. A 2-bit HARQ-ACK is sent in response to two codewords. HARQ-ACK responses include positive ACK (abbreviated as ACK), negative ACK (NACK), discontinuous transmission (DTX), or NACK / DTX. The term HARQ-ACK is used interchangeably with HARQ ACK / NACK and ACK / NACK.

[0088] - CSI (Channel State Information): Feedback information of the DL channel. Multiple-input multiple-output (MIMO) related feedback information includes RI and PMI.

[0089] Table 5 shows an exemplary PUCCH format. Based on PUCCH transmission duration, PUCCH formats may be divided into short PUCCH (formats 0 and 2) and long PUCCH (formats 1, 3, and 4).

[0090] [Table 5]

[0091]

[0092] PUCCH format 0 carries up to 2 bits of UCI and is mapped in a sequence-based manner for ease of transmission. Specifically, the UE sends specific UCI to the base station by sending one of multiple sequences on the PUCCH in PUCCH format 0. The UE sends PUCCH format 0 in the PUCCH resources configured for the corresponding SR only when it sends a positive SR.

[0093] PUCCH format 1 transmits up to 2 bits of UCI, and the modulation symbols of the UCI are spread in the time domain using an orthogonal cover code (OCC) (configured differently depending on whether frequency hopping is performed). DMRS is transmitted in symbols where no modulation symbols are transmitted (i.e., transmitted using time division multiplexing (TDM)).

[0094] PUCCH format 2 transmits more than 2 bits of UCI, and the modulation symbols for the DCI are sent using frequency division multiplexing (FDM) using DMRS. DMRS are located in symbols #1, #4, #7, and #10 of a given RB at a density of 1 / 3. A pseudo-noise (PN) sequence is used for the DMRS sequence. Frequency hopping can be enabled for the 2-symbol PUCCH format 2.

[0095] PUCCH format 3 does not support UE multiplexing in the same PRBS and transmits UCI exceeding 2 bits. In other words, the PUCCH resource of PUCCH format 3 does not include OCC. Modulation symbols are sent in TDM using DMRS.

[0096] PUCCH format 4 supports multiplexing of up to 4 UEs in the same PRBS and transmits more than 2 bits of UCI. In other words, the PUCCH resource of PUCCH format 3 includes OCC. Modulation symbols are sent in TDM using DMRS.

[0097] At least one of the one or two or more cells configured in the UE may be configured for PUCCH transmission. At least the primary cell may be configured as a cell for PUCCH transmission. Based on the at least one cell configured for PUCCH transmission, at least one PUCCH cell group may be configured in the UE, and each PUCCH cell group includes one or more cells. The PUCCH cell group may be referred to as a PUCCH group. PUCCH transmission may be configured for the SCell as well as the primary cell. The primary cell belongs to the primary PUCCH group, and the PUCCH-SCell configured for PUCCH transmission belongs to the secondary PUCCH group. The PUCCH on the primary cell can be used for cells belonging to the primary PUCCH group, and the PUCCH on the PUCCH-SCell can be used for cells belonging to the secondary PUCCH group.

[0098] The PUSCH transmits UL data (e.g., UL shared channel transport blocks (UL-SCH TBs)) and / or UCI based on a CP-OFDM waveform or a DFT-s-OFDM waveform. When the PUSCH is transmitted using a DFT-s-OFDM waveform, the UE transmits the PUSCH using transform precoding. For example, when transform precoding is not possible (e.g., disabled), the UE may transmit the PUSCH using a CP-OFDM waveform, whereas when transform precoding is possible (e.g., enabled), the UE may transmit the PUSCH using a CP-OFDM or DFT-s-OFDM waveform. PUSCH transmissions can be dynamically scheduled via a UL grant in the DCI or semi-statically scheduled via higher-layer (e.g., RRC) signaling (and / or Layer 1 (L1) signaling such as the PDCCH) (configured scheduling or configured grants). PUSCH transmissions can be performed in a codebook-based or non-codebook-based manner.

[0099] Figure 5 An exemplary ACK / NACK transmission process is shown. Figure 5 , the UE can detect the PDCCH in time slot #n. The PDCCH includes DL scheduling information (e.g., DCI format 1_0 or DCI format 1_1). The PDCCH indicates DL assignments with PDSCH offset K0 and PDSCH with HARQ-ACK report offset K1. For example, DCI format 1_0 and DCI format 1_1 may include the following information.

[0100] - Frequency domain resource assignment: Indicates the set of RBs assigned to PDSCH.

[0101] - Time domain resource assignment: Indicates K0 and the starting position (e.g., OFDM symbol index) and length (e.g., number of OFDM symbols) of the PDSCH in the time slot.

[0102] - PDSCH-to-HARQ_feedback timing indicator: indicates K1.

[0103] - HARQ process number (4 bits): Indicates the HARQ process ID of the data (eg, PDSCH or TB).

[0104] - PUCCH resource indicator (PRI): indicates the PUCCH resource to be used for UCI transmission among multiple PUCCH resources in a PUCCH resource set.

[0105] After receiving PDSCH in slot #(n+K0) according to the scheduling information of slot #n, the UE may send UCI on PUCCH in slot #(n+K1). UCI may include HARQ-ACK response to PDSCH. For convenience, Figure 5Based on the assumption that the SCS of PDSCH is equal to the SCS of PUCCH and slot #n1=slot #(n+K0), this should not be construed as limiting the present disclosure. When the SCSs are different, K1 may be indicated / interpreted based on the SCS of PUCCH.

[0106] When the PDSCH is configured to carry up to one TB, the HARQ-ACK response can be configured in one bit. When the PDSCH is configured to carry up to two TBs, the HARQ-ACK response can be configured in two bits if spatial bundling is not configured, and in one bit if spatial bundling is configured. When slot #(n+K1) is designated as the HARQ-ACK transmission timing for multiple PDSCHs, the UCI transmitted in slot #(n+K1) includes the HARQ-ACK responses for the multiple PDSCHs.

[0107] Whether the UE should perform spatial bundling for HARQ-ACK responses may be configured for each cell group (e.g., via RRC / higher layer signaling). For example, spatial bundling may be configured for each individual HARQ-ACK response sent on the PUCCH and / or for a HARQ-ACK response sent on the PUSCH.

[0108] When up to two (or two or more) TBs (or codewords) can be received at a time in the corresponding serving cell (which can be scheduled by one DCI) (for example, when the high-layer parameter maxNrofCodeWordsScheduledByDCI indicates 2 TBs), spatial bundling can be supported. More than four layers can be used for 2TB transmission, and up to four layers can be used for 1TB transmission. As a result, when spatial bundling is configured for the corresponding cell group, spatial bundling can be performed for serving cells in the cell group that can schedule more than four layers. A UE that wants to send a HARQ-ACK response through spatial bundling can generate a HARQ-ACK response by performing a (bitwise) logical AND operation on the A / N bits of multiple TBs.

[0109] For example, assuming that a UE receives DCI that schedules two TBs and receives the two TBs on the PDSCH based on the DCI, the UE performing spatial bundling can generate a single A / N bit by performing a logical AND operation between the first A / N bit of the first TB and the second A / N bit of the second TB. As a result, when both the first TB and the second TB are ACK, the UE reports an ACK bit value to the BS, and when at least one TB is NACK, the UE reports a NACK bit value to the BS.

[0110] For example, when only one TB is actually scheduled in a serving cell configured to receive two TBs, the UE may generate a single A / N bit by performing a logical AND operation on the A / N bits of the one TB and a bit value 1. As a result, the UE reports the A / N bits of the one TB to the BS.

[0111] Multiple parallel DL HARQ processes exist at the BS / UE for DL transmissions. These processes allow for continuous DL transmissions while the BS awaits HARQ feedback indicating the success or failure of the previous DL transmission. Each HARQ process is associated with a HARQ buffer in the media access control (MAC) layer. Each DL HARQ process manages state variables such as the number of MAC physical data unit (PDU) transmissions, HARQ feedback for MAC PDUs in the buffer, and the current redundancy version. Each HARQ process is identified by a HARQ process ID.

[0112] Figure 6 An exemplary PUSCH transmission process is shown. Figure 6 , the UE can detect the PDCCH in time slot #n. The PDCCH includes DL scheduling information (e.g., DCI format 1_0 or 1_1). DCI format 1_0 or 1_1 may include the following information.

[0113] - Frequency domain resource assignment: Indicates the set of RBs assigned to PUSCH.

[0114] - Time domain resource assignment: Indicates the slot offset K2 and the starting position (e.g., OFDM symbol index) and duration (e.g., number of OFDM symbols) of the PUSCH in the slot. The starting symbol and length of the PUSCH can be indicated by the start and length indicator value (SLIV) or separately.

[0115] The UE may then transmit the PUSCH in slot #(n+K2) according to the scheduling information in slot #n. The PUSCH includes the UL-SCH TB.

[0116] CSI-related operations

[0117] Figure 7 An example of a CSI-related process is shown.

[0118] The UE receives CSI-related configuration information from the BS via RRC signaling (710). The CSI-related configuration information may include at least one of channel state information-interference measurement (CSI-IM) related information, CSI measurement related information, CSI resource configuration related information, CSI-RS resource related information, or CSI report configuration related information.

[0119] - CSI-IM resources can be configured for UE interference measurement (IM). In the time domain, CSI-IM resource sets can be configured as periodic, semi-persistent, or aperiodic. CSI-IM resources can be configured as zero-power (ZP) CSI-RS for the UE. ZP CSI-RS can be configured to be distinguished from non-zero-power (NZP) CSI-RS.

[0120] - The UE may assume that the CSI-RS resources used for channel measurement and the CSI-IM / NZP CSI-RS resources used for interference measurement configured for one CSI report have a QCL relationship with respect to the "QCL-TypeD" of each resource (when the NZP CSI-RS resources are used for interference measurement).

