Method and apparatus for signal transmission and reception in wireless communication system

Abstract

A method performed by a first device in a wireless communication system according to at least one of embodiments disclosed in the present specification may comprise the steps of: transmitting, to a second device, a first signal including information on a first artificial intelligence / machine learning (AI / ML) model configured in the first device; and receiving, from the second device, a second signal including information on a second AI / ML model configured in the second device in response to the first signal, in which the first AI / ML model configured in the first device and the second AI / ML model configured in the second device are linked to each other, and the second signal includes at least one of information about a reference AI / ML model used in the second device in order to evaluate performance of the first AI / ML model or information about conditions for applying the second AI / ML model.

Description

Technical Field

[0001] This disclosure relates to wireless communication systems, and more specifically, to a method and apparatus for transmitting / receiving uplink / downlink wireless signals in a wireless communication system. Background Technology

[0002] Typically, wireless communication systems are evolving to provide communication services such as audio communication and data communication by covering a wider range of areas. Wireless communication is a multiple access system capable of supporting communication with multiple users by sharing available system resources (e.g., bandwidth, transmission power, etc.). For example, a multiple access system can be any of the following: 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). Summary of the Invention

[0003] Technical issues

[0004] The purpose of this disclosure is to provide a method and apparatus for performing the processing of transmitting and receiving wireless signals more accurately and efficiently.

[0005] The intended purpose is not limited to this; other purposes can be inferred from the disclosure in this specification.

[0006] Technical solution

[0007] According to one aspect of this disclosure, a method performed by a first device in a wireless communication system may include: transmitting to a second device a first signal including information about a first artificial intelligence / machine learning (AI / ML) model configured in the first device; and, in response to the first signal, receiving from the second device a second signal including information about a second AI / ML model configured in the second device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device may be associated with each other, and the second signal may include at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about conditions for applying the second AI / ML model.

[0008] Information regarding the conditions for applying the second AI / ML model may include at least one of a region identifier or a dataset identifier.

[0009] The second AI / ML model can be determined based on information about the first AI / ML model.

[0010] The second signal may include information about the performance of the first AI / ML model, which is evaluated based on a reference AI / ML model.

[0011] The second signal may include information about the level of uncertainty in the performance evaluation of the first AI / ML model based on the reference AI / ML model.

[0012] The first device can check the compatibility between the first AI / ML model and the second AI / ML model.

[0013] Based on compatibility being less than a threshold, the first device can request the second device to switch or update the second AI / ML model.

[0014] Based on compatibility being below a threshold, the first device can operate in a fallback mode that is not related to AI / ML.

[0015] The first device can send an acknowledgment (ACK) message to the second device indicating that the first AI / ML model and the second AI / ML model are compatible.

[0016] The first device may be a user equipment (UE).

[0017] The second device can be a base station (BS) or a network node.

[0018] According to another aspect, a computer-readable recording medium containing a program for performing the above-described methods can be provided.

[0019] According to another aspect of this disclosure, a first device in a wireless communication system may include: a memory configured to store instructions; and a processor configured to perform operations by executing the instructions. The operations performed by the processor may include: sending a first signal to a second device including information about a first AI / ML model configured in the first device, and, in response to the first signal, receiving a second signal from the second device including information about a second AI / ML model configured in the second device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device may be associated with each other, and the second signal may include at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about conditions for applying the second AI / ML model.

[0020] The first device may also include a transceiver.

[0021] The first device can be a UE.

[0022] The second device can be a network node or a BS.

[0023] According to another aspect of this disclosure, a method performed by a second device in a wireless communication system may include: receiving from a first device a first signal including information about a first AI / ML model configured in the first device; and, in response to the first signal, sending to the first device a second signal including information about a second AI / ML model configured in the second device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device may be associated with each other, and the second signal may include at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about conditions for applying the second AI / ML model.

[0024] According to another aspect of this disclosure, a second device in a wireless communication system may include: a memory configured to store instructions; and a processor configured to perform operations by executing the instructions. The operations performed by the processor may include: receiving from a first device a first signal including information about a first AI / ML model configured in the first device; and, in response to the first signal, sending to the first device a second signal including information about a second AI / ML model configured in a second device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device may be associated with each other, and the second signal may include at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about conditions for applying the second AI / ML model.

[0025] Beneficial effects

[0026] According to the implementation method, wireless signal transmission and reception can be performed efficiently in a wireless communication system.

[0027] Other effects not mentioned above can be inferred from the description. Attached Figure Description

[0028] Figure 1 The physical channel used in the 3rd Generation Partnership Project (3GPP) system, which serves as an exemplary wireless communication system, and the general signal transmission method using it are illustrated.

[0029] Figure 2 The structure of a radio frame is shown.

[0030] Figure 3 The resource grid for the time slot is shown.

[0031] Figure 4 An exemplary mapping of physical channels in a time slot is shown.

[0032] Figure 5 An exemplary physical downlink shared channel (PDSCH) and ACK / NACK transmission and reception process is shown.

[0033] Figure 6 An exemplary Physical Uplink Shared Channel (PUSCH) transmission process is shown.

[0034] Figure 7 An example of a Channel State Information (CSI) related process is shown.

[0035] Figure 8 It is a diagram used to illustrate the concepts of artificial intelligence / machine learning (AI / ML) / deep learning.

[0036] Figures 9 to 12 This demonstrates various AI / ML models based on deep learning.

[0037] Figure 13 This is a diagram illustrating the splitting of AI reasoning.

[0038] Figure 14 This is a diagram illustrating the framework of 3GPP Radio Access Network (RAN) intelligence.

[0039] Figures 15 to 17 This demonstrates the AI ​​model training and inference environment.

[0040] Figure 18 This is a diagram used to illustrate AI-based CSI feedback.

[0041] Figure 19 An exemplary process for model alignment between two-sided models based on AI / ML is shown.

[0042] Figure 20 Another exemplary process for model alignment between two-sided models based on AI / ML is shown.

[0043] Figure 21 This illustrates AI / ML function / model failures and an exemplary recovery process for AI / ML function / model failures.

[0044] Figure 22 An exemplary process for model alignment between two-sided models based on AI / ML is shown.

[0045] Figure 23 An implementation example of the operation of the first device according to the embodiment is shown.

[0046] Figure 24 An example of the operation of the second device according to the embodiment is shown.

[0047] Figures 25 to 28 Examples of the communication system 1 and wireless device applied to this disclosure are shown. Detailed Implementation

[0048] The embodiments disclosed herein are applicable to various radio 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 radio technologies such as Universal Terrestrial Radio Access (UTRA) or CDMA2000. TDMA can be implemented as radio technologies such as Global System for Mobile Communications (GSM) / General Packet Radio Service (GPRS) / Enhanced Data Rate GSM Evolution (EDGE). OFDMA can be implemented as radio technologies such as IEEE 802.11 (Wireless Fidelity (Wi-Fi)), IEEE 802.16 (Global Microwave Access Interoperability (WiMAX)), IEEE 802.20, and Evolved UTRA (E-UTRA). UTRA is part of the Universal Mobile Telecommunications System (UMTS). The 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE) is part of Evolved UMTS using E-UTRA (E-UMTS), and 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.

[0049] As more and more communication devices require greater communication capacity, there is a need for enhanced mobile broadband communications compared to traditional radio access technologies (RATs). Furthermore, the ability to provide various services anytime, anywhere by connecting multiple devices and objects is another important consideration for next-generation communications. Communication system designs considering reliability and latency-sensitive services / UEs are also being discussed. Therefore, the introduction of new radio access technologies that consider enhanced mobile broadband (eMBB), massive MTC, and ultra-reliable low-latency communication (URLLC) is being discussed. In this disclosure, for simplicity, this technology will be referred to as NR (New Radio or New RAT).

[0050] For the sake of brevity, this disclosure primarily describes 3GPP NR, but the technical concepts herein are not limited thereto.

[0051] In this disclosure, the term "settings" may be replaced with "configuration," and the two are used interchangeably. Furthermore, conditional expressions (e.g., "if," "in the case of," or "when") may be replaced by "based on" or "in the state of." Additionally, the operation or software / hardware (SW / HW) configuration of the user equipment (UE) / base station (BS) can be derived / understood based on the satisfaction of corresponding conditions. When the processing on the receiving (or transmitting) side can be derived / understood from the processing on the transmitting (or receiving) side in signal transmission / reception between wireless communication devices (e.g., BS and UE), its description may be omitted. For example, signal determination / generation / encoding / transmission on the transmitting side can be understood as signal monitoring reception / decoding / determination on the receiving side. Furthermore, when referring to the UE performing (or not performing) a specific operation, this can also be interpreted as the BS expecting / assuming (or not expecting / assuming) the UE to perform that specific operation. Similarly, when referring to the BS performing (or not performing) a specific operation, this can also be interpreted as the UE expecting / assuming (or not expecting / assuming) the BS to perform that specific operation. In the following description, for ease of description, sections, implementation methods, examples, options, methods, and solutions are distinguished and indexed, but this does not mean that each of them necessarily constitutes an independent invention or that each of them should be implemented only individually. Unless explicitly contradictory, it can be deduced / understood that at least some sections, implementation methods, examples, options, methods, and solutions can be combined or omitted.

[0052] 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 UE includes data and various control information, and varies depending on the type / purpose of the information transmitted and received by the UE and BS, encompassing various physical channels.

[0053] Figure 1 The physical channel used in a 3GPP NR system and the general signal transmission method using it are shown.

[0054] When the UE is powered on again from a power-off state or enters a new cell, in step S101, the UE performs an initial cell search procedure (e.g., establishing synchronization with the BS). For this purpose, the UE receives a synchronization signal block (SSB) from the BS. The SSB includes a primary synchronization signal (PSS), a secondary synchronization signal (SSS), and a physical broadcast channel (PBCH). The UE establishes synchronization with the BS based on the PSS / SSS and obtains information such as the cell identifier (ID). The UE can obtain broadcast information in the cell based on the PBCH. The UE can receive a DL reference signal (RS) during the initial cell search procedure to monitor the DL channel status.

