Method and apparatus for changing performance monitoring period of artificial intelligence and / or machine learning model

Abstract

Provided are a method and an apparatus for changing a performance monitoring period of an artificial intelligence (AI) / machine learning (ML) model. A terminal receives configuration information for performance monitoring of the AI / ML model from a base station. Here, the configuration information comprises information on at least one monitoring period and a period change condition associated with a reference signal (RS). In addition, the terminal performs performance monitoring of the AI / ML model on the basis of a first monitoring period among the at least one monitoring period. Thereafter, the terminal performs performance monitoring of the AI / ML model on the basis of a second monitoring period among the at least one monitoring period according to satisfaction of the period change condition.

Description

Method and device for changing the performance monitoring cycle of an artificial intelligence and / or machine learning model

[0001] This specification relates to wireless communications applicable to 5G NR, 5G-Advanced and 6G.

[0002] As more and more communication devices demand ever-increasing communication traffic, the need for next-generation 5G systems, which offer enhanced wireless broadband communication capabilities over existing LTE systems, is growing. This next-generation 5G system, known as NewRAT, differentiates communication scenarios into Enhanced Mobile BroadBand (eMBB), Ultra-reliability and low-latency communication (URLLC), and Massive Machine-Type Communications (mMTC).

[0003] Here, eMBB is a next-generation mobile communication scenario with characteristics such as High Spectrum Efficiency, High User Experienced Data Rate, and High Peak Data Rate; URLLC is a next-generation mobile communication scenario with characteristics such as Ultra Reliable, Ultra Low Latency, and Ultra High Availability (e.g., V2X, Emergency Service, and Remote Control); and mMTC is a next-generation mobile communication scenario with characteristics such as Low Cost, Low Energy, Short Packet, and Massive Connectivity (e.g., IoT).

[0004] The present disclosure provides a method and device for measuring the accuracy of AI / ML deployed for a terminal capable of performing beam management using AI / ML in a wireless communication system, wherein at least one model monitoring cycle and cycle change condition are set by a base station, and adaptively changing the model monitoring cycle when the set condition is satisfied.

[0005] One embodiment of the present specification provides a method for, in a wireless communication system, a terminal receives configuration information for performance monitoring of an artificial intelligence (AI) / machine learning (ML) model from a base station. The configuration information includes information on at least one monitoring period and a period change condition associated with a reference signal (RS). The terminal performs performance monitoring of the AI / ML model based on a first monitoring period among the at least one monitoring period. Thereafter, the terminal performs performance monitoring of the AI / ML model based on a second monitoring period among the at least one monitoring period, depending on satisfaction of the period change condition.

[0006] In addition, one embodiment of the present specification provides a method in which, in a wireless communication system, a base station transmits configuration information for performance monitoring of an AI / ML model to a terminal. Here, the configuration information includes information on at least one monitoring period and a period change condition associated with a reference signal (RS). Furthermore, the base station transmits the reference signal based on a first monitoring period among the at least one monitoring period. Subsequently, the base station transmits the reference signal based on a second monitoring period among the at least one monitoring period, depending on satisfaction of the period change condition.

[0007] In addition, one embodiment of the present invention provides a wireless communication system comprising at least one processor, and at least one memory storing instructions and being operably electrically connectable to the at least one processor, wherein the operation performed based on the instructions being executed by the at least one processor is: receiving configuration information for performance monitoring of an AI / ML model from a base station. Here, the configuration information includes information on at least one monitoring period and a period change condition associated with a reference signal (RS). Then, performance monitoring of the AI / ML model is performed based on a first monitoring period among the at least one monitoring period. Thereafter, a terminal is provided that performs performance monitoring of the AI / ML model based on a second monitoring period among the at least one monitoring period, depending on satisfaction of the period change condition.

[0008] In addition, one embodiment of the present invention provides a wireless communication system comprising at least one processor, and at least one memory storing instructions and being operably electrically connectable to the at least one processor, wherein the operation performed based on the instructions being executed by the at least one processor is: transmitting configuration information for performance monitoring of an AI / ML model to a terminal. Here, the configuration information includes information on at least one monitoring period and a period change condition associated with a reference signal (RS). Then, the reference signal is transmitted based on a first monitoring period among the at least one monitoring period. Thereafter, a base station is provided that transmits the reference signal based on a second monitoring period among the at least one monitoring period, depending on satisfaction of the period change condition.