[0121] - The CSI resource configuration may include at least one of a CSI-IM resource for interference measurement, an NZP CSI-RS resource for interference measurement, and an NZP CSI-RS resource for channel measurement. The channel measurement resource (CMR) may be an NZP CSI-RS used for CSI acquisition, and the interference measurement resource (IMR) may be an NZP CSI-RS used for CSI-IM and IM.

[0122] - CSI-RS can be configured for one or more UEs. Each UE can be provided with a different CSI-RS configuration, or the same CSI-RS configuration can be provided to multiple UEs. CSI-RS can support up to 32 antenna ports. CSI-RS corresponding to N (N is 1 or greater) antenna ports can be mapped to N RE positions within the time-frequency unit corresponding to one slot and one RB. When N is 2 or greater, the N-port CSI-RS can be multiplexed using CDM, FDM, and / or TDM. CSI-RS can be mapped to the remaining REs excluding those mapped with the CORESET, DMRS, and SSB. In the frequency domain, CSI-RS can be configured for the entire bandwidth, a fractional bandwidth part (BWP), or a fractional bandwidth. CSI-RS can be transmitted in each RB within the bandwidth in which the CSI-RS is configured (i.e., density = 1), or in every third RB (e.g., even or odd RBs) (i.e., density = 1 / 2). When CSI-RS is used as a tracking reference signal (TRS), a single-port CSI-RS can be mapped to three subcarriers in each resource block (i.e., density = 3). One or more CSI-RS resource sets can be configured for a UE in the time domain. Each CSI-RS resource set can include one or more CSI-RS configurations. Each CSI-RS resource set can be configured as periodic, semi-persistent, or aperiodic.

[0123] - A CSI report configuration may include configurations for feedback type, measurement resources, and report type. An NZP-CSI-RS resource set may be used for the CSI report configuration corresponding to the UE. An NZP-CSI-RS resource set may be associated with a CSI-RS or SSB. Multiple periodic NZP-CSI-RS resource sets may be configured as TRS resource sets. (i) Feedback types include channel quality indicator (CQI), precoding matrix indicator (PMI), CSI-RS resource indicator (CRI), SSB resource block indicator (SSBRI), layer indicator (LI), rank indicator (RI), and layer 1 (L1) reference signal received strength (RSRP). (ii) Measurement resources may include configurations for downlink signals and / or downlink resources on which the UE performs measurements to determine feedback information. Measurement resources may be configured as ZP and / or NZP CSI-RS resource sets associated with the CSI report configuration. An NZP CSI-RS resource set may include a CSI-RS set or an SSB set. For example, L1-RSRP can be measured for a CSI-RS set or an SSB set. (iii) The report type may include the time at which the UE performs reporting and the configuration of the uplink channel. The reporting time may be configured as periodic, semi-persistent, or aperiodic. Periodic CSI reports may be sent on the PUCCH. Semi-persistent CSI reports may be sent on the PUCCH or PUSCH based on a MAC CE indicating activation / deactivation. Aperiodic CSI reports may be indicated by DCI signaling. For example, the CSI request field of an uplink grant may indicate one of various report trigger sizes. Aperiodic CSI reports may be sent on the PUSCH.

[0124] The UE measures CSI based on the CSI-related configuration information. The CSI measurement may include receiving a CSI-RS (720) and obtaining CSI by calculating the received CSI-RS (730).

[0125] The UE may send a CSI report to the BS (740). For the CSI report, the time resources and frequency resources available to the UE are controlled by the BS. The channel state information (CSI) includes at least one of a channel quality indicator (CQI), a precoding matrix indicator (PMI), a CSI-RS resource indicator (CRI), an SS / PBCH block resource indicator (SSBRI), a layer indicator (LI), a rank indicator (RI), L1-RSRP, and / or L-SINR.

[0126] The time domain behavior of CSI reporting supports periodic, semi-persistent, and aperiodic reporting. i) Periodic CSI reporting is performed on the short PUCCH and long PUCCH. The periodicity and slot offset of periodic CSI reporting can be configured by RRC and refer to the CSI-ReportConfig IE. ii) SP (semi-periodic) CSI reporting is performed on the short PUCCH, long PUCCH, or PUSCH. For SP CSI in short / long PUCCH, the periodicity and slot offset are configured by RRC, and CSI reporting is enabled / disabled via a separate MAC CE / DCI. For SP CSI in PUSCH, the periodicity of SP CSI reporting is configured by RRC, but the slot offset is not configured by RRC, and SP CSI reporting is enabled / disabled by DCI (format 0_1). For SP CSI reporting in PUSCH, a separate RNTI (SP-CSI C-RNTI) is used. The initial CSI reporting timing follows the PUSCH time domain allocation value indicated by the DCI, and the subsequent CSI reporting timing follows the periodicity configured by RRC. DCI format 0_1 may include a CSI request field and enable / disable a specific configured SP-CSI triggering state. SP CSI reporting uses the same or similar mechanism for enabling / disabling SP CSI reporting as with data transmission in SPS PUSCH. iii) Aperiodic CSI reporting is performed in PUSCH and triggered by DCI. In this case, information related to the triggering of aperiodic CSI reporting can be transmitted / indicated / configured via MAC-CE. For AP CSI with AP CSI-RS, the AP CSI-RS timing is configured by RRC, and the timing of the AP CSI report is dynamically controlled by DCI.

[0127] The CSI codebooks (e.g., PMI codebooks) defined in the NR specification can be broadly categorized into Type I and Type II codebooks. Type I codebooks primarily support single-user MIMO (SU-MIMO), supporting both high-order and low-order operations. Type II codebooks primarily support multi-user MIMO (MU-MIMO), capable of handling up to two layers. While Type II codebooks provide more accurate CSI than Type I codebooks, they can also increase signaling overhead. On the other hand, enhanced Type II codebooks were introduced to address the CSI overhead associated with existing Type II codebooks. These enhanced Type II codebooks can reduce the codebook payload by accounting for correlation in the frequency domain.

[0128] The CSI report on the PUSCH can be configured as part 1 and part 2. Part 1 has a fixed payload size and is used to identify the number of information bits in part 2. Part 1 is completely transmitted before part 2.

[0129] - For Type I CSI feedback, Part 1 includes the RI (if reported), CRI (if reported), and CQI for the first codeword. Part 2 includes the PMI, and when RI>4, Part 2 also includes the CQI.

[0130] - For Type II CSI feedback, Part 1 includes the RI (if reported), CQI, and an indication of the number of non-zero WB amplitude coefficients for each layer of the Type II CSI. Part 2 includes the PMI of the Type II CSI.

[0131] - For enhanced Type II CSI feedback, Part 1 includes the RI (if reported), CQI, and an indication of the total number of non-zero WB amplitude coefficients for all layers of the enhanced Type II CSI. Part 2 includes the PMI of the enhanced Type II CSI.

[0132] For PUSCH, the CSI report consists of two parts. If the CSI payload to be reported is smaller than the payload size provided by the PUSCH resources allocated for CSI reporting, the UE may drop part 2 of the CSI.

[0133] Semi-persistent CSI reporting performed in PUCCH format 3 or 4 supports Type II CSI feedback, but only supports part 1 of Type I CSI feedback.

[0134] Quasi-isotopic luminescence (QCL)

[0135] Two antenna ports are quasi-collocated when the channel properties of one antenna port are to be inferred from the channel of another antenna port. The channel properties may include one or more of delay spread, Doppler spread, frequency / Doppler shift, average received power, receive timing / average delay, and spatial RX parameters.

[0136] A list of multiple TCI state configurations can be configured in the UE via the higher-layer parameter PDSCH-Config. Each TCI state is linked to the QCL configuration parameters between one or two DL reference signals and the DM-RS ports of the PDSCH. The QCL may include QCL-Type 1 for the first DL RS and QCL-Type 2 for the second DL RS. The QCL type may correspond to one of the following:

[0137] - "QCL-TypeA": {Doppler shift, Doppler spread, average delay, delay spread}

[0138] - "QCL-TypeB": {Doppler shift, Doppler spread}

[0139] - "QCL-TypeC": {Doppler shift, average delay}

[0140] - "QCL-TypeD": {spatial Rx parameters}

[0141] Beam Management (BM)

[0142] BM refers to a series of processes for acquiring and maintaining a BS beam set (Transmit Reception Point (TRP) beam) and / or a UE beam set that can be used for DL and UL transmission / reception. BM may include the following processes and terms.

[0143] - Beam measurement: An operation in which a BS or UE measures the characteristics of a received beamformed signal.

[0144] - Beam determination: The operation by which a BS or UE selects its Tx / Rx beam.

[0145] - Beam sweeping: The operation of covering the spatial domain using Tx and / or Rx beams within a prescribed time interval according to a predetermined method.

[0146] - Beam reporting: An operation in which the UE reports information about a beamformed signal based on beam measurements.

[0147] The BM process can be divided into (1) a DL BM process using SSB or CSI-RS and (2) a UL BM process using SRS. In addition, each BM process may include Tx beam sweeping for determining a Tx beam and Rx beam sweeping for determining an Rx beam.

[0148] The DL BM procedure may include (1) transmission of a beamformed DL RS (eg, CSI-RS or SSB) from the BS and (2) beam reporting from the UE.

[0149] The beam report may include a preferred DL RS ID and a reference signal received power (RSRP) corresponding to the preferred DL RS ID. The DL RS ID may be a SSB resource indicator (SSBRI) or a CSI-RS resource indicator (CRI).