[0055] After the initial cell search, in step S102, the UE can obtain more specific system information by receiving the Physical Downlink Control Channel (PDCCH) and receiving the Physical Downlink Shared Channel (PDSCH) based on the information in the PDCCH.

[0056] In steps S103 to S106, the UE may perform a random access procedure to access the BS. For random access, the UE may send a preamble to the BS on the Physical Random Access Channel (PRACH) (S103) and receive a response message for the preamble on the PDCCH and the corresponding PDSCH (S104). In the case of contention-based random access, the UE may further perform a contention resolution procedure by sending a PRACH (S105) and receiving the PDCCH and the corresponding PDSCH (S106).

[0057] Following the aforementioned process, the UE can receive the PDCCH / PDSCH (S107) and transmit the Physical Uplink Shared Channel (PUSCH) / Physical Uplink Control Channel (PUCCH) (S108), as part of the general downlink / uplink signal transmission process. The control information sent from the UE to the BS is called Uplink Control Information (UCI). UCI includes Hybrid Automatic Repeat and Request Acknowledgment / Nack Acknowledgment (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 usually transmitted on the PUCCH, it can be transmitted on the PUSCH when control information and service data need to be transmitted simultaneously. Additionally, UCI can be transmitted aperiodically via the PUSCH according to network requests / commands.

[0058] Figure 2 The radio frame structure is shown. In NR, uplink and downlink transmissions are configured in frames. Each radio frame is 10ms long and is divided into two 5ms half-frames (HF). Each half-frame is further divided into five 1ms subframes (SF). Subframes are divided into one or more time slots, and the number of time slots in a subframe depends on the subcarrier spacing (SCS). Depending on the cyclic prefix (CP), each time slot includes 12 or 14 orthogonal frequency division multiplexing (OFDM) symbols. When using a normal CP, each time slot includes 14 OFDM symbols. When using an extended CP, each time slot includes 12 OFDM symbols.

[0059] Table 1 illustrates, for example, how the number of symbols per slot, the number of slots per frame, and the number of slots per subframe vary according to SCS when using normal CP.

[0060] [Table 1] N slot symb Number of symbols in a time slot N frame,u slot Number of time slots in a frame N subframe,u slot Number of time slots in a subframe Table 2 shows the number of symbols per slot, the number of slots per frame, and the number of slots per subframe that vary according to SCS when using extended CP.

[0061] [Table 2]

[0062] The frame structure is only an example. The number of subframes, time slots, and symbols in a frame can vary.

[0063] In NR systems, OFDM parameter sets (e.g., SCS) can be configured differently for multiple cells aggregated for a single UE. Therefore, the (absolute time) duration of time resources (e.g., SF, time slots, or TTI) (referred to as time units (TU) for simplicity) consisting of the same number of symbols can be configured differently among the aggregated cells. Here, symbols can include OFDM symbols (or CP-OFDM symbols) and SC-FDMA symbols (or Discrete Fourier Transform-Extended-OFDM (DFT-s-OFDM) symbols).

[0064] Figure 3 The resource grid shows a time slot. A time slot comprises multiple symbols in the time domain. For example, when using a normal CP, a time slot comprises 14 symbols. However, when using an extended CP, a time slot comprises 12 symbols. A carrier comprises 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 portion (BWP) can be defined as multiple consecutive physical RBs (PRBs) in the frequency domain and corresponds to a single set of parameters (e.g., SCS, CP length, etc.). A carrier can include up to N (e.g., five) BWPs. Data communication can be performed through enabled BWPs, and only one BWP can be enabled for a UE. In the resource grid, individual elements are called resource elements (REs), and a complex symbol can be mapped to individual REs.

[0065] Figure 4This illustrates an exemplary mapping of physical channels within a time slot. The PDCCH can be transmitted in the DL control area, and the PDSCH can be transmitted in the DL data area. The PUCCH can be transmitted in the UL control area, and the PUSCH can be transmitted in the UL data area. The guard period (GP) provides a time gap for transmit-to-receive mode or receive-to-transmit mode switching at the BS and UE. Some symbols in the subframe during DL-UL handover can be configured as GP.

[0066] The physical channels will be described in more detail below.

[0067] The PDCCH transmits the DCI. For example, the PDCCH (i.e., the DCI) may carry information about the transmission format and resource allocation of the DL-SCH, resource allocation information for the Uplink Shared Channel (UL-SCH), paging information for the PCH, system information for the DL-SCH, resource allocation information for higher-layer control messages (e.g., RARs transmitted on the PDCCH), transmission power control commands, information about enabling / releasing configured schedules, etc. The DCI includes Cyclic Redundancy Check (CRC). The CRC is masked using various identifiers (IDs) (e.g., Radio Network Temporary Identifiers (RNTIs)) depending on the owner or purpose of the PDCCH. For example, if the PDCCH is used for a specific UE, the CRC is masked using the UE ID (e.g., Cell-RNTI (C-RNTI)). If the PDCCH is used for paging messages, the CRC is masked using the Paging-RNTI (P-RNTI). If the PDCCH is used for system information (e.g., System Information Block (SIB)), the CRC is masked by the System Information RNTI (SI-RNTI). When the PDCCH is used for RAR, the CRC is masked by the Random Access-RNTI (RA-RNTI).

[0068] The PDCCH comprises 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 the radio channel state. A CCE comprises six Resource Element Groups (REGs), each REG defined by one OFDM symbol × one (P)RB. The PDCCH is transmitted in a Control Resource Set (CORESET). A CORESET is defined as a set of REGs with a given set of parameters (e.g., SCS, CP length, etc.). Multiple CORESETs for a UE can overlap in the time / frequency domain. A CORESET can be configured by 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 symbols (up to 3) in a CORESET can be configured via higher-layer signaling.

[0069] For PDCCH reception / detection, the UE monitors PDCCH candidates. PDCCH candidates are the Common Enquiry Points (CCEs) that the UE should monitor to detect the PDCCH. Each PDCCH candidate is defined as having 1, 2, 4, 8, or 16 CCEs according to the Algorithm (AL). Monitoring includes (blind) decoding of the PDCCH candidates. The set of PDCCH candidates decoded by the UE is defined as the PDCCH Search Space (SS). The SS can be a Common Search Space (CSS) or a UE-Specific Search Space (USS). The UE can obtain the 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 a CORESET. SSs can be defined based on the following parameters.

[0070] - controlResourceSetId: CORESET associated with SS.

[0071] - monitoringSlotPeriodicityAndOffset: PDCCH monitoring periodicity (slot) and PDCCH monitoring offset (slot).

[0072] - monitoringSymbolsWithinSlot: PDCCH monitoring symbols within a slot (e.g., the first symbol of CORESET).

[0073] - 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}.

[0074] The timing when a UE needs to monitor PDCCH candidates (e.g., time / frequency resources) is defined as a PDCCH (monitoring) timing. One or more PDCCH (monitoring) timings can be configured in a time slot.

[0075] Table 3 shows the characteristics of each SS.

[0076] [Table 3]

[0077] Table 4 shows the DCI format transmitted on the PDCCH.

[0078] [Table 4]

[0079] DCI format 0_0 can be used to schedule PUSCH based on TB (or TB level), and DCI format 0_1 ​​can be used to schedule PUSCH based on TB (or TB level) or PUSCH based on code block group (CBG) (or CBG level). DCI format 1_0 can be used to schedule PDSCH based on TB (or TB level), and DCI format 1_1 can be used to schedule PDSCH based on TB (or TB level) or PDSCH based on CBG (or CBG level) (or DL-licensed DCI). DCI formats 0_0 / 0_1 can be referred to as UL-licensed DCI or UL scheduling information, and DCI formats 1_0 / 1_1 can be referred to as DL-licensed DCI or DL ​​scheduling information. DCI format 2_0 is used to transmit dynamic slot format information (e.g., dynamic slot format indicator (SFI)) to the UE, and DCI format 2_1 is used to transmit DL preemption information to the UE. DCI formats 2_0 and / or DCI format 2_1 can be transmitted to the corresponding group of UEs on the group common PDCCH (PDCCH pointing to a group of UEs).

[0080] DCI formats 0_0 and 1_0 can be referred to as fallback DCI formats, while DCI formats 0_1 and 1_1 can be referred to as non-fallback DCI formats. In fallback DCI formats, the DCI size / field configuration remains the same regardless of the UE configuration. Conversely, in non-fallback DCI formats, the DCI size / field configuration varies depending on the UE configuration.

[0081] PDSCH transmits DL data (e.g., DL Shared Channel Transport Block (DL-SCH TB)) using modulation schemes such as Quadrature Phase Shift Keying (QPSK), 16-element Quadrature Amplitude Modulation (16QAM), 64QAM, or 256QAM. TBs are encoded as codewords. PDSCH can transmit up to two codewords. Scrambling and modulation mapping can be performed on a codeword basis, and modulation symbols generated from each codeword can be mapped to one or more layers. Each layer, along with a demodulation reference signal (DMRS), is mapped to a resource, and OFDM symbol signals are generated from the layer mapped with the DMRS and transmitted through the corresponding antenna port.

[0082] PUCCH transmits uplink control information (UCI). UCI includes the following information.

[0083] - SR (Schedule Request): Information used to request UL-SCH resources.

[0084] - HARQ (Hybrid Automatic Repeat Request) - ACK (Acknowledgement): A response to a DL data packet (e.g., a codeword) on the PDSCH. HARQ-ACK indicates whether the DL data packet was successfully received. A 1-bit HARQ-ACK can be sent in response to a single codeword. A 2-bit HARQ-ACK can be sent in response to two codewords. HARQ-ACK responses include positive ACK (simply ACK), negative ACK (NACK), discontinuous transmission (DTX), or NACK / DTX. The term HARQ-ACK is used interchangeably with HARQ ACK / NACK and ACK / NACK.