[0009] The above periodic change conditions may include aperiodic reporting conditions of performance monitoring results of the AI / ML model.

[0010] Meanwhile, the above-described configuration information further includes timer information, and a timer can be started based on the timer information. While the timer is running, if an aperiodic reporting condition for the performance monitoring result of the AI / ML model is satisfied, the terminal can transmit the performance monitoring result of the AI / ML model to the base station, and the base station can receive it. The terminal can restart the timer upon transmission of the performance monitoring result of the AI / ML model, and the base station can restart the timer upon reception of the performance monitoring result of the AI / ML model. Here, the second monitoring period can be shorter than the first monitoring period.

[0011] On the other hand, the above-mentioned setting information further includes timer information, and when a timer started based on the timer information expires, transmission of a reference signal and performance monitoring of the AI / ML model based on the second monitoring period can be performed. Here, the second monitoring period may be longer than the first monitoring period.

[0012] On the other hand, after the COUNT value associated with the performance monitoring of the AI / ML model increases, and the aperiodic reporting condition of the performance monitoring result of the AI / ML model is satisfied, the terminal may transmit the performance monitoring result of the AI / ML model to the base station, and the base station may receive it. Upon transmission / reception of the performance monitoring result of the AI / ML model, the COUNT value may be reset. Here, the second monitoring period may be shorter than the first monitoring period.

[0013] On the other hand, the above setting information further includes information on a COUNT MAX value associated with performance monitoring of the AI / ML model, and when the increased COUNT value associated with performance monitoring of the AI / ML model reaches the COUNT MAX value, transmission of a reference signal based on the second monitoring period and performance monitoring of the AI / ML model can be performed. Here, the second monitoring period can be longer than the first monitoring period.

[0014] Communication techniques utilizing AI / ML models require continuous model monitoring. However, excessively frequent model monitoring can increase DL (downlink) reference signal (RS) overhead and increase the measurement burden on the terminal, which can lead to battery drain. To address this, the disclosure of this specification efficiently performs model monitoring using one or more model monitoring cycles, thereby effectively resolving the trade-off between the DL RS overhead and measurement burden associated with model monitoring and model performance monitoring.

[0015] Figure 1 is a diagram illustrating a wireless communication system.

[0016] Figure 2 illustrates the structure of a radio frame used in NR.

[0017] Figures 3a to 3c are exemplary diagrams showing exemplary architectures for wireless communication services.

[0018] Figure 4 illustrates the slot structure of an NR frame.

[0019] Figure 5 illustrates an example of subframe types in NR.

[0020] Figure 6 illustrates the structure of a self-contained slot.

[0021] Figure 7 shows an example of initial beam measurement and selection in NR.

[0022] Figure 8 shows an example of an initial connection procedure between a terminal and a base station in NR.

[0023] Figure 9 shows an example of candidate beam settings in NR.

[0024] Figures 10a to 10c illustrate three procedures for beam management in NR.

[0025] Figures 11a to 11c illustrate examples of beam reporting procedures in NR.

[0026] Fig. 12 is a flowchart illustrating a method of operating a terminal according to one embodiment of the present specification.

[0027] FIG. 13 is an example showing multi-level model monitoring according to one embodiment of the present specification.

[0028] FIG. 14 is an example showing a timer-based model monitoring cycle change according to one embodiment of the present specification.

[0029] FIG. 15 is an example showing a timer-based model monitoring cycle change according to another embodiment of the present specification.

[0030] FIG. 16 is an example showing a COUNT-based model monitoring cycle change according to one embodiment of the present specification.

[0031] Fig. 17 is a flowchart illustrating a method of operating a terminal according to another embodiment of the present specification.

[0032] Fig. 18 is a flowchart illustrating an operation method of a base station according to one embodiment of the present specification.

[0033] Figure 19 illustrates a device according to one embodiment of the present specification.

[0034] Figure 20 is a block diagram showing the configuration of a terminal according to one embodiment of the present specification.

[0035] Figure 21 shows a block diagram of a processor in which the disclosure of this specification is implemented.

[0036] Fig. 22 is a block diagram showing in detail the transmitter / receiver of the first device illustrated in Fig. 19 or the transmitter / receiver unit of the device illustrated in Fig. 20.