[0150] Artificial Intelligence / Machine Learning (AI / ML)

[0151] With the advancement of AI / ML technologies, the nodes and UEs that comprise wireless communication networks are becoming increasingly intelligent and advanced. In particular, network / BS intelligence is expected to rapidly optimize and derive / apply various network / BS-determined parameter values (e.g., the transmit and receive power of each BS, the transmit power of each UE, the precoder / beam of the BS / UE, the time / frequency resource allocation for each UE, or the duplexing method of the BS) based on various environmental parameters (e.g., the distribution / location of BSs, the distribution / location / material of buildings / furniture, the location / movement direction / speed of UEs, and climate information). In line with this trend, many standardization organizations (e.g., 3GPP and O-RAN) are considering the introduction of network / BS-determined parameter values, and research on this topic is actively underway.

[0152] In a narrow sense, AI / ML can be easily referred to as artificial intelligence based on deep learning, but the concept is shown in Figure 8 middle.

[0153] - Artificial Intelligence: This corresponds to all automation where machines replace human work.

[0154] - Machine Learning: Machines learn decision-making patterns from data without explicitly programmed rules.

[0155] Deep learning: This is an AI / ML model based on artificial neural networks, where machines perform everything from unstructured data to feature extraction and determination all at once. The algorithm relies on biological neural systems, i.e., a multi-layered network of interconnected nodes inspired by neural networks for feature extraction and transformation. Common deep learning network architectures include deep neural networks (DNNs), recurrent neural networks (RNNs), and convolutional neural networks (CNNs).

[0156] Classification of AI / ML types according to various references

[0157] 1. Offline vs. online

[0158] (1) Offline learning: This follows a sequential process of database collection, learning, and prediction. In other words, collection and learning are performed offline, and the completed program can be installed on-site and used for prediction. In most cases, this offline learning method is used. In offline learning, the system does not learn incrementally. Instead, it uses all available collected data to learn, and the results are applied without further learning. If new data needs to be learned, learning can be restarted using the new, complete data.

[0159] (2) Online Learning: Online learning is a method that improves performance little by little by incrementally learning from newly generated data, based on the fact that data for learning is continuously generated via the internet. Learning is performed in real time on specific data units (batches) collected online, allowing the system to quickly adapt to changing data.

[0160] To build an AI system, learning can be performed solely through online learning using only real-time generated data. Alternatively, after offline learning using a specific dataset, additional learning can be performed using subsequently generated real-time data (online + offline learning).

[0161] 2. Classification based on AI / ML framework concepts

[0162] (1) Centralized learning: When training data collected from multiple different nodes are reported to a centralized node, all data resources / storage / learning (e.g., supervised, unsupervised, and reinforcement learning) are performed by one central node.

[0163] (2) Federated Learning: Configuring a collective AI / ML model based on data across decentralized data owners. Instead of using data in the AI / ML model, local nodes / devices collect data and train their own copies of the AI / ML model, eliminating the need to report source data to a central node. In federated learning, the parameters / weights of the AI / ML model can be sent back to a centralized node to support general AI / ML model training. Advantages of federated learning include increased computing speed and information security. That is, personal data does not need to be uploaded to a central server for processing, preventing the leakage and abuse of personal information.

[0164] (3) Distributed Learning: Machine learning processing represents the concept of scaling and deployment across a cluster of nodes. AI / ML model training is split and shared across multiple nodes operating simultaneously to accelerate AI / ML model training.

[0165] 3. Classification by learning method

[0166] (1) Supervised Learning: Supervised learning is a machine learning task that aims to learn a mapping function from input to output given a labeled dataset. The input data is called training data and has known labels or outcomes. Examples of supervised learning include: (i) regression: linear regression, logistic regression; (ii) instance-based algorithms: k-nearest neighbor (KNN); (iii) decision tree algorithms: CART; (iv) support vector machines: SVM; (v) Bayesian algorithms: Naive Bayes; and (vi) ensemble algorithms: extreme gradient boosting, packing: random forest. Supervised learning can be further divided into two categories: classification and prediction of labels, and regression prediction of quantities.

[0167] (2) Unsupervised learning: This is a machine learning task that aims to learn a function that describes the hidden structure in unlabeled data. The input data is unlabeled and the outcome is not known. Some examples of unsupervised learning include K-means clustering, principal component analysis (PCA), nonlinear independent component analysis (ICA), and LSTM.

[0168] (3) Reinforcement Learning: In reinforcement learning (RL), an agent aims to optimize a long-term goal by interacting with the environment through a trial-and-error process, which is goal-oriented learning based on interaction with the environment. Examples of RL algorithms include (i) Q-learning, (ii) multi-armed bandit learning, (iii) deep Q-networks, state-action-reward-state-action (SARSA), (iv) temporal difference learning, (v) actuator-critic reinforcement learning, (vi) deep deterministic policy gradient, and (vii) Monte Carlo tree search. RL can be further grouped into AI / ML model-based RL and AI / ML model-free RL. Model-based RL is an RL algorithm that uses a predictive AI / ML model to obtain the transition probabilities between states using various dynamic states of the environment and an AI / ML model of the rewards these states lead to. Model-free RL is a value-based or policy-based RL algorithm that maximizes future rewards, which has low computational complexity in terms of multi-agent environments / states and does not require an accurate representation of the environment. RL algorithms can also be categorized as value-based RL versus policy-based RL, policy-based RL versus off-policy RL, etc.

[0169] AI / ML models

[0170] Figure 9 Shows an example of a feedforward neural network (FFNN) AI / ML model. Figure 9 ,FFNN AI / ML model includes input layer, hidden layer and output layer.

[0171] Figure 10 Shows an example of a recurrent neural network (RNN) AI / ML model. Figure 10 , the RNN AI / ML model is an artificial neural network in which hidden nodes are connected to directed edges to form a directed cycle. It is an AI / ML model suitable for processing data that appears sequentially (for example, speech or text). One type of RNN is the long short-term memory (LSTM), which is a structure that adds the cell state to the hidden state of the RNN. In detail, in LSTM, an input gate, a forget gate, and an output gate are added to the RNN cell, and the cell state is added. Figure 10 In this example, A represents a neural network, and x t Indicates the input value, h t Indicates the output value. Here, h t It can represent the current state value based on time, and h t-1 Can represent the previous state value.

[0172] Figure 11A convolutional neural network (CNN) AI / ML model is shown. CNN is used for two purposes, including reducing the complexity of the AI / ML model and extracting good features by applying convolution calculations commonly used in video processing or image processing. Figure 11 A kernel or filter refers to a unit / structure that applies weights to the input within a specific range / unit. Kernels (or filters) can be modified through learning. The stride is the range by which the kernel moves within the input. A feature map is the result of applying a kernel to the input. Padding refers to a value added to adjust the size of the feature map. To enhance robustness to distortion and variation, multiple feature maps can be extracted. Pooling refers to the computation of reducing the size of a feature map by downsampling it (for example, max pooling or average pooling).

[0173] Figure 12 Shows the autoencoder AI / ML model. Figure 12 An autoencoder is a type of unsupervised learning where a neural network receives a feature vector x and outputs the same or a similar vector x'. The input and output nodes have the same features. The autoencoder reconstructs the input, so the output can be called a reconstruction. The loss function can be expressed as Equation 1 below.

[0174] [Formula 1]

[0175] ,in

[0176] In Equation 12, the loss function of the autoencoder is calculated based on the difference between the input and the output. Based on the loss function of the autoencoder, the degree of loss in the input is evaluated, and an optimization process is applied to the autoencoder to minimize the loss.

[0177] Figure 13 is a diagram illustrating split AI reasoning.

[0178] Figure 13 It shows a case where the model inference function is performed collaboratively between a terminal device such as a UE and a network AI / ML endpoint during split AI operation.

[0179] In addition to model inference, model training, executors, and data collection can each be split into multiple parts based on the current task and environment. These functions can be performed through the collaboration of multiple entities.

[0180] For example, computationally and energy-intensive portions can be executed at a network endpoint, while privacy-sensitive and latency-sensitive portions can be executed on a terminal device. In this scenario, the terminal device executes the task / model based on input data up to a specific portion / layer, then sends the intermediate data to the network endpoint. The network endpoint executes the remaining portions / layers and provides inference output to one or more devices performing the operation / task.

[0181] The following describes the functional framework for AI operations.

[0182] In this article, the following terms are defined to explain AI (or AI / ML) in more detail.

[0183] - Data Collection: Data collected from network nodes, management entities, or UEs, which is used as the basis for AI model training, data analysis, and inference.

[0184] - AI Model: A data-driven algorithm in which AI techniques are applied to generate a set of outputs, including prediction information and / or decision parameters, based on a set of inputs.

[0185] - AI / ML training: The online or offline process of training AI models by learning features and patterns to best represent the data and obtaining an AI / ML model trained for inference.

[0186] - AI / ML Inference: The process of using trained AI models to make predictions or draw decisions based on collected data and AI models.

[0187] Reference Figure 14 , the data collection function 10 collects input data and provides the processed input data to the model training function 20 and the model inference function 30.

[0188] For example, input data may include measurements from UEs or other network entities, feedback from actuators, and output from AI models.

[0189] The data collection function 10 performs data preparation based on the input data and provides the input data processed by the data preparation. Here, the data collection function 10 does not perform specific data preparation (e.g., data preprocessing and cleaning, forming and transforming) for each AI algorithm, but the data collection function 10 can perform data preparation common to the AI algorithms.

[0190] After completing the data preparation process, the data collection function 10 provides the training data 11 to the model training function 20 and provides the inference data 12 to the model inference function 30. Here, the training data 11 is the input data required by the AI model training function 20. The inference data 12 is the input data required by the AI model inference function 30.

[0191] The data collection function 10 may be performed by a single entity (e.g., UE, RAN node, network node, etc.) or by multiple entities. In this case, the training data 11 and the inference data 12 may be provided from the multiple entities to the model training function 20 and the model inference function 30, respectively.

[0192] As part of the AI model testing process, the model training function 20 is responsible for performing the AI model training, validation, and testing required to generate model performance metrics. If necessary, the model training function 20 can handle data preparation (e.g., data preprocessing and cleaning, shaping, and transformation) based on the training data 11 provided by the data collection function 10.