[0085] - CSI (Channel State Information): Feedback information for the DL channel. MIMO-related feedback information includes RI and PMI.

[0086] Table 5 shows exemplary PUCCH formats. Based on the PUCCH transmission duration, PUCCH formats can be divided into short PUCCH (formats 0 and 2) and long PUCCH (formats 1, 3, and 4).

[0087] [Table 5]

[0088] PUCCH format 0 transmits up to 2 bits of UCI and is mapped in a sequence-based manner for easy transmission. Specifically, the UE sends a specific UCI to the BS by transmitting one of multiple sequences on the PUCCH of PUCCH format 0. The UE only transmits the PUCCH of PUCCH format 0 in the PUCCH resource configured for the corresponding SR when the UE sends an affirmative SR.

[0089] PUCCH format 1 transmits up to 2 bits of UCI, and the modulation symbols of UCI are spread in the time domain with orthogonal overlay code (OCC) (configured differently depending on whether frequency hopping is performed). DMRS is transmitted in symbols that do not transmit modulation symbols (i.e., transmitted in time division multiplexing (TDM)).

[0090] PUCCH format 2 transmits more than 2 bits of UCI, and the modulation symbols of the DCI are transmitted using DMRS in frequency division multiplexing (FDM). The DMRS is located at a density of 1 / 3 in symbols #1, #4, #7, and #10 of a given RB. A pseudo-noise (PN) sequence is used for the DMRS sequence. Frequency hopping can be enabled for 2-symbol PUCCH format 2.

[0091] PUCCH format 3 does not support UE multiplexing within the same PRBS and transmits more than 2 bits of UCI. In other words, PUCCH resources in PUCCH format 3 do not include OCC. Modulation symbols are transmitted in TDM using DMRS.

[0092] PUCCH format 4 supports multiplexing of up to four UEs in the same PRBS and transmits more than 2 bits of UCI. In other words, PUCCH resources in PUCCH format 3 include OCC. Modulation symbols are transmitted in TDM using DMRS.

[0093] At least one of one or more cells configured in the UE can be configured for PUCCH transmission. At least the primary cell can be configured as a cell for PUCCH transmission. Based on at least one cell configured for PUCCH transmission, at least one PUCCH cell group can be configured in the UE, and each PUCCH cell group includes one or more cells. A PUCCH cell group may be simply referred to as a PUCCH group. PUCCH transmission can be configured for SCells and 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. PUCCH on the primary cell can be used by cells belonging to the primary PUCCH group, and PUCCH on the PUCCH-SCell can be used by cells belonging to the secondary PUCCH group.

[0094] PUSCH transmits UL data (e.g., UL Shared Channel Transport Block (UL-SCH TB)) and / or UCI based on CP-OFDM or DFT-s-OFDM waveforms. When PUSCH is transmitted in DFT-s-OFDM waveform, the UE transmits PUSCH via transform precoding. For example, when transform precoding is not possible (e.g., disabled), the UE may transmit PUSCH using CP-OFDM waveform, while when transform precoding is possible (e.g., enabled), the UE may transmit PUSCH using either CP-OFDM or DFT-s-OFDM waveform. PUSCH transmission can be dynamically scheduled via UL authorization in DCI, or semi-statically scheduled via higher-layer (e.g., RRC) signaling (and / or Layer 1 (L1) signaling such as PDCCH) (configured scheduling or configured authorization). PUSCH transmission can be performed in a codebook-based or non-codebook-based manner.

[0095] Figure 5 This illustrates an exemplary ACK / NACK transmission process. (Refer to...) Figure 5 The UE can detect the PDCCH in slot #n. The PDCCH includes DL scheduling information (e.g., DCI format 1_0 or DCI format 1_1). The PDCCH indicates the DL assignment offset K0 with the PDSCH and the PDSCH offset K1 with the HARQ-ACK report. For example, DCI format 1_0 and DCI format 1_1 may include the following information.

[0096] - Frequency domain resource assignment: Indicates the set of RBs assigned to the PDSCH.

[0097] - Time-domain resource assignment: Indicates the starting position (e.g., OFDM symbol index) and length (e.g., number of OFDM symbols) of K0 and the PDSCH in the time slot.

[0098] - PDSCH-to-HARQ_feedback timer indicator: Indicates K1.

[0099] - HARQ process ID (4 bits): The HARQ process ID that indicates the data (e.g., PDSCH or TB).

[0100] - PUCCH Resource Indicator (PRI): Indicates the PUCCH resource to be used for UCI transmission among multiple PUCCH resources in the PUCCH resource set.

[0101] After receiving the PDSCH in time slot #(n+K0) according to the scheduling information of time slot #n, the UE can send a UCI on the PUCCH in time slot #(n+K1). The UCI may include a HARQ-ACK response to the PDSCH. For convenience, Figure 5The assumption that the SCS of the PDSCH is equal to the SCS of the PUCCH and that slot #n1 = slot #(n+K0) should not be construed as limiting this disclosure. When the SCSs are different, K1 can be indicated / interpreted based on the SCS of the PUCCH.

[0102] When the PDSCH is configured to carry a maximum of one TB, the HARQ-ACK response can be configured in one bit. When the PDSCH is configured to carry a maximum of two TBs, the HARQ-ACK response can be configured in 2 bits if spatial bundling is not configured, and in 1 bit if spatial bundling is configured. When slot #(n+K1) is designated as the timing for HARQ-ACK transmission of multiple PDSCHs, the UCI transmitted in slot #(n+K1) includes HARQ-ACK responses for multiple PDSCHs.

[0103] Whether a UE should perform spatial binding in response to a HARQ-ACK response can be configured for each cell group (e.g., via RRC / higher-layer signaling). For example, spatial binding can be configured for individual HARQ-ACK responses sent on the PUCCH and / or on the PUSCH.

[0104] When up to two (or more) TBs (or codewords) can be received at once in the corresponding serving cell (which may be scheduled by a DCI) (e.g., when higher-layer parameters...). maxNrofCodeWordsScheduledByDCI Spatial binding is supported when 2 TB is specified. More than four layers can be used for 2TB of transmission, and up to four layers can be used for 1TB of transmission. As a result, when spatial binding is configured for a corresponding cell group, spatial binding can be performed on serving cells within the cell group that can be scheduled for more than four layers. A UE that wants to send a HARQ-ACK response via spatial binding can generate a HARQ-ACK response by performing a (bit-wise) logical AND operation on the A / N bits of multiple TBs.

[0105] For example, suppose a UE receives a 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 the ACK bit value to the BS, and when at least one TB is NACK, the UE reports the NACK bit value to the BS.

[0106] For example, when only one TB is actually scheduled in a serving cell configured to receive two TBs, the UE can generate a single A / N bit by performing a logical AND operation on the A / N bit and bit value 1 of a TB. As a result, the UE reports the A / N bit of one TB to the BS.

[0107] Multiple parallel DL HARQ processes exist at the BS / UE for DL ​​transmission. While the BS awaits HARQ feedback indicating the success or failure of a previous DL transmission, multiple parallel HARQ processes allow for continuous 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 to MAC PDUs in the buffer, and the current redundant version. Each HARQ process is identified by a HARQ process ID.

[0108] Figure 6 An exemplary PUSCH transmission process is shown. (Refer to...) 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.

[0109] - Frequency domain resource assignment: Indicates the set of RBs assigned to PUSCH.

[0110] - Time-domain resource assignment: Indicates the slot offset K2 and the start position (e.g., OFDM symbol index) and duration (e.g., number of OFDM symbols) of the PUSCH within the slot. The start symbol and length of the PUSCH can be indicated by the start and length indicator value (SLIV) or separately.

[0111] Then, the UE can send the PUSCH in time slot #(n+K2) according to the scheduling information in time slot #n. The PUSCH includes the UL-SCH TB.

[0112] CSI related operations

[0113] Figure 7 An example of a CSI-related process is shown.

[0114] 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 the following: 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.

[0115] - CSI-IM resources can be configured for UE interference measurements (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 UE zero-power (ZP)-CSI-RS. ZP-CSI-RS can be configured to be distinguished from non-zero-power (NZP)-CSI-RS.

[0116] - The UE may assume that the CSI-RS resources configured for channel measurements and the CSI-IM / NZP CSI-RS resources configured for interference measurements for a CSI report have a QCL relationship with respect to the "QCL-TypeD" of the respective resources (when the NZP CSI-RS resources are used for interference measurements).

[0117] - CSI resource configuration may include at least one of CSI-IM resources for interference measurement, NZP CSI-RS resources for interference measurement, and NZP CSI-RS resources for channel measurement. The channel measurement resource (CMR) may be an NZP CSI-RS for CSI acquisition, and the interference measurement resource (IMR) may be an NZP CSI-RS for both CSI-IM and IM.

[0118] - CSI-RS can be configured for one or more UEs. Different CSI-RS configurations can be provided for each UE, 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 locations within a time-frequency cell corresponding to one time slot and one RB. When N is 2 or greater, N-port CSI-RS can be multiplexed using CDM, FDM, and / or TDM methods. CSI-RS can be mapped to the remaining REs except for those mapped to CORESET, DMRS, and SSB. In the frequency domain, CSI-RS can be configured for the entire bandwidth, a portion of the bandwidth (BWP), or a portion of the bandwidth. CSI-RS can be transmitted in each RB within the bandwidth where CSI-RS is configured (i.e., density = 1), or CSI-RS can be transmitted in every two RBs (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 onto three subcarriers in each resource block (i.e., density = 3). One or more CSI-RS resource sets can be configured for the UE in the time domain. Each CSI-RS resource set may include one or more CSI-RS configurations. Each CSI-RS resource set can be configured as periodic, semi-persistent, or aperiodic.