[0037] It should be noted that the technical terms used in this specification are used merely to describe specific embodiments and are not intended to limit the contents of this specification. In addition, unless specifically defined otherwise herein, the technical terms used in this specification should be interpreted as having a meaning generally understood by those skilled in the art to which this specification pertains, and should not be interpreted in an excessively broad or narrow sense. In addition, if a technical term used in this specification is an incorrect technical term that does not accurately express the contents and ideas of this specification, it should be replaced with a technical term that can be correctly understood by a person skilled in the art. In addition, general terms used in this specification should be interpreted according to their dictionary definitions or according to the preceding and following context, and should not be interpreted in an excessively narrow sense.

[0038] Additionally, the singular expressions used herein include plural expressions unless the context clearly dictates otherwise. In this application, terms such as "consist of" or "have" should not be construed to necessarily include all of the components or steps described in the specification, and should be construed to mean that some of the components or steps may not be included, or that additional components or steps may be included.

[0039] Additionally, terms including ordinal numbers, such as "first" and "second," used herein may be used to describe various components, but these components should not be limited by these terms. These terms are used solely to distinguish one component from another. For example, without departing from the scope of the present invention, a first component could be referred to as a second component, and similarly, a second component could also be referred to as a first component.

[0040] When a component is referred to as being connected or connected to another component, it may be directly connected or connected to that other component, but there may also be other components intervening. Conversely, when a component is referred to as being directly connected or connected to another component, it should be understood that there are no other components intervening.

[0041] Hereinafter, embodiments will be described in detail with reference to the attached drawings. Regardless of the drawing reference numerals, identical or similar components will be given the same reference numerals, and redundant descriptions thereof will be omitted. In addition, when describing the contents of this specification, if a detailed description of a related known technology is judged to obscure the gist of this specification, the detailed description thereof will be omitted. In addition, it should be noted that the attached drawings are only intended to make the contents and ideas of this specification easily understandable, and should not be construed as limiting the contents and ideas of this specification by the attached drawings. The contents and ideas of this specification should be construed to extend to all changes, equivalents, and substitutes other than the attached drawings.

[0042] In this specification, “A or B” can mean “only A,” “only B,” or “both A and B.” In other words, “A or B” in this specification can be interpreted as “A and / or B.” For example, “A, B or C” in this specification can mean “only A,” “only B,” “only C,” or “any combination of A, B, and C.”

[0043] As used herein, a slash ( / ) or a comma can mean "and / or." For example, "A / B" can mean "A and / or B." Accordingly, "A / B" can mean "only A," "only B," or "both A and B." For example, "A, B, C" can mean "A, B, or C."

[0044] In this specification, “at least one of A and B” may mean “only A,” “only B,” or “both A and B.” Additionally, in this specification, the expressions “at least one of A or B” or “at least one of A and / or B” may be interpreted identically to “at least one of A and B.”

[0045] Additionally, in this specification, “at least one of A, B and C” can mean “only A,” “only B,” “only C,” or “any combination of A, B and C.” Additionally, “at least one of A, B or C” or “at least one of A, B and / or C” can mean “at least one of A, B and C.”

[0046] Additionally, parentheses used in this specification may mean “for example.” Specifically, when “control information (PDCCH)” is indicated, “PDCCH (Physical Downlink Control Channel)” may be suggested as an example of “control information.” In other words, “control information” in this specification is not limited to “PDCCH,” and “PDDCH” may be suggested as an example of “control information.” Furthermore, even when indicated as “control information (i.e., PDCCH),” “PDCCH” may be suggested as an example of “control information.”

[0047] Technical features individually described in a single drawing in this specification may be implemented individually or simultaneously.

[0048] Although the attached drawing illustrates a UE (User Equipment) as an example, the illustrated UE may also be referred to as a terminal, ME (Mobile Equipment), etc. In addition, the UE may be a portable device such as a laptop, mobile phone, PDA, smart phone, multimedia device, etc., or a non-portable device such as a PC or vehicle-mounted device.

[0049] Hereinafter, the term "UE" is used as an example of a device capable of wireless communication (e.g., a wireless communication device, a wireless device, or a wireless device). The operations performed by the UE can be performed by any device capable of wireless communication. A device capable of wireless communication may also be referred to as a wireless communication device, a wireless device, or a wireless device.

[0050] The term base station used below generally refers to a fixed station that communicates with wireless devices, and can be used as a comprehensive term that includes eNodeB (evolved-NodeB), eNB (evolved-NodeB), BTS (Base Transceiver System), Access Point, gNB (Next generation NodeB), RRH (remote radio head), TP (transmission point), RP (reception point), relay, etc.