[0193] Here, model deployment / update 13 is used to initially deploy the trained, validated, and tested AI model to the model reasoning function 30 or to provide an updated model to the model reasoning function 30.

[0194] The model inference function 30 is responsible for providing AI model inference output 16 (e.g., predictions or decisions). If applicable, the model inference function 30 may also provide model performance feedback 14 to the model training function 20. Additionally, if necessary, the model inference function 30 may handle data preparation (e.g., data preprocessing and cleaning, shaping, and transformation) based on the inference data 12 provided by the data collection function 10.

[0195] Here, output 16 refers to the inference output of the AI model generated by the model inference function 30, and the details of the inference output may vary depending on the use case.

[0196] When available, model performance feedback 14 can be used to monitor the performance of the AI model. However, feedback can be omitted.

[0197] The executor function 40 receives the output 16 from the model inference function 30 and triggers or performs relevant tasks / operations. The executor function 40 may trigger tasks / operations of other entities (eg, one or more UEs, one or more RAN nodes, one or more network nodes, etc.) or itself.

[0198] Feedback 15 can be used to derive training data 11 and inference data 12 or monitor the performance of the AI model, its impact on the network, etc.

[0199] The definitions of training, validation, and testing in datasets used in AI / ML can be distinguished as follows:

[0200] - Training data: Training data refers to the dataset used to train the model.

[0201] - Validation data: Validation data is a dataset used to validate a trained model. In other words, validation data is a dataset used to prevent overfitting to a typical training dataset.

[0202] Validation data is a dataset used to select the best model among multiple trained models during the training process. Therefore, validation data can also be considered a type of learning.

[0203] - Test data: Test data refers to the dataset used for final evaluation. Test data is independent of training data.

[0204] In the case of the above datasets, it is common to split the training set so that the training and validation data are divided in a ratio of 8:2 or 7:3. When including test data, the training, validation, and test data can be divided in a ratio of 6:2:2 (train:validation:test).

[0205] Depending on the capabilities of the AI / ML functions between the BS and the UE, the cooperation levels can be defined as follows: Multiple levels can be combined or the levels can be modified by separating any one level.

[0206] Cat 0a) No collaborative framework: AI / ML algorithms are purely implementation-based and do not require changes to the wireless interface.

[0207] Cat 0b) This level involves wireless interfaces suitable for modification based on effectively implemented AI / ML algorithms, but this level corresponds to a framework without collaboration.

[0208] Category 1) provides inter-node support to enhance the AI / ML algorithms of each node. This level applies when the UE receives support (for training, adaptation, etc.) from the gNB and vice versa. At this level, no model exchange between network nodes is required.

[0209] Category 2) enables collaborative ML tasks between the UE and gNB. This level requires the exchange of AI / ML model commands between network nodes.

[0210] Figure 14 The functions shown may be implemented in a RAN node (eg, a BS, a TRP, a Central Unit (CU) of a BS), a network node, Operations Administration and Maintenance (OAM) of a network operator, or a UE.

[0211] Alternatively, two or more entities among the RAN node, the network node, the network operator's OAM or the UE may collaborate to implement Figure 14 For example, an entity may perform Figure 14 Some of the functions in the , and another entity can perform the remaining functions. Thus, if Figure 14 If some of the functions shown are performed by a single entity (e.g., a UE, a RAN node, or a network node), the transmission / provision of data / information between the functions may be omitted. For example, if the model training function 20 and the model inference function 30 are performed by the same entity, the transmission / provision of the model deployment / update 13 and the model performance feedback 14 may be omitted.

[0212] Alternatively, Figure 14Any of the functions shown may be performed by collaboration between two or more entities in a RAN node, a network node, a network operator's OAM, or a UE. This may be referred to as split AI operation.

[0213] Figure 15 The scenario is shown in which the AI model training function is performed by a network node (e.g., a core network node, a network operator's OAM, etc.), while the AI model inference function is performed by a RAN node (e.g., a BS, a TRP, or a CU of a BS).

[0214] Step 1: RAN Node 1 and RAN Node 2 send input data (i.e., training data) for AI model training to the network node. Here, RAN Node 1 and RAN Node 2 may send data collected from UEs (e.g., UE measurements related to RSRP, RSRQ or SINR of the serving cell and neighboring cells, UE location, speed, etc.) to the network node.

[0215] Step 2: The network nodes train the AI model using the received training data.

[0216] Step 3: The network node deploys / updates the AI model to RAN node 1 and / or RAN node 2. RAN node 1 (and / or RAN node 2) may continue to perform model training based on the received AI model.

[0217] For ease of explanation, it is assumed that the AI model is only deployed / updated to RAN node 1.

[0218] Step 4: RAN node 1 receives input data for AI model inference (i.e., inference data) from UE and RAN node 2.

[0219] Step 5: RAN node 1 uses the received inference data to perform AI model inference to generate output data (e.g., prediction or decision).

[0220] Step 6: If applicable, the RAN node 1 may send model performance feedback to the network node.

[0221] Step 7: RAN node 1, RAN node 2, and UE (or RAN node 1 and UE, or RAN node 1 and RAN node 2) perform actions based on the output data. For example, in the case of a load balancing operation, the UE may move from RAN node 1 to RAN node 2.

[0222] Step 8: RAN node 1 and RAN node 2 send feedback information to the network node.

[0223] Figure 16 A scenario is shown in which both the AI model training function and the AI model inference function are performed by a RAN node (e.g., a BS, a TRP, or a CU of a BS).

[0224] Step 1: UE and RAN node 2 send input data (i.e., training data) for AI model training to RAN node 1.

[0225] Step 2: RAN node 1 uses the received training data to train the AI model.

[0226] Step 3: RAN node 1 receives input data for AI model inference (i.e., inference data) from UE and RAN node 2.

[0227] Step 4: RAN node 1 uses the received inference data to perform AI model inference to generate output data (e.g., prediction or decision).

[0228] Step 5: RAN node 1, RAN node 2, and the UE (or RAN node 1 and the UE, or RAN node 1 and RAN node 2) perform actions based on the output data. For example, in the case of a load balancing operation, the UE may move from RAN node 1 to RAN node 2.

[0229] Step 6: RAN node 2 sends feedback information to RAN node 1.

[0230] Figure 17 This section illustrates a scenario in which the AI model training function is performed by a RAN node (e.g., a BS, a TRP, or a CU of a BS), while the AI model inference function is performed by a UE.

[0231] Step 1: The UE sends input data (i.e., training data) for AI model training to the RAN node. Here, the RAN node may collect data from various UEs and / or other RAN nodes (e.g., UE measurements related to RSRP, RSRQ or SINR of the serving cell and neighboring cells, UE location, speed, etc.).

[0232] Step 2: The RAN node uses the received training data to train the AI model.

[0233] Step 3: The RAN node deploys / updates the AI model to the UE. The UE can continue to perform model training based on the received AI model.

[0234] Step 4: The UE receives input data for AI model inference (i.e., inference data) from the RAN node (and / or from other UEs).

[0235] Step 5: The UE performs AI model inference using the received inference data to generate output data (e.g., prediction or decision).

[0236] Step 6: If applicable, the UE may send model performance feedback to the RAN node.

[0237] Step 7: The UE and RAN nodes perform actions based on the output data.

[0238] Step 8: The UE sends feedback information to the RAN node.

[0239] AI / ML model management for AI / ML-based CSI prediction

[0240] The present disclosure proposes management and updating of AI / ML models for AI / ML-based CSI prediction reporting.

[0241] While there are various topics related to enhancements to CSI feedback, such as AI / ML-based joint prediction, improvements to existing codebooks, and CSI compression in the time-space-frequency (TSF) domain, this disclosure specifically focuses on CSI prediction in the time domain based on unilateral AI / ML models.

[0242] Figure 18 is a diagram illustrating predictions in the time domain based on a unilateral AI / ML model.

[0243] refer to Figure 18 , the unilateral AI / ML model on the network side or UE side predicts and outputs CSI based on the input of past CSI channel measurements.

[0244] Further discussion is needed on CSI prediction, and similar discussions are underway for MIMO, taking into account high mobility. AI / ML-based CSI prediction may be suitable for Rel-18 Type II CSI. However, if prediction is performed based on low-performance methods such as sample-and-hold, it is difficult to properly evaluate the performance improvement.

[0245] A consensus has recently been reached in standardization meetings regarding AI / ML-based predictions for future CSI, as shown in Table 6.

[0246] [Table 6]

[0247]

[0248] The following describes an AI / ML model management method for efficiently supporting AI / ML-based CSI prediction, assuming a single-side AI / ML model. Single-side AI / ML model training and inference can be performed by a single node, but both training and inference do not necessarily have to be performed by the same node. The node performing training / inference may include, for example, a gNB, an AI / ML server (e.g., non-3GPP), or a UE. For simplicity, this disclosure primarily describes UE-side AI / ML models, but embodiments are not limited thereto. The proposals described below are applicable to general single-side AI / ML models. AI / ML model lifecycle management (LCM) includes a series of operations, such as AI / ML model configuration, AI / ML model deployment, AI / ML model registration, AI / ML model training, AI / ML model inference, AI / ML model monitoring, and / or AI / ML model updates. AI / ML is data-driven. Therefore, if the training dataset becomes outdated or if the characteristics of the inference dataset differ significantly from those of the training dataset, the performance of the AI / ML model will degrade.

[0249] Therefore, the entity responsible for managing the AI / ML model (e.g., the network or UE) is instructed / configured / defined to monitor the AI / ML model using monitoring metrics (e.g., intermediate KPIs (key performance indicators), SGCS (squared generalized cosine similarity), NMSE (normalized mean square error) and / or performance KPIs (e.g., throughput, accuracy), and perform updates or fine-tuning of the model when necessary. This series of operations is referred to as the LCM of the AI / ML model.