[0119] - CSI report configuration may include configuration of feedback type, measurement resources, report type, etc. NZP-CSI-RS resource sets can be used for the corresponding UE's CSI report configuration. NZP-CSI-RS resource sets can be associated with CSI-RS or SSB. Multiple periodic NZP-CSI-RS resource sets can 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), Layer 1 (L1) - Reference Received Strength (RSRP), etc. (ii) Measurement resources may include configuration of downlink signals and / or downlink resources that the UE performs measurements to determine feedback information. Measurement resources can be configured as ZP and / or NZP CSI-RS resource sets associated with the CSI report configuration. NZP CSI-RS resource sets may include CSI-RS sets or SSB sets. For example, L1-RSRP can be measured for a CSI-RS set or an SSB set. (iii) Report types may include the timing of the UE reporting and the configuration of the uplink channel. Reporting time can be configured to be periodic, semi-persistent, or aperiodic. Periodic CSI reports can be sent on the PUCCH. Semi-persistent CSI reports can be sent on the PUCCH or PUSCH based on an indication of enabled / disabled MAC CE. Aperiodic CSI reports can be indicated by DCI signaling. For example, the uplink-granted CSI request field can indicate one of various report trigger sizes. Aperiodic CSI reports can be sent on the PUSCH.

[0120] The UE measures CSI based on configuration information related to CSI. CSI measurement may include receiving CSI-RS (720) and obtaining CSI by calculating the received CSI-RS (730).

[0121] The UE may send a CSI report to the BS (740). For the CSI report, the time and frequency resources available to the UE are controlled by the BS. Channel State Information (CSI) includes at least one of the following: Channel Quality Indicator (CQI), Precoding Matrix Indicator (PMI), CSI-RS Resource Indicator (CRI), SS / PBCH Block Resource Indicator (SSBRI), Layer Indicator (LI), Rank Indicator (RI), L1-RSRP, and / or L-SINR.

[0122] CSI reporting supports periodic, semi-permanent, and aperiodic temporal behavior. i) Periodic CSI reporting is performed in short and long PUCCHs. The periodicity and slot offset of periodic CSI reports can be configured by RRC, and refer to CSI-ReportConfig IE. ii) SP (semi-periodic) CSI reporting is performed in short, long, or PUSCHs. For SP CSI in short / long PUCCHs, 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 PUSCHs, 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 PUSCHs, a separate RNTI (SP-CSI C-RNTI) is used. The initial CSI report timing follows the PUSCH temporal allocation value indicated by the DCI, and subsequent CSI report timing follows the periodicity configured by RRC. DCI format 0_1 ​​may include a CSI request field and enable / disable a specific configuration of SP-CSI triggering status. SP CSI reports have the same or similar enable / disable mechanism as data transmission in the SPS PUSCH. iii) Non-periodic CSI reporting is performed in the PUSCH and triggered by the DCI. In this case, information related to the triggering of non-periodic CSI reports 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 AP CSI reports is dynamically controlled by the DCI.

[0123] The CSI codebooks defined in the NR specification (e.g., PMI codebooks) can be broadly classified into Type I and Type II codebooks. Type I codebooks are primarily designed for single-user MIMO (SU-MIMO) supporting both high-order and low-order MIMO. Type II codebooks primarily support multi-user MIMO (MU-MIMO) capable of handling up to two layers. While Type II codebooks offer more accurate CSI compared to Type I, they may also increase signaling overhead. On the other hand, enhanced Type II codebooks are introduced to address the CSI overhead associated with existing Type II codebooks. Enhanced Type II codebooks can reduce the codebook's payload by considering correlations in the frequency domain.

[0124] 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 sent completely before Part 2.

[0125] - For Type I CSI feedback, Part 1 includes the RI (if reported), CRI (if reported), and CQI of the first codeword. Part 2 includes PMI, and when RI > 4, Part 2 also includes CQI.

[0126] - For Type II CSI feedback, Part 1 includes an indication of the RI (if reported), CQI, and the number of non-zero WB amplitude coefficients for each layer of Type II CSI. Part 2 includes the PMI for Type II CSI.

[0127] - For Enhanced Type II CSI feedback, Part 1 includes an indication of the RI (if reported), CQI, and the total number of non-zero WB amplitude coefficients for all layers of Enhanced Type II CSI. Part 2 includes the PMI for Enhanced Type II CSI.

[0128] 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 the CSI report, the UE may discard part 2 of the CSI.

[0129] Semi-persistent CSI reports executed in PUCCH format 3 or 4 support type II CSI feedback, but only support part 1 of type I ICSI feedback.

[0130] Quasi-isotope (QCL)

[0131] When the channel properties of one antenna port are to be inferred from the channel properties of another antenna port, the two antenna ports are quasi-co-located. Channel properties may include one or more of the following: delay spread, Doppler spread, frequency / Doppler shift, average received power, receive timing / average delay, and spatial RX parameters.

[0132] 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 port of the PDSCH. The QCL may include qcl-Type1 of the first DLRS and qcl-Type2 of the second DLRS. The QCL type may correspond to one of the following.

[0133] - “QCL-TypeA”: {Doppler frequency shift, Doppler spread, average delay, delay spread}

[0134] - "QCL-TypeB": {Doppler frequency shift, Doppler spread}

[0135] - "QCL-TypeC": {Doppler shift, average delay}

[0136] - "QCL-TypeD": {Space Rx parameter}

[0137] Beam management (BM)

[0138] BM refers to a series of processes used to acquire and maintain the BS beam set (transmit / receive point (TRP) beam) and / or the UE beam set that can be used for DL ​​and UL transmission / reception. BM may include the following processes and terms.

[0139] - Beam measurement: The operation of a BS or UE to measure the characteristics of the received beamforming signal.

[0140] - Beam determination: The operation of the BS or UE to select its Tx / Rx beam.

[0141] - Beam sweeping: The operation of covering a spatial domain with Tx and / or Rx beams within a specified time interval according to a predetermined method.

[0142] - Beam Report: The UE reports information about the signal based on beamforming based on beam measurement.

[0143] The BM process can be divided into (1) the DL BM process using SSB or CSI-RS and (2) the UL BM process using SRS. In addition, each BM process may include a Tx beam sweep for determining the Tx beam and an Rx beam sweep for determining the Rx beam.

[0144] The DL BM process may include (1) transmission of beamforming DL RS (e.g., CSI-RS or SSB) from the BS and (2) beam reporting from the UE.

[0145] The beam report may include a preferred DL RS ID and the corresponding Reference Signal Received Power (RSRP). The DL RS ID may be an SSB Resource Indicator (SSBRI) or a CSI-RS Resource Indicator (CRI).

[0146] Artificial Intelligence / Machine Learning (AI / ML)

[0147] With the development of AI / ML technologies, the nodes and UEs constituting wireless communication networks are becoming increasingly intelligent and advanced. Particularly due to the intelligence of the network / BS, it is expected that various network / BS-specific parameter values ​​(e.g., the distribution / location of BSs, the distribution / location / material of buildings / furniture, the location / movement direction / speed of UEs, and climate information) will be rapidly optimized and derived / applied based on various environmental parameters (e.g., the transmit and receive power of each BS, the transmission power of each UE, the precoder / beam of the BS / UE, the time / frequency resource allocation of each UE, or the duplex method of the BS) will be implemented. In line with this trend, many standardization organizations (e.g., 3GPP or O-RAN) are considering introducing network / BS-specific parameter values, and research in this area is actively underway.

[0148] In a narrow sense, AI / ML can easily be called artificial intelligence based on deep learning, but conceptually it is shown to be... Figure 8 middle.

[0149] - Artificial intelligence: This can be interpreted as all automation that replaces human jobs with machines.

[0150] Machine learning: Machines can learn decision-making patterns from data without explicit programming rules.

[0151] - Deep Learning: This is an AI / ML model based on artificial neural networks, where the machine performs everything from unstructured data to feature extraction and determination in one go, and the algorithm relies on biological neural systems, i.e., multi-layered interconnected networks of nodes inspired by neural networks for feature extraction and transformation. Common deep learning network architectures can include deep neural networks (DNNs), recurrent neural networks (RNNs), and convolutional neural networks (CNNs).

[0152] AI / ML type classification based on various references

[0153] 1. Offline vs. Online

[0154] (1) Offline learning: This follows the 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. This offline learning method is used in most cases. In offline learning, the system does not learn incrementally. Instead, it uses all available collected data for learning and applies the results without further learning. If new data is needed for learning, learning can begin again using the new complete data.

[0155] (2) Online learning: Online learning is a method that improves performance little by little by learning incrementally from newly generated data that is continuously generated via the Internet. Specific units (batches) of data collected online are learned in real time, allowing the system to quickly adapt to changing data.

[0156] 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).

[0157] 2. Classification based on AI / ML framework concepts

[0158] (1) Centralized learning: When training data collected from multiple different nodes is reported to a centralized node, all data resources / storage / learning (e.g., supervised, unsupervised and reinforcement learning) are performed by a central node.

[0159] (2) Federated Learning: This involves configuring collective AI / ML models based on data from distributed data owners. Instead of using data in AI / ML models, local nodes / individual devices collect data and train copies of their AI / ML models, thus 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. The advantages of federated learning include increased computational speed and enhanced information security. Specifically, it eliminates the need to upload personal data to a central server for processing, preventing the leakage and misuse of personal information.

[0160] (3) Distributed learning: Machine learning processing represents the concept of scaling and deployment across node clusters. AI / ML model training is split and shared across multiple concurrently operating nodes to accelerate AI / ML model training.