[0051] Although this specification describes embodiments using LTE systems, LTE-A systems, and NR systems, these embodiments may be applied to any communication system falling within the above definitions.

[0052] Wireless Communication System

[0053] Building on the success of LTE (long term evolution) / LTE-Advanced (LTE-A) for 4th generation mobile communications, commercialization of the next generation, or 5th generation (so-called 5G) mobile communications, and follow-up research are also ongoing.

[0054] The International Telecommunication Union (ITU) defines 5G mobile communications as providing data transfer speeds of up to 20 Gbps and a perceived transmission speed of at least 100 Mbps everywhere. Its official name is "IMT-2020."

[0055] ITU proposes three usage scenarios: eMBB (enhanced Mobile BroadBand), mMTC (massive Machine Type Communication), and URLLC (Ultra Reliable and Low Latency Communications).

[0056] URLLC addresses usage scenarios that require high reliability and low latency. For example, services such as autonomous driving, factory automation, and augmented reality require high reliability and low latency (e.g., sub-1ms). Current 4G (LTE) latency is statistically 21-43ms (best 10%) and 33-75ms (median). This is insufficient to support services requiring sub-1ms latency. Next, eMBB usage scenarios address usage scenarios that require mobile ultra-wideband.

[0057] In other words, the 5th generation mobile communication system can support higher capacity than the current 4G LTE, increase the density of mobile broadband users, and support D2D (Device to Device), high reliability, and MTC (Machine-type communication). 5G research and development also aims for lower latency and lower battery consumption than 4G mobile communication systems to better implement the Internet of Things. For this 5G mobile communication, a new radio access technology (New RAT or NR) may be proposed.

[0058] The NR frequency band can be defined by two types of frequency ranges (FR1, FR2). The numerical values ​​of the frequency ranges can be changed, and for example, the two types of frequency ranges (FR1, FR2) can be as shown in Table 1 below. For convenience of explanation, among the frequency ranges used in the NR system, FR1 can mean the “sub 6 GHz range”, and FR2 can mean the “above 6 GHz range” and can be called millimeter wave (mmW).

[0059] Frequency Range designationCorresponding frequency rangeSubcarrier SpacingFR1410MHz - 7125MHz15, 30, 60kHzFR224250MHz - 52600MHz60, 120, 240kHz

[0060] The numerical value of the frequency range of the NR system can be changed. For example, FR1 can include a band from 410 MHz to 7125 MHz, as shown in Table 1. That is, FR1 can include frequency bands above 6 GHz (or 5850, 5900, 5925 MHz, etc.). For example, the frequency bands above 6 GHz (or 5850, 5900, 5925 MHz, etc.) included within FR1 can include unlicensed bands. Unlicensed bands can be used for various purposes, such as for vehicle communications (e.g., autonomous driving).

[0061] Meanwhile, 3GPP-based communication standards define downlink physical channels corresponding to resource elements that carry information originating from upper layers, and downlink physical signals corresponding to resource elements that are used by the physical layer but do not carry information originating from upper layers. For example, the physical downlink shared channel (PDSCH), physical broadcast channel (PBCH), physical multicast channel (PMCH), physical control format indicator channel (PCFICH), physical downlink control channel (PDCCH), and physical hybrid ARQ indicator channel (PHICH) are defined as downlink physical channels, and reference signals and synchronization signals are defined as downlink physical signals. A reference signal (RS), also referred to as a pilot, is a signal with a special predefined waveform known to the gNB and the UE. For example, cell specific RS, UE-specific RS (UE-RS), positioning RS (PRS), and channel state information RS (CSI-RS) are defined as downlink reference signals. The 3GPP LTE / LTE-A standard defines uplink physical channels corresponding to resource elements carrying information originating from higher layers, and uplink physical signals corresponding to resource elements used by the physical layer but not carrying information originating from higher layers.For example, a physical uplink shared channel (PUSCH), a physical uplink control channel (PUCCH), and a physical random access channel (PRACH) are defined as uplink physical channels, and a demodulation reference signal (DMRS) for uplink control / data signals and a sounding reference signal (SRS) used for uplink channel measurement are defined.