[0250] Hereinafter, a method for efficient LCM in CSI prediction based on a one-sided AI / ML model will be described. Although focusing on UE-side AI / ML, the embodiments are not limited thereto.

[0251] Proposal 1

[0252] For AI / ML model monitoring in unilateral AI / ML-based CSI prediction, the network may receive reports on some or all of the following information from the UE (e.g., NW-side monitoring), and / or the UE may use some or all of the following information (e.g., UE-side monitoring):

[0253] (i) The variance or upper / lower bound of the measurement of CQI or SINR variation over a specific time interval;

[0254] (ii) decoding statistics, e.g., PUCCH / PDSCH decoding ratio;

[0255] (iii) request a new (lower) MCS;

[0256] (iv) requesting to change the CSI prediction / observation window size;

[0257] (v) prediction accuracy (e.g., accuracy information calculated by the UE);

[0258] (vi) intermediate KPIs, e.g., SGCS, NMSE;

[0259] (vii) Performance KPIs, e.g., (predicted) UE throughput.

[0260] For example, regarding (ii) decoding statistics, the network can determine the MCS based on the CSI reported by the UE and schedule data according to the MCS. If the BS appropriately configures the MCS, the decoding performance of channels such as the PDSCH can be improved. Otherwise, the decoding performance may be reduced. Therefore, when the UE reports the reported decoding statistics, the network can use the reported information to monitor the AI / ML model. In another example, the UE may use the MCS configured by the BS, but the decoding performance may be poor. The BS can reconfigure the MCS, or the UE can provide / determine information related to the preferred MCS level.

[0261] For example, in case (iv), the CSI prediction / observation window can be associated with the input / output of the AI / ML model. Therefore, depending on changes in data statistics, the configuration of the prediction / observation window used for the initial setup (deployment) of the AI / ML model can be changed (e.g., the window size can be adjusted). In particular, when the configured prediction window is larger than required by the current channel conditions, some predicted CSI within the window may have low accuracy. Therefore, requests to change the prediction / observation window size can also be used for monitoring of the AI / ML model.

[0262] For example, regarding (v), prediction accuracy can be a metric calculated by the UE based on a specific definition, or a metric calculated as an intermediate KPI based on an AI / ML model. Specific definitions may include, but are not limited to, the similarity between actual measured and predicted CSI, or the mean square error (MSE) between predicted CSI and actual measurements. Additionally, a separate measurement RS can be configured within the prediction window to support LCM-related accuracy calculation / reporting.

[0263] In one example, (i) CQI / SINR variation may represent the CQI / SINR variation between some or all predicted CSIs within a prediction window, such as Figure 19 For example, you can use Figure 19 The difference (e.g., differential value) between the first and fifth CQIs in .

[0264] In addition, all CSIs within the prediction window are reported ( Figure 19Part 1 CSI can increase payload overhead by limiting the number of CSIs reported within the prediction window (the first to fifth CSIs in the prediction window). To address this issue, the base station can set the number of CSIs to be reported within the prediction window (K value) and / or the maximum number of CSIs to be reported for the UE (K_max). The UE can then select and report K CSIs or fewer than K_max CSIs within the prediction window. The UE can report information about the selected predicted CSIs and the number of predicted CSIs to the base station using, for example, a bitmap, and such information can be included in the Part 1 CSI.

[0265] Proposal 2

[0266] In unilateral AI / ML-based CSI prediction, when the network is the entity responsible for monitoring the AI / ML model, the signaling related to AI / ML model update / fine-tuning may include some or all of the following:

[0267] (i) AI / ML model update signaling

[0268] - When AI / ML model training is performed on the network side or by a third party (e.g., AI server): updated information about AI / ML parameters and / or structure (e.g., new parameterization), CSI prediction configuration (e.g., prediction / observation window information), and target values for intermediate KPIs / performance KPIs can be provided.

[0269] - When performing AI / ML model training on the UE side: CSI prediction configuration (e.g., prediction / observation window information) and target values for intermediate KPIs / performance KPIs can be provided.

[0270] - Dataset signaling (or signaling of a reference resource for retraining) can be included.

[0271] (ii) AI / ML model update period and / or AI / ML model activation / deactivation commands may be provided.

[0272] (iii) AI / ML model switching can be directional.

[0273] (iv) Can provide operation / configuration / instructions in fallback mode

[0274] Proposal 2 can also be applied when the UE performs inference while the network monitors the AI / ML model.

[0275] The BS may indicate / configure information related to the AI / ML model update for the UE and signal an activation command for the updated AI / ML model and / or an AI / ML model update completion report request to the UE. The activation command / completion report request may be explicitly indicated by a MAC-CE or the like, or an AI / ML model update timer or the like may be set to indicate activation based on the expiration of the AI / ML model update timer. The completion report request is a request made to the UE to report to the network whether the AI / ML model update has been successfully completed. If the AI / ML model update has been successfully completed, the UE may report an ACK. Otherwise, it may report a NACK.

[0276] For example, when information related to an AI / ML model update is signaled, an AI / ML model update timer may be started (or restarted). When the AI / ML model update timer expires, the BS may determine that the AI / ML model update or fine-tuning has been completed. The value of the AI / ML model update timer may be set / indicated based on the capabilities of the UE. Furthermore, if the UE successfully completes the AI / ML model update procedure before the AI / ML model update timer expires, the UE may send signaling (e.g., an ACK) to the BS indicating that the AI / ML model update was successfully completed before the timer expired.

[0277] In addition, when signaling related to the setup or activation of an AI / ML model update / fine-tuning is provided to the UE, the UE considers it to be the same as a deactivation command for an existing activated AI / ML model and suspends the operation of the existing AI / ML model / function until new signaling / command for activating / switching the AI / ML model is instructed / configured. In order to efficiently use the memory related to the AI / ML model or the UE's capacity related to the maximum number of AI / ML models that can be managed by the UE, the AI / ML model may be deleted / deregistered / released. In this case, the UE may report information related to the deletion / deregistration / release of the AI / ML model to the network (e.g., the BS or the AI / ML server).

[0278] As an example, in conjunction with the AI / ML model update completion operation, the UE may immediately activate the updated AI / ML model. This scenario may be limited to relatively simple procedures, such as fine-tuning rather than a complete update of the AI / ML model, or scenarios where only (some) parameters (rather than the structure) of the AI / ML model are updated during the AI / ML model update.

[0279] For example, the UE may treat the signaling as a deactivation command for an existing activated AI / ML model, suspend the operation of the existing AI / ML model / function until new signaling / command for activation / switching of the AI / ML model, etc. is provided, and operate in a fallback mode (e.g., non-AI mode, such as non-predictive CSI reporting in the case of CSI prediction).

[0280] In the above proposal, when the appropriate AI / ML model for each channel condition (e.g., configuration ID / area ID / scenario ID) has been pre-loaded in the UE but is currently inactive, the network may only need to instruct the UE to switch to the AI / ML model. The UE can also report to the BS whether the AI / ML model switching has been successfully completed.

[0281] Proposal 3

[0282] In unilateral AI / ML-based CSI prediction, when the UE is the entity responsible for AI / ML model monitoring, some or all of the following items may be signaled regarding updates / fine-tuning of the AI / ML model:

[0283] (i) Dataset-related signaling:

[0284] - Requests for new datasets for AI / ML model retraining / updates, or requests for reference signaling.

[0285] (ii) Requests to configure AI cycles / ML model updates / fine-tuning / retraining, or to set fallback mode.

[0286] (iii) Requesting reporting of AI / ML model update completion and / or activating the updated AI / ML model.

[0287] (iv) AI / ML model switching.

[0288] In Proposal 3, the inference object is the UE, and the UE can also be the object monitored by the AI / ML model.

[0289] As in Proposal 2, dataset-related signaling is necessary for retraining / fine-tuning / AI / ML model parameter / structure updates. However, if an appropriate AI / ML model for each channel condition (e.g., configuration ID / region ID / scene ID) is already available, requesting an AI / ML model switch or reporting the switch completion to the network may be sufficient.

[0290] Once the period for AI / ML model update / fine-tuning / retraining is set by the BS, the UE may suspend transmission / reception of at least some or all channels / signals during the set period for update / fine-tuning / retraining. Alternatively, the UE may be allowed to perform only minimal operations such as SSB reception during this period. The number of channels / signals to be transmitted and received and the channel signals to be transmitted and received may be determined based on the capabilities of the UE. Upon completion of the AI / ML model update / fine-tuning / retraining, or simultaneously with the corresponding signaling transmission, the suspended transmission / reception of the relevant channels / signals may be resumed. Alternatively, the UE may operate in a fallback mode (e.g., non-AI mode) during this period.

[0291] As another example, the UE may operate using an existing activated AI / ML model until the update of the AI / ML model is completed. To this end, information such as an AI / ML model ID and an AI / ML model version may be assigned to each AI / ML model.

[0292] Once the UE completes the AI / ML model update / fine-tuning / retraining, the UE may report the completion and / or request activation of the updated / fine-tuned / retrained AI / ML model.

[0293] In Proposal 2 / 3, the activation / update / switching of AI / ML models can be linked to RACH resources. For example, AI / ML model / version information can be linked to RACH resources. Once the UE transmits a random access preamble using RACH resources, the BS can indicate / configure the AI / ML model / version to be applied to the UE via a response (e.g., RAR).

[0294] Furthermore, even during handover, AI / ML model / version information can be provided to the target cell (each target cell) for prompt application of the AI / ML model. Along with or separately from the AI / ML model / version information, the target cell (each target cell) can be provided with information such as the area ID / configuration ID / scenario ID / site ID related to the training / inference of i) the AI / ML model associated with the UE's location information, ii) the AI / ML model currently used by the UE, and / or iii) the AI / ML model being served by the BS to the UE. The area ID / configuration ID / scenario ID / site ID can be associated with the dataset used for training / inference of the AI / ML model.