[0161] 3. According to the classification of learning methods

[0162] (1) Supervised Learning: Supervised learning is a machine learning task aimed at learning a mapping function from input to output given a labeled dataset. The input data is called training data and has known labels or results. Examples of supervised learning can include: (i) Regression: Linear Regression, Logistic Regression; (ii) Instance-based algorithms: k-Nearest Neighbors (KNN); (iii) Decision Tree algorithms: CART; (iv) Support Vector Machines: SVM; (v) Bayesian algorithms: Naive Bayes; and (vi) Ensemble algorithms: Extreme Gradient Boosting, Packaging: Random Forest. Due to the nature of regression and classification problems, supervised learning can be further grouped, with classification predicting labels and regression predicting the number of labels.

[0163] (2) Unsupervised learning: This is a machine learning task aimed at learning a function that describes the hidden structure in unlabeled data. The input data is unlabeled, and there is no known outcome. Some examples of unsupervised learning include K-means clustering, principal component analysis (PCA), nonlinear independent component analysis (ICA), and LSTM.

[0164] (3) Reinforcement Learning: In reinforcement learning (RL), an agent aims to optimize a long-term objective by interacting with the environment through trial-and-error processing. This is goal-oriented learning based on interaction with the environment. Examples of RL algorithms may include (i) Q-learning, (ii) multi-armed robber learning, (iii) deep Q-networks, State-Action-Reward-State-Action (SARSA), (iv) temporal difference learning, (v) enforcer-evaluator 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 predictive AI / ML models to obtain the transition probabilities between states by using various dynamic states of the environment and the AI / ML models that cause rewards from these states. Model-free RL is a value- or policy-based RL algorithm that achieves maximum future rewards. It has lower computational complexity in terms of multi-agent environments / states and does not require an accurate representation of the environment. RL algorithms can also be classified as value-based RL versus policy-based RL, policy-based RL versus non-policy RL, etc.

[0165] AI / ML models

[0166] Figure 9 An example of a feedforward neural network (FFNN) AI / ML model is shown. (See reference...) Figure 9 The FFNN AI / ML model consists of an input layer, a hidden layer, and an output layer.

[0167] Figure 10 This shows an example of a recurrent neural network (RNN) AI / ML model. (See reference...) Figure 10 RNN AI / ML models are artificial neural networks where hidden nodes are connected to directed edges to form directed loops. They are suitable for processing sequential data (e.g., speech or text). One type of RNN is Long Short-Term Memory (LSTM), which is a structure that adds cell states to the hidden states of an RNN. Specifically, in an LSTM, input gates, forget gates, and output gates are added to the RNN cells, along with the cell states. Figure 10 In the diagram, A represents a neural network, and x... t h represents the input value. t This represents the output value. Here, h t It can represent the current state value based on time, while h t-1 It can represent the previous state value.

[0168] Figure 11This illustrates a Convolutional Neural Network (CNN) AI / ML model. CNNs serve two purposes: reducing the complexity of AI / ML models by applying convolutional computations commonly used in video or image processing, and extracting good features. (See reference...) Figure 11 A kernel or filter refers to a unit / structure that applies weights to the input within a specific range / cell. Kernels (or filters) can be modified through learning. Stride is the range of movement the kernel makes within the input. A feature map is the result of applying the kernel to the input. Padding refers to values ​​added to adjust the size of the feature map. To enhance robustness to distortion and variations, multiple feature maps can be extracted. Pooling refers to the computation of reducing the size of a feature map by downsampling it (e.g., max pooling or average pooling).

[0169] Figure 12 An AI / ML model of an autoencoder is shown. (Refer to...) Figure 12 An autoencoder is a neural network that receives a feature vector x and outputs the same or similar vector x', and its input and output nodes share the same features—a form of unsupervised learning. The autoencoder reconstructs the input; therefore, the output can be called the reconstruction. The loss function can be expressed according to Equation 1.

[0170] [Formula 1] ,in

[0171] In Equation 12, the loss function of the autoencoder is calculated based on the difference between the input and output. Based on the loss function of the autoencoder, the degree of loss in the input is evaluated, and optimization processing is applied to the autoencoder to minimize the loss.

[0172] Figure 13 This is a diagram illustrating the splitting of AI reasoning.

[0173] Figure 13 This illustrates a scenario where model inference functions are collaboratively performed between a terminal device, such as a UE, and a network AI / ML endpoint during split AI operations.

[0174] In addition to model inference, model training, executors, and data collection can each be broken down into multiple parts based on the current task and environment. These functions can be executed through the collaboration of multiple entities.

[0175] For example, computationally intensive and energy-intensive parts can be executed at the network endpoint, while privacy-sensitive and latency-sensitive parts can be executed on the terminal device. In this case, the terminal device executes the task / model up to a specific part / layer based on the input data, and then sends intermediate data to the network endpoint. The network endpoint executes the remaining part / layer and provides inference output to one or more devices that performed the operation / task.

[0176] The following describes the functional framework used for AI operations.

[0177] In this article, the following terms may be defined for more detailed explanation of AI (or AI / ML).

[0178] - 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.

[0179] - AI Model: A data-driven algorithm in which AI technology is applied to generate an output set based on an input set, including predictive information and / or decision parameters.

[0180] - AI / ML Training: Online or offline processing of training AI models by learning features and patterns to best represent data and obtain AI / ML models for inference training.

[0181] - AI / ML inference: The process of using trained AI models to make predictions or make decisions based on collected data and the AI ​​model.

[0182] 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.

[0183] For example, input data may include measurements from the UE or other network entities, feedback from actuators, and outputs from AI models.

[0184] Data collection function 10 performs data preparation based on input data and provides input data processed through data preparation. Here, data collection function 10 does not perform specific data preparation (e.g., data preprocessing and cleaning, shaping and transformation) for each AI algorithm, but data collection function 10 can perform common data preparation for AI algorithms.

[0185] After completing the data preparation and processing, the data collection function 10 provides the training data 11 to the model training function 20 and 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, and the inference data 12 is the input data required by the AI ​​model inference function 30.

[0186] The data collection function 10 can be performed by a single entity (e.g., UE, RAN node, network node, etc.) or by multiple entities. In this case, training data 11 and inference data 12 can be provided to the model training function 20 and the model inference function 30 from multiple entities, respectively.

[0187] 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 may perform data preparation (e.g., data preprocessing and cleaning, shaping, and transformation) based on the training data 11 provided by the data collection function 10.

[0188] Here, model deployment / update 13 is used to initially deploy the trained, validated, and tested AI model to the model inference function 30 or to provide an updated model to the model inference function 30.

[0189] The model inference function 30 is responsible for providing AI model inference output 16 (e.g., prediction or decision). 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 perform data preparation (e.g., data preprocessing and cleaning, shaping and transformation) based on the inference data 12 provided by the data collection function 10.

[0190] 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 usage.

[0191] When available, model performance feedback 14 can be used to monitor the performance of the AI ​​model. However, the feedback can be omitted.

[0192] Actuator function 40 receives output 16 from model inference function 30 and triggers or executes related tasks / operations. Actuator function 40 may trigger tasks / operations of other entities (e.g., one or more UEs, one or more RAN nodes, one or more network nodes, etc.) or itself.

[0193] Feedback 15 can be used to derive training data 11 and inference data 12, or to monitor the performance of the AI ​​model and its impact on the network.

[0194] The definitions of training, validation, and testing in datasets used in AI / ML can be distinguished as follows: - Training data: Training data refers to the dataset used to train the model.

[0195] - Validation data: Validation data refers to the dataset used to validate a trained model. In other words, validation data is a dataset used to prevent overfitting from the typical training dataset.

[0196] Additionally, validation data refers to the dataset used to select the best model among multiple trained models during the training process. Therefore, validation data can also be considered a form of learning.

[0197] - Test data: Test data refers to the dataset used for the final evaluation. Test data is independent of training data.

[0198] In the case of the datasets described above, it is common practice to split the training set so that the training and validation data are divided in an 8:2 or 7:3 ratio. When test data is included, the training, validation, and test data can be divided in a 6:2:2 ratio (training:validation:test).

[0199] Based on the AI / ML capabilities between the BS and UE, the collaboration level can be defined as follows. Multiple levels can also be combined, or a level can be modified by separating any one of them.

[0200] Cat 0a) No collaborative framework: AI / ML algorithms are purely based on the implementation and do not require changes to the wireless interface.

[0201] Cat 0b) This level relates to wireless interfaces suitable for modifications based on efficient implementations of AI / ML algorithms, but this level corresponds to a framework without cooperation.

[0202] Cat 1) provides support between nodes to enhance the AI / ML algorithms of each node. This level is applied when the UE receives support from the gNB (for training, adaptation, etc.) and vice versa. At this level, model exchange between network nodes is not required.

[0203] Cat 2) enables collaborative ML tasks to be performed between the UE and gNB. This level requires AI / ML model command exchange between network nodes.

[0204] Figure 14 The functions shown can be implemented in RAN nodes (e.g., BS, TRP, BS central unit (CU)), network nodes, network operator operations management and maintenance (OAM) or UE.

[0205] Alternatively, two or more entities among the RAN node, network node, network operator's OAM, or UE can collaborate to achieve this. Figure 14 The functions shown. For example, an entity can perform... Figure 14 One entity can perform some of the functions, while 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., UE, RAN node, or network node), then the transmission / providing of data / information between the functions can be omitted. For example, if model training function 20 and model inference function 30 are performed by the same entity, then the transmission / providing of model deployment / update 13 and model performance feedback 14 can be omitted.

[0206] Alternatively, Figure 14Any of the functions shown can be performed through collaboration between two or more entities in the RAN node, network node, network operator's OAM, or UE. This can be referred to as split AI operation.

[0207] Figure 15 This illustrates a scenario where AI model training is performed by network nodes (e.g., core network nodes, network operator's OAM, etc.), while AI model inference is performed by RAN nodes (e.g., BS, TRP, or BS's CU).

[0208] Step 1: RAN Node 1 and RAN Node 2 send the 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 the UE (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.

[0209] Step 2: Network nodes use the received training data to train AI models.