[0062] In this specification, PDCCH (Physical Downlink Control CHannel) / PCFICH (Physical Control Format Indicator CHannel) / PHICH ((Physical Hybrid automatic retransmit request Indicator CHannel) / PDSCH (Physical Downlink Shared CHannel) mean a set of time-frequency resources or a set of resource elements that carry DCI (Downlink Control Information) / CFI (Control Format Indicator) / downlink ACK / NACK (ACKnowlegement / Negative ACK) / downlink data, respectively. In addition, PUCCH (Physical Uplink Control CHannel) / PUSCH (Physical Uplink Shared CHannel) / PRACH (Physical Random Access CHannel) mean a set of time-frequency resources or a set of resource elements that carry UCI (Uplink Control Information) / uplink data / random access signals, respectively.

[0063] Figure 1 is a diagram illustrating a wireless communication system.

[0064] As can be seen from FIG. 1, the wireless communication system includes at least one base station (BS). The BS is divided into a gNodeB (or gNB) (20a) and an eNodeB (or eNB) (20b). The gNB (20a) supports 5th generation mobile communications. The eNB (20b) supports 4th generation mobile communications, i.e., long term evolution (LTE).

[0065] Each base station (20a and 20b) provides communication services for a specific geographic area (commonly referred to as a cell) (20-1, 20-2, 20-3). The cell may be further divided into multiple areas (referred to as sectors).

[0066] A UE (user equipment) typically belongs to a single cell, and the cell to which the UE belongs is called a serving cell. The base station that provides communication services for the serving cell is called a serving base station (BS). Since the wireless communication system is a cellular system, there are other cells adjacent to the serving cell. These other cells adjacent to the serving cell are called neighbor cells. The base station that provides communication services to the neighbor cell is called a neighbor BS. The serving cell and neighbor cells are determined relative to the UE.

[0067] Hereinafter, downlink refers to communication from a base station (20) to a UE (10), and uplink refers to communication from a UE (10) to a base station (20). In downlink, the transmitter may be part of the base station (20), and the receiver may be part of the UE (10). In uplink, the transmitter may be part of the UE (10), and the receiver may be part of the base station (20).

[0068] Meanwhile, wireless communication systems can be broadly divided into frequency division duplex (FDD) and time division duplex (TDD). In FDD, uplink and downlink transmissions occupy different frequency bands and occur at different times. In TDD, uplink and downlink transmissions occupy the same frequency band but occur at different times. The channel response in TDD is essentially reciprocal, meaning that the downlink and uplink channel responses are nearly identical in a given frequency range. Therefore, in TDD-based wireless communication systems, the downlink channel response can be derived from the uplink channel response. In TDD, uplink and downlink transmissions are time-divided across the entire frequency band, so downlink transmission by the base station and uplink transmission by the UE cannot be performed simultaneously. In TDD systems, where uplink and downlink transmissions are divided into subframes, uplink and downlink transmissions are performed in different subframes.

[0069] Figure 2 illustrates the structure of a radio frame used in NR.

[0070] In NR, uplink and downlink transmissions are structured as frames. A radio frame is 10ms long and is defined by two 5ms half-frames (HF). Each half-frame is defined by five 1ms subframes (SF). A subframe is divided into one or more slots, and the number of slots in a subframe depends on the subcarrier spacing (SCS). Each slot contains 12 or 14 OFDM(A) symbols, depending on the cyclic prefix (CP). When a normal CP is used, each slot contains 14 symbols. When an extended CP is used, each slot contains 12 symbols. Here, the symbols may include OFDM symbols (or CP-OFDM symbols), SC-FDMA symbols (or DFT-s-OFDM symbols).

[0071] Support for various numerologies

[0072] In NR systems, multiple numerologies may be provided to terminals as wireless communication technologies advance. For example, an SCS of 15 kHz supports a wide area in traditional cellular bands. An SCS of 30 kHz / 60 kHz supports dense urban environments, lower latency, and wider carrier bandwidth. An SCS of 60 kHz or higher supports a bandwidth greater than 24.25 GHz to overcome phase noise.

[0073] The above numerology can be defined by the cycle prefix (CP) length and subcarrier spacing (SCS). A single cell can provide multiple numerologies to a terminal. When the numerology index is represented by μ, each subcarrier spacing and the corresponding CP length can be as shown in the table below.

[0074] μ△f=2 μ 15 [kHz]CP015 General 130 General 260 General, Extended 3120 General 4240 General 5480 General 6960 General

[0075] For general CP, when the index of the numerology is represented by μ, the number of OFDM symbols per slot (N slot symb ), number of slots per frame (N frame,μ slot ) and the number of slots per subframe (N subframe,μ slot ) is as shown in the table below.