[0295] Although Proposals 1 / 2 / 3 have been described based on AI / ML-based CSI prediction, they can also be applied to time beam prediction and general LCM of unilateral AI / ML models.

[0296] Figure 20is a diagram illustrating management of an AI / ML model for CSI prediction in a wireless communication system according to an embodiment.

[0297] refer to Figure 20 The UE may receive one or more RRC configurations from the network (A05). The RRC configurations may include, for example, configurations for CSI measurement / reporting and / or configurations for AI / ML models. The configurations for CSI measurement / reporting may be related to CSI prediction based on the AI / ML model.

[0298] The UE configures an AI / ML model based on the RRC configuration (A10). The AI / ML model may be a unilateral AI / ML model configured for CSI prediction.

[0299] The UE may monitor the AI / ML model (A20). By monitoring the AI / ML model, the UE may obtain information about the performance of CSI prediction. The information about the performance of CSI prediction may be determined based on at least one of a change in a channel quality indicator (CQI) or a signal-to-interference-and-noise ratio (SINR) during a specific period, CSI prediction accuracy, or downlink channel decoding performance. The specific period may be a CSI prediction window that includes the CSI prediction time point.

[0300] When the UE determines that the size of the CSI prediction window needs to be changed to monitor the AI / ML model, it may request the network to change the size of the CSI prediction window (A15).

[0301] Based on the information about the performance of the CSI prediction, the UE may send a first request (A25) for lifecycle management (LCM) of the AI / ML model to the network. As an example, the first request may include at least one of the following: (i) sending information for requesting a dataset for updating the AI / ML model, (ii) sending information for requesting setting a period for performing the update of the AI / ML model, (iii) sending information for requesting switching of the AI / ML model, or (iv) sending information for requesting a fallback to non-AI / ML-based operation.

[0302] In response to the first request from the UE, the network may send a first response to the LCM of the AI / ML model (A30). For example, the first response may include at least one of: (i) sending a dataset for updating the AI / ML model, (ii) sending information for setting a time period in which the AI / ML model is to be updated, (iii) sending information instructing switching of the AI / ML model, or (iv) sending information instructing a fallback to non-AI / ML-based operation. The dataset may be used for training or retraining the AI / ML model.

[0303] The UE may perform an update / switching of the AI / ML model based on the first response (A35). During a period in which the update or switching of the AI / ML model is performed, transmission or reception of at least one signal may be suspended.

[0304] Based on the update or switching of the AI / ML model, the UE may send a second request for the LCM of the AI / ML model to the network (A40). For example, the second request may include at least one of the following: (v) sending information notifying completion of the update or switching of the AI / ML model, or (vi) sending information requesting activation of the updated or switched AI / ML model.

[0305] In response to the second request from the UE, the network may send a second response to the LCM of the AI / ML model (A30). For example, the second response may include information indicating activation of the updated or switched AI / ML model.

[0306] Figure 21 is a diagram illustrating an operation of a terminal according to an embodiment.

[0307] refer to Figure 21 , the UE can configure the artificial intelligence / machine learning (AI / ML) model (B05).

[0308] The UE may obtain information about the performance of the channel state information (CSI) prediction of the AI / ML model through monitoring of the AI / ML model (B10).

[0309] The UE may perform lifecycle management (LCM) related procedures for the AI / ML model based on the acquired information on the performance of the CSI prediction (B15).

[0310] The LCM-related process may include at least one of: (i) sending information for requesting a data set for updating of the AI / ML model, (ii) sending information for requesting setting of a period in which the updating of the AI / ML model is to be performed, (iii) sending information for requesting switching of the AI / ML model, or (iv) sending information for requesting fallback to non-AI / ML based operation.

[0311] The LCM-related process may further include at least one of: (v) transmitting information notifying completion of update or switching of the AI / ML model, or (vi) transmitting information for requesting activation of the updated or switched AI / ML model.

[0312] The information on the performance of the CSI prediction may be determined based on at least one of a change in a channel quality indicator (CQI) or a signal to interference and noise ratio (SINR) during a specific period, accuracy of the CSI prediction, and decoding performance of a downlink channel.

[0313] The specific period may be a CSI prediction window including a CSI prediction time point.

[0314] The UE may transmit information for requesting a change in the size of a CSI prediction window for monitoring an AI / ML model.

[0315] Datasets can be used to train or retrain AI / ML models.

[0316] During a period in which updating of the AI / ML model is to be performed, transmission or reception of at least one signal may be suspended.

[0317] The AI / ML model can be a unilateral AI / ML model configured for the UE through network signaling.

[0318] Figure 22 is a diagram illustrating an operation of a base station according to an embodiment.

[0319] Reference Figure 22 , the BS may send information (C05) for configuring an artificial intelligence / machine learning (AI / ML) model to the UE.

[0320] The BS may perform a lifecycle management (LCM)-related procedure for an AI / ML model based on the performance of channel state information (CSI) prediction of the AI / ML model configured for the UE (C10).

[0321] The LCM-related process may include at least one of the following: (i) sending a data set for updating of the AI / ML model, (ii) sending information for setting a period in which the updating of the AI / ML model is to be performed, (iii) sending information indicating switching of the AI / ML model, or (iv) sending information indicating fallback to non-AI / ML-based operation.

[0322] The LCM-related process may further include at least one of: (v) receiving information notifying completion of update or switching of the AI / ML model, and (vi) sending information indicating activation of the updated or switched AI / ML model.

[0323] The information on the performance of the CSI prediction may be determined based on at least one of a change in a channel quality indicator (CQI) or a signal to interference and noise ratio (SINR) during a specific period, accuracy of the CSI prediction, and decoding performance of a downlink channel.

[0324] The specific period may be a CSI prediction window including a CSI prediction time point.

[0325] The BS may receive information requesting a change in the size of a CSI prediction window for monitoring an AI / ML model from the UE.

[0326] Datasets can be used to train or retrain AI / ML models.

[0327] During a period in which updating of the AI / ML model is to be performed, transmission or reception of at least one signal may be suspended.

[0328] The AI / ML model can be a unilateral AI / ML model configured for the UE through network signaling.

[0329] Figure 23 A communication system 1 applied to the present disclosure is shown.

[0330] refer to Figure 23 The communication system 1 applied to the present disclosure includes wireless devices, a base station (BS), and a network. Herein, a wireless device refers to a device that performs communication using a radio access technology (RAT) (e.g., 5G New RAT (NR) or Long Term Evolution (LTE)) and may be referred to as a communication / radio / 5G device. Wireless devices may include, but are not limited to, a robot 100a, vehicles 100b-1 and 100b-2, an extended reality (XR) device 100c, a handheld device 100d, a home appliance 100e, an Internet of Things (IoT) device 100f, and an artificial intelligence (AI) device / server 400. For example, a vehicle may include a vehicle with wireless communication capabilities, an autonomous vehicle, and a vehicle capable of performing communication between vehicles. Herein, a vehicle may include an unmanned aerial vehicle (UAV) (e.g., a drone). XR devices may include augmented reality (AR) / virtual reality (VR) / mixed reality (MR) devices and may be implemented in the form of head-mounted devices (HMDs), vehicle-mounted heads-up displays (HUDs), televisions, smartphones, computers, wearable devices, home appliances, digital signage, vehicles, robots, and the like. Handheld devices may include smartphones, smartpads, wearable devices (e.g., smartwatches or smart glasses), and computers (e.g., laptops). Home appliances may include TVs, refrigerators, and washing machines. IoT devices may include sensors and smart meters. For example, a base station (BS) and a network may be implemented as wireless devices, and a specific wireless device 200a may operate as a base station (BS) / network node relative to other wireless devices.

[0331] Wireless devices 100a to 100f can connect to a network 300 via a base station (BS) 200. AI technology can be applied to wireless devices 100a to 100f, and wireless devices 100a to 100f can connect to an AI server 400 via the network 300. Network 300 can be configured using a 3G network, a 4G network (e.g., LTE), or a 5G network (e.g., NR). While wireless devices 100a to 100f can communicate with each other via BS 200 / network 300, wireless devices 100a to 100f can also communicate directly with each other (e.g., sidelink communication) without going through the BS / network. For example, vehicles 100b-1 and 100b-2 can communicate directly (e.g., vehicle-to-vehicle (V2V) / vehicle-to-everything (V2X) communication). IoT devices (e.g., sensors) can communicate directly with other IoT devices (e.g., sensors) or other wireless devices 100a to 100f.

[0332] Wireless communication / connections 150a, 150b, or 150c may be established between wireless devices 100a to 100f / BS 200, or between BS 200 / BS 200. Here, the wireless communication / connections may be established via various RATs (e.g., 5G NR), such as uplink / downlink communication 150a, sidelink communication 150b (or D2D communication), or inter-BS communication (e.g., relay, integrated access backhaul (IAB)). The wireless devices and BSs / wireless devices may transmit / receive radio signals to / from each other via the wireless communication / connections 150a and 150b. For example, the wireless communication / connections 150a and 150b may transmit / receive signals via various physical channels. To this end, at least a portion of various configuration information for configuring processes for transmitting / receiving radio signals, various signal processing processes (e.g., channel coding / decoding, modulation / demodulation, and resource mapping / demapping), and resource allocation processes may be implemented based on various proposals of the present disclosure.

[0333] Figure 24 A wireless device suitable for use with the present disclosure is shown.

[0334] refer to Figure 24 , the first wireless device 100 and the second wireless device 200 may transmit radio signals via various RATs (eg, LTE and NR). Herein, {the first wireless device 100 and the second wireless device 200} may correspond to Figure 23 {wireless device 100x and BS 200} and / or {wireless device 100x and wireless device 100x}.