[0210] Step 3: The network nodes deploy / update the AI ​​model to RAN node 1 and / or RAN node 2. RAN node 1 (and / or RAN node 2) can then continue to perform model training based on the received AI model.

[0211] For ease of explanation, assume that the AI ​​model is only deployed / updated to RAN node 1.

[0212] Step 4: RAN Node 1 receives input data (i.e., inference data) for AI model inference from UE and RAN Node 2.

[0213] Step 5: RAN node 1 uses the received inference data to perform AI model inference to generate output data (e.g., predictions or decisions).

[0214] Step 6: If applicable, RAN node 1 may send model performance feedback to the network nodes.

[0215] 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 load balancing operation, the UE may move from RAN Node 1 to RAN Node 2.

[0216] Step 8: RAN Node 1 and RAN Node 2 send feedback information to the network nodes.

[0217] Figure 16 This illustrates a scenario where both AI model training and AI model inference are performed by RAN nodes (e.g., BS, TRP, or CU of BS).

[0218] Step 1: The UE and RAN Node 2 send the input data (i.e., training data) for AI model training to RAN Node 1.

[0219] Step 2: RAN node 1 uses the received training data to train the AI ​​model.

[0220] Step 3: RAN Node 1 receives input data (i.e., inference data) for AI model inference from UE and RAN Node 2.

[0221] Step 4: RAN node 1 uses the received inference data to perform AI model inference to generate output data (e.g., predictions or decisions).

[0222] Step 5: 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 load balancing operation, the UE may move from RAN Node 1 to RAN Node 2.

[0223] Step 6: RAN node 2 sends feedback information to RAN node 1.

[0224] Figure 17 This illustrates a scenario where AI model training is performed by the RAN node (e.g., BS, TRP, or CU of BS), while AI model inference is performed by the UE.

[0225] Step 1: The UE sends input data (i.e., training data) to the RAN node for AI model training. 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.).

[0226] Step 2: The RAN node uses the received training data to train the AI ​​model.

[0227] Step 3: The RAN node deploys / updates the AI ​​model to the UE. The UE can then continue to perform model training based on the received AI model.

[0228] Step 4: The UE receives input data (i.e., inference data) for AI model inference from the RAN node (and / or from other UEs).

[0229] Step 5: The UE uses the received inference data to perform AI model inference to generate output data (e.g., predictions or decisions).

[0230] Step 6: If applicable, the UE may send model performance feedback to the RAN node.

[0231] Step 7: The UE and RAN node perform actions based on the output data.

[0232] Step 8: The UE sends feedback information to the RAN node.

[0233] AI / ML model alignment process

[0234] AI / ML is being discussed to perform User Equipment (UE) / Network (NW) operations more efficiently and reduce UE / NW complexity. In 3GPP Release 18, research is underway to apply AI / ML to CSI feedback, beam management, and positioning. AI / ML can be implemented in various ways depending on the application and / or the AI / ML functionality / model.

[0235] Regarding CSI compression using a bilateral model, it may be necessary to allow the UE to select information from a UE-side AI / ML model (e.g., a CSI generation model) that is compatible with the AI / ML model used by the BS / NW (e.g., a CSI reconstruction model). Additionally, a process may be needed to align information between the UE and the BS / NW.

[0236] Therefore, this disclosure proposes model alignment between the BS / NW and UE for bilateral model operation. Model alignment can refer to the operation / process of pairing the UE-side model and the NW-side model with each other, such that above-a certain level of performance (e.g., intermediate KPI (cosine similarity) and performance KPI (throughput)) is derived through AI / ML models on both the transmitting and receiving sides (e.g., UE-side model and NW-side model), and the AI / ML models on both sides are compatible.

[0237] Figure 19 An exemplary process for model alignment between two-sided models based on AI / ML is shown.

[0238] Reference Figure 19 The UE sends information about the UE-side model and / or auxiliary information to the BS / NW (step 1a).

[0239] The BS / NW can select an appropriate NW-side model based on information sent by the UE. The BS / NW can send information about the selected NW-side AI / ML model to the UE (step 2a).

[0240] For example, in step 2a, the UE can determine whether the NW-side AI / ML model is compatible with the UE-side model based on the information about the NW-side AI / ML model sent by the BS / NW. The UE can send an acknowledgment (ACK) message to the BS / NW (step 3a). The ACK can be used to notify the UE that it has successfully received the information about the NW-side AI / ML model sent by the BS / NW in step 2a and / or that the NW-side model information is appropriate / compatible with the UE's UE-side model.

[0241] For example, when the NW model ID is sent to the UE in step 2a, the UE can independently determine and apply the UE model ID that is paired with the corresponding model ID (by selecting one of the UE models). Alternatively, the BS / NW can directly signal to the UE the UE model ID that is paired with the NW model it has selected.

[0242] The information about the UE / NW side AI / ML model sent and received in steps 1a / 2b may include, but is not limited to, at least some of the following information fragments.

[0243] (i) Backbone network: Information about the backbone network of the AI / ML model, which includes information about convolutional neural networks (CNN), deep neural networks (DNN), recurrent neural networks (RNN), long short-term memory (LSTM), etc.

[0244] (ii) Model structure / parameters: Model structure / parameters may include how many layers / nodes the corresponding AI / ML model has or what cost function is used (e.g., MSE or cross-entropy loss function).

[0245] (iii) Model ID: For example, the UE model ID can be an ID assigned by the NW when the UE registers the model. Alternatively, the model ID can be an ID that is unique to the UE / NW. In addition, the UE / NW model ID can be a pairing ID for a specific compatible UE / NW model (through prior coordination).

[0246] (iv) Uncertainty / accuracy level of the reference model: The reference model refers to the model (e.g., the NW-side model) used for performance evaluation of the UE-side model (through prior coordination such as UE testing). This may include information on the level of uncertainty (e.g., variance and range of values) of the UE performance derived from the reference model (e.g., intermediate KPIs (cosine similarity) and performance KPIs (throughput)).

[0247] (V) Model Applicability Conditions and / or Metadata: Information about the UE's channel environment and the environment in which the AI / ML model is applied. For example, it may be information about the region / zone / scene / environment / dataset (e.g., identifiers), and may also include information such as the configuration of the specific model used for applying the BS / NW.

[0248] Figure 20 Another exemplary process for model alignment between two bilateral models based on AI / ML is shown. Figure 20 In the NW, the model alignment can be controlled.

[0249] The UE can send UE capability information about at least one model ID / function ID / application condition to the NW / BS (step 1b).

[0250] The NW / BS can configure at least one model ID / function ID and / or other auxiliary information for the UE (step 2b). The auxiliary information may include at least one of the following: time information (timestamp, window, or timer) or area / site / dataset / configuration information (e.g., identifier).

[0251] NW / BS can prevent the activation / triggering of AI / ML models that are not supported by the UE.

[0252] For specific times / regions not supported by the UE, the NW / BS can dynamically activate / deactivate specific AI / ML models (e.g., via MAC-CE or DCI).

[0253] The UE can use the UE-side model indicated by the NW / BS based on timing / area information (step 3b).

[0254] When multiple compatible models are configured for a UE, the UE can select a model and report the model ID.

[0255] exist Figure 19 / Figure 20 In this process, the UE can determine / check (e.g., model monitoring) the NW-side model that the BS / NW has identified as appropriate / compatible. Based on the UE's determination, additional procedures may be needed to restore the pairing between the UE-side model and the NW-side model. For this purpose, the following options can be considered as well.

[0256] (i) Option 1: The UE can fall back to a non-AI operation (e.g., traditional CSI feedback, beam management, or positioning) and then report the fallback to the BS / NW.

[0257] (ii) Option 2. The UE can switch from NW request mode.

[0258] Regarding model switching requests, the UE can report to the BS / NW that monitoring metrics, etc., are outside the predefined / permitted range (e.g., outside the range). The BS / NW can identify these models as incorrectly paired based on the UE's report.

[0259] Before the model switch is completed, the UE can continue to operate based on the model configured by the BS / NW, but then proceed to the recovery process related to the function / model failure.

[0260] Figure 21 The diagram illustrates AI / ML function / model failures and an exemplary recovery process for such failures.

[0261] Recovery operations for functional / model failures can be defined similarly to those for radio link failures (RLF) / beam failures. For example, a recovery process for functional / model failures could consider some or all of the following.

[0262] The UE monitors the model based on monitoring metrics (e.g., intermediate KPIs (cosine similarity) and performance KPIs (throughput)) (configured by the BS / NW or pre-agreed upon) and detects functional / model faults based on model monitoring (step 1c). For example, the UE can detect a fault when the metric is equal to or less than a specific threshold configured by the BS / NW.

[0263] When multiple active / inactive NW-side models are configured as candidates, the UE can select the model suitable for itself from the candidate models / functions and report to the BS / NW to switch to the selected model (step 2c).

[0264] When a function / model failure is detected multiple times (e.g., by the number of times / counter configured by the BS / NW), the UE can request the BS / NW to switch / update / deactivate the corresponding model (step 3c).

[0265] In response to a UE's request, the BS / NW can send a command to the UE to perform a specific model switch / update / fine-tuning / deactivation (step 4c).

[0266] Figure 22 An exemplary process for model alignment between two-sided models based on AI / ML is shown. Figure 22 In this context, the UE can have control mode alignment.

[0267] The UE can send UE capability information about at least one model ID / function ID / application condition to the NW / BS (step 1d).

[0268] The NW / BS can configure / indicate information to the UE about at least one NW-side model and / or UE-NW pair model (e.g., proxy model / reference model) (step 2d).

[0269] The UE can select and report the UE-side model, and also report the given NW-side model and / or the UE-NW's suitability for the model (step 3d).

[0270] When the UE determines that none of the models are suitable, the UE can report that these models are unsuitable and fall back to non-AI-based operations.