[0335] The first wireless device 100 may include one or more processors 102 and one or more memories 104, and may further include one or more transceivers 106 and / or one or more antennas 108. The processor 102 may control the memory 104 and / or the transceiver 106 and may be configured to implement the descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed herein. For example, the processor 102 may process information within the memory 104 to generate first information / signals, and then transmit a radio signal including the first information / signals through the transceiver 106. The processor 102 may receive a radio signal including second information / signals through the transceiver 106, and then store information obtained by processing the second information / signals in the memory 104. The memory 104 may be connected to the processor 102 and may store various information related to the operation of the processor 102. For example, the memory 104 may store software code including commands for executing some or all of the processes controlled by the processor 102 or for executing the descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed herein. Herein, processor 102 and memory 104 may be part of a communication modem / circuit / chip designed to implement a RAT (e.g., LTE or NR). Transceiver 106 may be connected to processor 102 and transmit and / or receive radio signals via one or more antennas 108. Each transceiver 106 may include a transmitter and / or a receiver. Transceiver 106 may be used interchangeably with radio frequency (RF) units. In this disclosure, a wireless device may refer to a communication modem / circuit / chip.

[0336] The second wireless device 200 may include one or more processors 202 and one or more memories 204, and may also include one or more transceivers 206 and / or one or more antennas 208. The processor 202 may control the memory 204 and / or the transceiver 206 and may be configured to implement the descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed herein. For example, the processor 202 may process information within the memory 204 to generate third information / signals, and then transmit a radio signal including the third information / signals through the transceiver 206. The processor 202 may receive a radio signal including fourth information / signals through the transceiver 206, and then store information obtained by processing the fourth information / signals in the memory 204. The memory 204 may be connected to the processor 202 and may store various information related to the operation of the processor 202. For example, the memory 204 may store software code including commands for executing some or all of the processes controlled by the processor 202 or for executing the descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed herein. Herein, the processor 202 and memory 204 may be part of a communication modem / circuit / chip designed to implement a RAT (e.g., LTE or NR). The transceiver 206 may be connected to the processor 202 and transmit and / or receive radio signals via one or more antennas 208. Each transceiver 206 may include a transmitter and / or a receiver. The term "transceiver 206" may be used interchangeably with "RF unit." In this disclosure, a wireless device may refer to a communication modem / circuit / chip.

[0337] The hardware elements of wireless devices 100 and 200 will be described in more detail below. One or more protocol layers may be implemented by (but not limited to) one or more processors 102 and 202. For example, one or more processors 102 and 202 may implement one or more layers (e.g., functional layers such as PHY, MAC, RLC, PDCP, RRC, and SDAP). One or more processors 102 and 202 may generate one or more protocol data units (PDUs) and / or one or more service data units (SDUs) according to the descriptions, functions, procedures, proposals, methods, and / or operational flowcharts disclosed in this document. One or more processors 102 and 202 may generate messages, control information, data, or information according to the descriptions, functions, procedures, proposals, methods, and / or operational flowcharts disclosed in this document. The one or more processors 102 and 202 may generate a signal (e.g., a baseband signal) including a PDU, SDU, message, control information, data, or information according to the descriptions, functions, procedures, proposals, methods, and / or operational flowcharts disclosed in this document, and provide the generated signal to the one or more transceivers 106 and 206. The one or more processors 102 and 202 may receive a signal (e.g., a baseband signal) from the one or more transceivers 106 and 206 and obtain the PDU, SDU, message, control information, data, or information according to the descriptions, functions, procedures, proposals, methods, and / or operational flowcharts disclosed in this document.

[0338] The one or more processors 102 and 202 may be referred to as controllers, microcontrollers, microprocessors, or microcomputers. The one or more processors 102 and 202 may be implemented using hardware, firmware, software, or a combination thereof. For example, one or more application-specific integrated circuits (ASICs), one or more digital signal processors (DSPs), one or more digital signal processing devices (DSPDs), one or more programmable logic devices (PLDs), or one or more field-programmable gate arrays (FPGAs) may be included in the one or more processors 102 and 202. The descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed in this document may be implemented using firmware or software, and the firmware or software may be configured to include modules, processes, or functions. The firmware or software configured to execute the descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed in this document may be included in the one or more processors 102 and 202 or stored in the one or more memories 104 and 204 to be driven by the one or more processors 102 and 202. The descriptions, functions, processes, proposals, methods and / or operational flow charts disclosed in this document may be implemented in the form of codes, commands and / or command sets using firmware or software.

[0339] One or more memories 104 and 204 may be connected to one or more processors 102 and 202 and store various types of data, signals, messages, information, programs, codes, instructions, and / or commands. One or more memories 104 and 204 may be configured by read-only memory (ROM), random access memory (RAM), electrically erasable programmable read-only memory (EPROM), flash memory, a hard drive, registers, cache memory, a computer-readable storage medium, and / or a combination thereof. One or more memories 104 and 204 may be located internally and / or externally to one or more processors 102 and 202. One or more memories 104 and 204 may be connected to one or more processors 102 and 202 via various technologies, such as wired or wireless connections.

[0340] One or more transceivers 106 and 206 may transmit user data, control information, and / or radio signals / channels mentioned in the methods and / or operational flowcharts of this document to one or more other devices. One or more transceivers 106 and 206 may receive user data, control information, and / or radio signals / channels mentioned in the descriptions, functions, processes, proposals, methods, and / or operational flowcharts disclosed in this document from one or more other devices. For example, one or more transceivers 106 and 206 may be connected to one or more processors 102 and 202 and transmit and receive radio signals. For example, one or more processors 102 and 202 may execute control so that one or more transceivers 106 and 206 may transmit user data, control information, or radio signals to one or more other devices. One or more processors 102 and 202 may execute control so that one or more transceivers 106 and 206 may receive user data, control information, or radio signals from one or more other devices. One or more transceivers 106 and 206 may be connected to one or more antennas 108 and 208, and may be configured to transmit and receive user data, control information, and / or radio signals / channels as described in the descriptions, functions, processes, proposals, methods, and / or operational flow charts disclosed herein via the one or more antennas 108 and 208. In this document, one or more antennas may be multiple physical antennas or multiple logical antennas (e.g., antenna ports). The one or more transceivers 106 and 206 may convert received radio signals / channels, etc., from RF band signals to baseband signals so that the received user data, control information, radio signals / channels, etc., may be processed by the one or more processors 102 and 202. The one or more transceivers 106 and 206 may convert user data, control information, radio signals / channels, etc., processed by the one or more processors 102 and 202, from baseband signals to RF band signals. To this end, one or more of the transceivers 106 and 206 may include (analog) oscillators and / or filters.

[0341] Figure 25 Another example of a wireless device applicable to the present disclosure is shown. The wireless device can be used according to the usage / service (reference Figure 23 ) are implemented in various forms.

[0342] refer to Figure 25 , the wireless devices 100 and 200 may correspond to Figure 24The wireless devices 100 and 200 may be configured by various elements, components, units / parts and / or modules. For example, each of the wireless devices 100 and 200 may include a communication unit 110, a control unit 120, a memory unit 130 and an additional component 140. The communication unit may include a communication circuit 112 and a transceiver 114. For example, the communication circuit 112 may include Figure 24 One or more processors 102 and 202 and / or one or more memories 104 and 204. For example, the transceiver 114 may include Figure 24 The control unit 120 is electrically connected to the communication unit 110, the memory unit 130, and the additional components 140, and controls the overall operation of the wireless device. For example, the control unit 120 may control the electrical / mechanical operation of the wireless device based on the program / code / command / information stored in the memory unit 130. The control unit 120 may transmit information stored in the memory unit 130 to an external device (e.g., another communication device) via the communication unit 110 via a wireless / wired interface, or may store information received from an external device (e.g., another communication device) via the communication unit 110 via a wireless / wired interface in the memory unit 130.

[0343] The additional components 140 may be configured differently depending on the type of wireless device. For example, the additional components 140 may include at least one of a power supply unit / battery, an input / output (I / O) unit, a drive unit, and a computing unit. The wireless device may be configured in the following manner: Figure 23 100a), vehicles ( Figure 23 100b-1 and 100b-2), XR devices ( Figure 23 100c), handheld device ( Figure 23 100d), household appliances ( Figure 23 100e), IoT devices ( Figure 23 100f), digital broadcasting terminals, holographic devices, public safety devices, MTC devices, medical devices, fintech devices (or financial devices), security devices, climate / environmental devices, AI servers / devices ( Figure 23 400), BS ( Figure 23 200), network nodes, etc. The wireless device can be used in a mobile or fixed location depending on the use case / service.

[0344] exist Figure 25In the present disclosure, the various elements, components, units / portions, and / or modules within wireless devices 100 and 200 may all be connected to each other via wired interfaces, or at least a portion thereof may be wirelessly connected via communication unit 110. For example, in each of wireless devices 100 and 200, control unit 120 and communication unit 110 may be wired, and control unit 120 and first units (e.g., 130 and 140) may be wirelessly connected via communication unit 110. The various elements, components, units / portions, and / or modules within wireless devices 100 and 200 may also include one or more elements. For example, control unit 120 may be configured by a collection of one or more processors. As an example, control unit 120 may be configured by a collection of a communication control processor, an application processor, an electronic control unit (ECU), a graphics processing unit, and a memory control processor. As another example, memory unit 130 may be configured by random access memory (RAM), dynamic RAM (DRAM), read-only memory (ROM), flash memory, volatile memory, non-volatile memory, and / or a combination thereof.

[0345] Figure 26 A vehicle or autonomous driving vehicle applicable to the present disclosure is shown. The vehicle or autonomous driving vehicle may be implemented by a mobile robot, a car, a train, a manned / unmanned aerial vehicle (AV), a ship, etc.

[0346] refer to Figure 26 , the vehicle or autonomous driving vehicle 100 may include an antenna unit 108, a communication unit 110, a control unit 120, a drive unit 140a, a power supply unit 140b, a sensor unit 140c, and an autonomous driving unit 140d. The antenna unit 108 may be configured as a part of the communication unit 110. Blocks 110 / 130 / 140a to 140d correspond to Figure 25 Blocks 110 / 130 / 140.

[0347] The communication unit 110 can send and receive signals (e.g., data and control signals) to and from external devices such as other vehicles, base stations (e.g., gNBs and roadside units), and servers. The control unit 120 can perform various operations by controlling components of the vehicle or autonomous vehicle 100. The control unit 120 may include an electronic control unit (ECU). The drive unit 140a enables the vehicle or autonomous vehicle 100 to travel on a road. The drive unit 140a may include an engine, a motor, a powertrain, wheels, brakes, a steering system, and the like. The power supply unit 140b can supply power to the vehicle or autonomous vehicle 100 and may include wired / wireless charging circuitry, a battery, and the like. The sensor unit 140c can acquire vehicle status, surrounding environment information, user information, and the like. The sensor unit 140 c may include an inertial measurement unit (IMU) sensor, a collision sensor, a wheel sensor, a speed sensor, a slope sensor, a weight sensor, a heading sensor, a location module, a vehicle forward / backward sensor, a battery sensor, a fuel sensor, a tire sensor, a steering sensor, a temperature sensor, a depth sensor, an ultrasonic sensor, a lighting sensor, a pedal position sensor, etc. The autonomous driving unit 140 d may implement a technology for maintaining the lane in which the vehicle is traveling, a technology for automatically adjusting the speed (e.g., adaptive cruise control), a technology for autonomously traveling along a determined path, a technology for traveling by automatically setting a path if a destination is set, etc.

[0348] For example, the communication unit 110 may receive map data, traffic information data, and the like from an external server. The autonomous driving unit 140d may generate an autonomous driving route and driving plan based on the acquired data. The control unit 120 may control the drive unit 140a so that the vehicle or autonomous driving vehicle 100 moves along the autonomous driving route according to the driving plan (e.g., speed / direction control). During autonomous driving, the communication unit 110 may aperiodically or periodically acquire recent traffic information data from an external server and acquire surrounding traffic information data from neighboring vehicles. During autonomous driving, the sensor unit 140c may acquire vehicle status and / or surrounding environment information. The autonomous driving unit 140d may update the autonomous driving route and driving plan based on the newly acquired data / information. The communication unit 110 may transmit information regarding the vehicle's location, autonomous driving route, and / or driving plan to the external server. The external server may predict traffic information data based on information collected from the vehicle or autonomous driving vehicle using AI technology, etc., and provide the predicted traffic information data to the vehicle or autonomous driving vehicle.

[0349] Figure 27 is a diagram illustrating a DRX operation of a UE according to an embodiment of the present disclosure.

[0350] The UE may perform DRX operations in the procedures and / or methods described / proposed above. A UE configured with DRX can reduce power consumption by discontinuously receiving downlink signals. DRX may be performed in the RRC_IDLE state, the RRC_INACTIVE state, and the RRC_CONNECTED state. The UE performs DRX in the RRC_IDLE state and the RRC_INACTIVE state to discontinuously receive paging signals. DRX in the RRC_CONNECTED state (RRC_CONNECTED DRX) will be described below.

[0351] refer to Figure 27 , a DRX cycle includes an on-duration and a DRX opportunity. The DRX cycle defines the time interval between periodic repetitions of the on-duration. The on-duration is the time period during which the UE monitors the PDCCH. When a UE is configured with DRX, the UE performs PDCCH monitoring during the on-duration. When the UE successfully detects a PDCCH during the PDCCH monitoring period, the UE starts an inactivity timer and remains awake. Conversely, when the UE fails to detect any PDCCH during the PDCCH monitoring period, the UE transitions to a sleep state after the on-duration. Therefore, when DRX is configured, PDCCH monitoring / reception may be performed discontinuously in the time domain in the processes and / or methods described / proposed above. For example, when DRX is configured, PDCCH reception opportunities (e.g., time slots with PDCCH SSs) may be configured discontinuously according to the DRX configuration in the present disclosure. Conversely, when DRX is not configured, PDCCH monitoring / reception may be performed continuously in the time domain. For example, when DRX is not configured, PDCCH reception opportunities (e.g., time slots with PDCCH SSs) may be configured continuously in the present disclosure. Regardless of whether DRX is configured, PDCCH monitoring may be restricted during time periods configured as measurement gaps.

[0352] The DRX configuration information is received through higher layer signaling (eg, RRC signaling), and DRX on / off is controlled by DRX commands from the MAC layer. Once DRX is configured, the UE may discontinuously perform PDCCH monitoring while executing the above-described / proposed procedures and / or methods.

[0353] The above-mentioned embodiments correspond to the combination of elements and features of the present disclosure in a prescribed form. Furthermore, unless explicitly mentioned, each element or feature may be considered as optional. Each element or feature may be implemented in a form that is not combined with other elements or features. In addition, it is possible to implement the embodiments of the present disclosure by partially combining elements and / or features together. The order of operations described for each embodiment of the present disclosure may be modified. Some configurations or features of one embodiment may be included in another embodiment, or may replace the corresponding configuration or features of another embodiment. Furthermore, it will be apparent that the embodiments are configured by combining claims that do not have a clear reference relationship in the appended claims, or may be included as new claims by amendment after submitting the application.

[0354] Those skilled in the art will appreciate that the present disclosure may be implemented in other specific forms than those set forth herein without departing from the spirit and essential characteristics of the present disclosure. The above-described embodiments are therefore to be construed in all respects as illustrative and not restrictive. The scope of the present disclosure is to be determined by the appended claims and their legal equivalents, not by the foregoing description, and all changes coming within the meaning and range of equivalence of the appended claims are intended to be encompassed therein.

[0355] Industrial Applicability

[0356] The present disclosure is applicable to UE, BS or other devices in a wireless mobile communication system.

Claims

1. A method performed by a terminal in a wireless communication system, the method comprising: Configure artificial intelligence / machine learning (AI / ML) models; Obtaining information about the performance of channel state information (CSI) prediction of the AI / ML model through monitoring of the AI / ML model; and performing lifecycle management (LCM) related processes for the AI / ML model based on the acquired information about the performance of the CSI prediction, The LCM-related process includes at least one of the following: (i) sending information for requesting a data set for updating the AI / ML model; (ii) sending information for requesting setting of a time period in which the update of the AI / ML model is to be performed; (iii) sending information for requesting switching of the AI / ML model; or (iv) sending information for requesting a fallback to non-AI / ML-based operation.

2. The method according to claim 1, wherein The LCM-related process further includes at least one of: (v) sending information notifying completion of the update or switch of the AI / ML model, or (vi) sending information for requesting activation of the updated or switched AI / ML model.

3. The method according to claim 1, wherein The information about the performance of the CSI prediction is determined based on at least one of: a change in a channel quality indicator (CQI) or a signal to interference and noise ratio (SINR) during a specific period; an accuracy of the CSI prediction; or decoding performance of a downlink channel.

4. The method according to claim 3, wherein: The specific period is a CSI prediction window including a CSI prediction time point.

5. The method according to claim 4, further comprising: Information for requesting a change in the size of the CSI prediction window for monitoring the AI / ML model is sent.

6. The method according to claim 1, wherein The dataset is a dataset used to train or retrain the AI / ML model.

7. The method according to claim 1, wherein The transmission or reception of at least one signal is suspended during a period in which the update of the AI / ML model is to be performed.

8. The method according to claim 1, wherein The AI / ML model is a unilateral AI / ML model configured for the terminal through network signaling.

9. A computer-readable recording medium having recorded thereon a program for executing the method according to claim 1.

10. An apparatus for wireless communication, the apparatus comprising: a memory configured to store instructions; as well as a processor configured to perform operations by executing the instructions, The operation of the processor includes: Configure artificial intelligence / machine learning (AI / ML) models; Obtaining information about the performance of channel state information (CSI) predictions of the AI / ML model through monitoring of the AI / ML model; and performing lifecycle management (LCM) related processes for the AI / ML model based on the acquired information about the performance of the CSI prediction, The LCM-related process includes at least one of the following: (i) sending information for requesting a data set for updating the AI / ML model; (ii) sending information for requesting setting of a time period in which the update of the AI / ML model is to be performed; (iii) sending information for requesting switching of the AI / ML model; or (iv) sending information for requesting a fallback to non-AI / ML-based operation.

11. The apparatus according to claim 10, further comprising a transceiver, in, The apparatus is a terminal operating in a wireless communication system.

12. The device according to claim 10, wherein The apparatus is a processing device configured to control a terminal operating in a wireless communication system.

13. A method performed by a base station in a wireless communication system, the method comprising: Send information to the endpoint for configuring artificial intelligence / machine learning (AI / ML) models; as well as performing a lifecycle management (LCM)-related process for the AI / ML model based on performance of channel state information (CSI) prediction of the AI / ML model configured for the terminal; The LCM-related process includes at least one of the following: (i) sending a data set for updating the AI / ML model; (ii) sending information for setting a time period in which the update of the AI / ML model is to be performed; (iii) sending information indicating switching of the AI / ML model; or (iv) sending information indicating a fallback to non-AI / ML-based operation.

14. A base station for wireless communication, the base station comprising: a memory configured to store instructions; as well as a processor configured to perform operations by executing the instructions, The operation of the processor includes: Sending information to the endpoint for configuring artificial intelligence / machine learning (AI / ML) models; and performing a lifecycle management (LCM)-related process for the AI / ML model based on performance of channel state information (CSI) prediction of the AI / ML model configured for the terminal; The LCM-related process includes at least one of the following: (i) sending a data set for updating the AI / ML model; (ii) sending information for setting a time period in which the update of the AI / ML model is to be performed; (iii) sending information indicating switching of the AI / ML model; or (iv) sending information indicating a fallback to non-AI / ML-based operation.

Images