[0271] Since step 2d may require complex processes such as model transfer / transfer, the BS / NW can instruct / configure the first metric calculation value based on the model reported by the UE / for the UE in step 1d, and the UE can calculate the second metric based on the UE model implemented according to the information provided by the BS / NW in step 2d. The UE can perform model alignment operations between the BS / NW and the UE by operations such as comparing the first and second metric values.

[0272] As described above, when a BS / NW employing AI / ML technology or a high-performance BS / NW (or otherwise) is implemented, dynamic optimization for interference / services becomes possible through real-time collaboration and / or integrated control between BS / NWs. Control and AI / ML-related signaling methods for this purpose have been proposed in this disclosure. Although this disclosure has been described based on a 3GPP NR system, the proposed techniques can be applied to other communication systems.

[0273] Figure 23 An exemplary implementation example of the operation of a first device according to an embodiment is shown. The first device may be a UE.

[0274] Reference Figure 23 The first device can send a first signal (A05) to the second device, including information about the first AI / ML model configured in the first device.

[0275] In response to the first signal, the first device can receive a second signal (A10) from the second device, which includes information about a second AI / ML model configured in the second device.

[0276] The first AI / ML model configured in the first device and the second AI / ML model configured in the second device can be associated with each other.

[0277] The second signal may include at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about the conditions under which the second AI / ML model is applied.

[0278] Information regarding the conditions for applying the second AI / ML model may include at least one of a region identifier or a dataset identifier.

[0279] The second AI / ML model can be determined based on information about the first AI / ML model.

[0280] The second signal may include information about the performance of the first AI / ML model, which is evaluated based on the reference AI / ML model.

[0281] The second signal may include information about the level of uncertainty in the performance evaluation of the first AI / ML model based on the reference AI / ML model.

[0282] The first device can check the compatibility between the first AI / ML model and the second AI / ML model.

[0283] Based on compatibility being less than a threshold, the first device can request the second device to switch or update the second AI / ML model.

[0284] Based on compatibility being below a threshold, the first device can operate in a fallback mode that is not related to AI / ML.

[0285] The first device can send ACK information to the second device indicating that the first AI / ML model and the second AI / ML model are compatible.

[0286] Figure 24 An exemplary implementation example of the operation of a second device according to an embodiment is shown. The second device may be a BS or a network node.

[0287] Reference Figure 24 The second device can receive a first signal (B05) from the first device, including information about the first AI / ML model configured in the first device.

[0288] In response to the first signal, the second device may send a second signal (B10) to the first device, including information about the second AI / ML model configured in the second device.

[0289] The first AI / ML model configured in the first device and the second AI / ML model configured in the second device can be associated with each other.

[0290] The second signal may include at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about the conditions under which the second AI / ML model is applied.

[0291] Information regarding the conditions for applying the second AI / ML model may include at least one of a region identifier or a dataset identifier.

[0292] The second AI / ML model can be determined based on information about the first AI / ML model.

[0293] The second signal may include information about the performance of the first AI / ML model, which is evaluated based on a reference AI / ML model.

[0294] The second signal may include information about the level of uncertainty in the performance evaluation of the first AI / ML model based on the reference AI / ML model.

[0295] The second device can receive ACK information from the first device indicating that the first AI / ML model and the second AI / ML model are compatible.

[0296] Figure 25 A communication system 1 applied to this disclosure is shown.

[0297] Reference Figure 25 The communication system 1 applied to this disclosure includes wireless devices, base stations (BS), and networks. 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) robots 100a, vehicles 100b-1 and 100b-2, extended reality (XR) devices 100c, handheld devices 100d, home appliances 100e, Internet of Things (IoT) devices 100f, and artificial intelligence (AI) devices / servers 400. For example, vehicles may include vehicles with wireless communication capabilities, autonomous vehicles, and vehicles capable of performing communication between vehicles. Herein, vehicles may include unmanned aerial vehicles (UAVs) (e.g., drones). XR devices may include augmented reality (AR) / virtual reality (VR) / mixed reality (MR) devices, and may take the form of head-mounted displays (HMDs), head-up displays (HUDs) installed in vehicles, televisions, smartphones, computers, wearable devices, home appliances, digital signage, vehicles, robots, etc. Handheld devices may include smartphones, smart tablets, 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, the BS and network may be implemented as wireless devices, and a particular wireless device 200a may operate as a BS / network node relative to other wireless devices.

[0298] Wireless devices 100a to 100f can connect to network 300 via BS 200. AI technology can be applied to wireless devices 100a to 100f, and wireless devices 100a to 100f can connect to AI server 400 via network 300. Network 300 can be configured using a 3G network, a 4G (e.g., LTE) network, or a 5G (e.g., NR) network. Although wireless devices 100a to 100f can communicate with each other via BS 200 / network 300, wireless devices 100a to 100f can perform direct communication with each other (e.g., sidelink communication) without going through the BS / network. For example, vehicles 100b-1 and 100b-2 can perform direct communication (e.g., vehicle-to-vehicle (V2V) / vehicle-to-everything (V2X) communication). IoT devices (e.g., sensors) can perform direct communication with other IoT devices (e.g., sensors) or other wireless devices 100a to 100f.

[0299] Wireless communication / connections 150a, 150b, or 150c can be established between wireless devices 100a to 100f / BS 200 or between BS 200 and BS 200. In this document, wireless communication / connections can 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)). Wireless devices and BS / wireless devices can transmit / receive radio signals to / from each other via wireless communication / connections 150a and 150b. For example, wireless communication / connections 150a and 150b can transmit / receive signals via various physical channels. For this purpose, at least a portion of the configuration information for configuring the process of transmitting / receiving radio signals, various signal processing processes (e.g., channel coding / decoding, modulation / demodulation, and resource mapping / demapping), and resource allocation processes can be performed based on various proposals of this disclosure.

[0300] Figure 26 A wireless device applicable to this disclosure is shown.

[0301] Reference Figure 26 The first wireless device 100 and the second wireless device 200 can transmit radio signals via various RATs (e.g., LTE and NR). In this document, {first wireless device 100 and second wireless device 200} can correspond to... Figure 25 {Wireless Device 100x and BS 200} and / or {Wireless Device 100x and Wireless Device 100x}.

[0302] The first wireless device 100 may include one or more processors 102 and one or more memories 104, and additionally include one or more transceivers 106 and / or one or more antennas 108. The processors 102 may control the memories 104 and / or the transceivers 106, and may be configured to implement the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed herein. For example, the processor 102 may process information in the memory 104 to generate a first information / signal, and then transmit a radio signal including the first information / signal via the transceivers 106. The processor 102 may receive a radio signal including a second information / signal via the transceivers 106, and then store the information obtained by processing the second information / signal in the memory 104. The memory 104 may be connected to the processor 102 and may store various information relating to the operation of the processor 102. For example, the memory 104 may store software code including commands for performing some or all of the processes controlled by the processor 102 or for performing the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed herein. In this document, processor 102 and memory 104 may be part of a communication modem / circuit / chip designed to implement 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.

[0303] The second wireless device 200 may include one or more processors 202 and one or more memories 204, and additionally include one or more transceivers 206 and / or one or more antennas 208. The processors 202 may control the memories 204 and / or the transceivers 206, and may be configured to implement the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed in this document. For example, the processors 202 may process information in the memories 204 to generate a third information / signal, and then transmit a radio signal including the third information / signal via the transceivers 206. The processors 202 may receive a radio signal including a fourth information / signal via the transceivers 206, and then store the information obtained by processing the fourth information / signal in the memories 204. The memories 204 may be connected to the processors 202 and may store various information relating to the operation of the processors 202. For example, the memories 204 may store software code including commands for executing some or all of the processes controlled by the processors 202 or for executing the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed in this document. In this document, processor 202 and memory 204 may be part of a communication modem / circuit / chip designed to implement RAT (e.g., LTE or NR). Transceiver 206 may be connected to 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. Transceiver 206 may be used interchangeably with an RF unit. In this disclosure, a wireless device may refer to a communication modem / circuit / chip.

[0304] 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 are 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, processes, proposals, methods, and / or operation 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, processes, proposals, methods, and / or operation flowcharts disclosed in this document. One or more processors 102 and 202 may generate signals (e.g., baseband signals) including PDUs, SDUs, messages, control information, data, or information, according to the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed in this document, and provide the generated signals to one or more transceivers 106 and 206. One or more processors 102 and 202 may receive signals (e.g., baseband signals) and acquire PDUs, SDUs, messages, control information, data, or information from one or more transceivers 106 and 206, according to the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed in this document.

[0305] One or more processors 102 and 202 may be referred to as controllers, microcontrollers, microprocessors, or microcomputers. One or more processors 102 and 202 may be implemented by hardware, firmware, software, or a combination thereof. As an example, one or more application-specific integrated circuits (ASICs), one or more digital signal processors (DSPs), one or more digital signal processing devices (DSPDs), one or more programmable logic devices (PLDs), or one or more field-programmable gate arrays (FPGAs) may be included in one or more processors 102 and 202. The descriptions, functions, processes, proposals, methods, and / or operation 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. Firmware or software configured to execute the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed in this document may be included in one or more processors 102 and 202 or stored in one or more memories 104 and 204 to be driven by one or more processors 102 and 202. The descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed in this document can be implemented using firmware or software in the form of code, commands, and / or command sets.

[0306] 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, code, instructions, and / or commands. One or more memories 104 and 204 may be configured with read-only memory (ROM), random access memory (RAM), electrically erasable programmable read-only memory (EPROM), flash memory, hard disk drive, registers, cache memory, computer-readable storage media, and / or combinations 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.

[0307] One or more transceivers 106 and 206 may transmit user data, control information, and / or radio signals / channels mentioned in the methods and / or operation 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 operation 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 perform controls to enable one or more transceivers 106 and 206 to transmit user data, control information, or radio signals to one or more other devices. One or more processors 102 and 202 may perform controls to enable one or more transceivers 106 and 206 to 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 one or more transceivers 106 and 206 may be configured to transmit and receive user data, control information, and / or radio signals / channels mentioned in the descriptions, functions, processes, proposals, methods, and / or operation flowcharts disclosed herein via 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). One or more transceivers 106 and 206 may convert received radio signals / channels, etc., from RF band signals to baseband signals for processing by one or more processors 102 and 202. One or more transceivers 106 and 206 may convert user data, control information, radio signals / channels, etc., processed by one or more processors 102 and 202 from baseband signals to RF band signals. For this purpose, one or more transceivers 106 and 206 may include (analog) oscillators and / or filters.

[0308] Figure 27 Another example of a wireless device applied to this disclosure is shown. The wireless device may vary depending on usage / service (see reference). Figure 25 It is realized in various forms.

[0309] Reference Figure 27 Wireless devices 100 and 200 can correspond to Figure 26The wireless devices 100 and 200 are configured with 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 26 One or more processors 102 and 202 and / or one or more memories 104 and 204. For example, transceiver 114 may include Figure 26 The device comprises one or more transceivers 106 and 206 and / or one or more antennas 108 and 208. Control unit 120 is electrically connected to communication unit 110, memory unit 130, and add-on components 140, and controls the overall operation of the wireless device. For example, control unit 120 may control the electrical / mechanical operation of the wireless device based on programs / code / commands / information stored in memory unit 130. Control unit 120 may transmit information stored in memory unit 130 to an external source (e.g., other communication devices) via communication unit 110 through a wireless / wired interface, or store information received from an external source (e.g., other communication devices) via communication unit 110 in memory unit 130 via a wireless / wired interface.

[0310] The additional component 140 can be configured differently depending on the type of wireless device. For example, the additional component 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 can be configured according to (but is not limited to) a robot ( Figure 25 100a), vehicles ( Figure 25 100b-1 and 100b-2), XR device ( Figure 25 100c), handheld device ( Figure 25 100d), home appliances ( Figure 25 100e), IoT devices ( Figure 25 100f), digital broadcasting terminals, holographic devices, public safety devices, MTC devices, medical devices, fintech devices (or financial devices), security devices, climate / environment devices, AI servers / devices ( Figure 25 400), BS ( Figure 25 This can be achieved through 200 network nodes, etc. Wireless devices can be used in mobile or fixed locations depending on the use case / service.

[0311] exist Figure 27In wireless devices 100 and 200, all elements, components, units / parts, and / or modules may be interconnected 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 connected, 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 / parts, and / or modules within wireless devices 100 and 200 may also include one or more elements. For example, control unit 120 may be configured as a collection of one or more processors. As an example, control unit 120 may be configured as a collection of communication control processors, application processors, electronic control units (ECUs), graphics processing units, and memory control processors. As another example, memory unit 130 may be configured as random access memory (RAM), dynamic RAM (DRAM), read-only memory (ROM), flash memory, volatile memory, non-volatile memory, and / or combinations thereof.

[0312] Figure 28 The illustration shows a vehicle or autonomous vehicle applicable to this disclosure. The vehicle or autonomous vehicle may be a mobile robot, car, train, manned / unmanned aerial vehicle (AV), vessel, etc.

[0313] Reference Figure 28 The vehicle or autonomous 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 part of the communication unit 110. Blocks 110 / 130 / 140a to 140d respectively correspond to... Figure 27 Blocks 110 / 130 / 140.

[0314] Communication unit 110 can send and receive signals (e.g., data and control signals) to and from external devices such as other vehicles, BSs (e.g., gNBs and roadside units), and servers. Control unit 120 can perform various operations by controlling the components of the vehicle or autonomous vehicle 100. Control unit 120 may include an electronic control unit (ECU). Drive unit 140a enables the vehicle or autonomous vehicle 100 to move on a road. Drive unit 140a may include an engine, motor, powertrain, wheels, brakes, steering mechanism, etc. Power supply unit 140b can supply power to the vehicle or autonomous vehicle 100 and includes wired / wireless charging circuitry, battery, etc. Sensor unit 140c can acquire vehicle status, surrounding environment information, user information, etc. Sensor unit 140c may include inertial measurement unit (IMU) sensors, collision sensors, wheel sensors, speed sensors, slope sensors, weight sensors, heading sensors, position modules, vehicle forward / reverse sensors, battery sensors, fuel sensors, tire sensors, steering sensors, temperature sensors, depth sensors, ultrasonic sensors, lighting sensors, pedal position sensors, etc. Autonomous driving unit 140d can implement technologies for maintaining the vehicle within its lane, technologies for automatically adjusting speed (e.g., adaptive cruise control), technologies for autonomously driving along a determined path, and technologies for automatically setting a route if a destination is set, etc.

[0315] For example, communication unit 110 can receive map data, traffic information data, etc., from an external server. Autonomous driving unit 140d can generate an autonomous driving path and driving plan from the acquired data. Control unit 120 can control drive unit 140a, enabling the vehicle or autonomous vehicle 100 to move along the autonomous driving path according to the driving plan (e.g., speed / direction control). During autonomous driving, communication unit 110 can periodically or non-periodically acquire recent traffic information data from an external server and acquire surrounding traffic information data from neighboring vehicles. During autonomous driving, sensor unit 140c can acquire vehicle status and / or surrounding environment information. Autonomous driving unit 140d can update the autonomous driving path and driving plan based on newly acquired data / information. Communication unit 110 can transmit information about vehicle location, autonomous driving path, and / or driving plan to an external server. The external server can predict traffic information data using AI technology, etc., based on information collected from the vehicle or autonomous vehicle, and provide the predicted traffic information data to the vehicle or autonomous vehicle.

[0316] The above embodiments correspond to combinations of the elements and features of this disclosure in a prescribed form. Furthermore, unless explicitly stated otherwise, each element or feature may be considered optional. Each element or feature may be implemented without combination with other elements or features. Moreover, embodiments of this disclosure can be implemented by partially combining elements and / or features together. The order of operations described for various embodiments of this disclosure may be modified. Some configurations or features of one embodiment may be included in another embodiment, or may replace corresponding configurations or features of another embodiment. Additionally, it will be readily understood that embodiments are configured by combining claims not explicitly referenced in the appended claims, or may be included as new claims after filing the application.

[0317] Those skilled in the art will understand that this disclosure may be practiced in other specific forms besides those set forth herein without departing from the spirit and essential characteristics of this disclosure. Therefore, the above embodiments should be construed in all respects as illustrative rather than restrictive. The scope of this disclosure should be determined by the appended claims and their legal equivalents, and not by the foregoing description, and all changes falling within the meaning and scope of the appended claims are intended to be covered therewith.

[0318] Industrial applicability

[0319] This disclosure applies to UE, BS or other devices in wireless mobile communication systems.

Claims

1. A method performed by a first device in a wireless communication system, the method comprising the following steps: Sending a first signal to the second device, including information about a first artificial intelligence / machine learning (AI / ML) model configured in the first device; and In response to the first signal, a second signal including information about a second AI / ML model configured in the second device is received from the second device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device are associated with each other, and The second signal includes at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about the conditions for applying the second AI / ML model.

2. The method according to claim 1, wherein, The information regarding the conditions for applying the second AI / ML model includes at least one of a region identifier or a dataset identifier.

3. The method according to claim 1, wherein, The second AI / ML model is determined based on the information about the first AI / ML model.

4. The method according to claim 1, wherein, The second signal includes information about the performance of the first AI / ML model as evaluated based on the reference AI / ML model.

5. The method according to claim 1, wherein, The second signal includes information about the level of uncertainty in the performance evaluation of the first AI / ML model based on the reference AI / ML model.

6. The method according to claim 1, further comprising: Check the compatibility between the first AI / ML model and the second AI / ML model.

7. The method according to claim 6, further comprising: Based on the compatibility being less than the threshold, a request is made from the second device to switch or update the second AI / ML model.

8. The method according to claim 6, wherein, Based on the compatibility being less than the threshold, the first device operates in a fallback mode that is not related to AI / ML.

9. The method according to claim 1, further comprising: Send an ACK message to the second device indicating that the first AI / ML model and the second AI / ML model are compatible.

10. The method according to claim 1, wherein, The first device is a user equipment (UE), and The second device is a base station (BS) or a network node.

11. A computer-readable recording medium having a program for performing the method according to claim 1.

12. A first device in a wireless communication system, the first device comprising: A memory configured to store instructions; as well as A processor configured to perform operations by executing the instructions. The operations performed by the processor include: Sending a first signal to the second device, including information about a first artificial intelligence / machine learning (AI / ML) model configured in the first device; and In response to the first signal, a second signal including information about a second AI / ML model configured in the second device is received from the second device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device are associated with each other, and The second signal includes at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about the conditions for applying the second AI / ML model.

13. The first device according to claim 12, wherein, The first device is a user equipment (UE), and The second device is a network node or a base station (BS).

14. A method performed by a second device in a wireless communication system, the method comprising the following steps: Receive a first signal from the first device including information about a first artificial intelligence / machine learning (AI / ML) model configured in the first device; and In response to the first signal, a second signal including information about the second AI / ML model configured in the second device is sent to the first device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device are associated with each other, and The second signal includes at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about the conditions for applying the second AI / ML model.

15. A second device in a wireless communication system, the second device comprising: A memory configured to store instructions; as well as A processor configured to perform operations by executing the instructions. The operations performed by the processor include: Receive a first signal from the first device including information about a first artificial intelligence / machine learning (AI / ML) model configured in the first device; and In response to the first signal, a second signal including information about the second AI / ML model configured in the second device is sent to the first device. The first AI / ML model configured in the first device and the second AI / ML model configured in the second device are associated with each other, and The second signal includes at least one of information about a reference AI / ML model used by the second device for performance evaluation of the first AI / ML model or information about the conditions for applying the second AI / ML model.

Images