[0076] μ△f=2 μ 15 [kHz]N slot symb N frame,μ slot N subframe,μ slot 015141011301420226014404312014808424014160165480143203269601464064

[0077] For extended CP, when the index of the numerology is represented by μ, the number of OFDM symbols per slot (N slot symb ), number of slots per frame (N frame,μ slot ) and the number of slots per subframe (N subframe,μ slot ) is as shown in the table below.

[0078] μSCS (15*2 u )N slot symb N frame,μ slot N subframe,μslot 260KHz (u=2)12404

[0079] In an NR system, OFDM(A) numerology (e.g., SCS, CP length, etc.) may be set differently between multiple cells that are merged into a single terminal. Accordingly, the (absolute time) interval of a time resource (e.g., SF, slot, or TTI) (conveniently referred to as TU (Time Unit)) consisting of the same number of symbols may be set differently between the merged cells.

[0080] Figures 3a to 3c are exemplary diagrams showing exemplary architectures for wireless communication services.

[0081] Referring to FIG. 3a, the UE is connected to an LTE / LTE-A-based cell and an NR-based cell in a DC (dual connectivity) manner.

[0082] The above NR-based cell is connected to the core network for existing 4th generation mobile communication, i.e. Evolved Packet Core (EPC).

[0083] Referring to FIG. 3b, unlike FIG. 3a, the LTE / LTE-A-based cell is connected to a core network for 5th generation mobile communication, i.e., a 5G core network.

[0084] A service method based on an architecture as illustrated in Figures 3a and 3b above is called NSA (non-standalone).

[0085] Referring to Figure 3c, the UE is connected only to NR-based cells. A service method based on this architecture is called SA (standalone).

[0086] Meanwhile, in the above NR, it may be considered that reception from the base station utilizes a downlink subframe, and transmission to the base station utilizes an uplink subframe. This method can be applied to paired and unpaired spectrums. A pair of spectrums means that two carrier spectrums are included for downlink and uplink operations. For example, in a pair of spectrums, one carrier may include a downlink band and an uplink band that are paired with each other.

[0087] Figure 4 illustrates the slot structure of an NR frame.

[0088] A slot contains multiple symbols in the time domain. For example, in the case of a normal CP, one slot contains 14 symbols, but in the case of an extended CP, one slot contains 12 symbols. A carrier contains multiple subcarriers in the frequency domain. An RB (Resource Block) is defined as multiple (e.g., 12) consecutive subcarriers in the frequency domain. A BWP (Bandwidth Part) is defined as multiple consecutive (physical, P)RBs in the frequency domain, and can correspond to a single numerology (e.g., SCS, CP length, etc.). A terminal can be configured with up to N (e.g., 4) BWPs in the downlink and uplink, respectively. Downlink or uplink transmission is performed through an activated BWP, and at a given time, only one BWP among the BWPs configured for the terminal can be activated. In the resource grid, each element is referred to as a Resource Element (RE), to which one complex symbol can be mapped.

[0089] Figure 5 illustrates an example of subframe types in NR.

[0090] The transmission time interval (TTI) illustrated in FIG. 5 may be referred to as a subframe or slot for NR (or new RAT). The subframe (or slot) of FIG. 5 may be used in a TDD system of NR (or new RAT) to minimize data transmission delay. As illustrated in FIG. 5, a subframe (or slot) includes 14 symbols. The symbols in the front of the subframe (or slot) may be used for a downlink (DL) control channel, and the symbols in the back of the subframe (or slot) may be used for an uplink (UL) control channel. The remaining symbols may be used for DL ​​data transmission or UL data transmission. According to this subframe (or slot) structure, downlink transmission and uplink transmission may be sequentially performed in one subframe (or slot). Therefore, downlink data may be received within a subframe (or slot), and an uplink acknowledgment (ACK / NACK) may be transmitted within the subframe (or slot).

[0091] The structure of these subframes (or slots) can be called self-contained subframes (or slots).

[0092] Specifically, the first N symbols in a slot are used to transmit a DL control channel (hereinafter, DL control region), and the last M symbols in the slot can be used to transmit a UL control channel (hereinafter, UL control region). N and M are each integers greater than or equal to 0. A resource region (hereinafter, data region) between the DL control region and the UL control region can be used for DL ​​data transmission or UL data transmission. For example, a physical downlink control channel (PDCCH) can be transmitted in the DL control region, and a physical downlink shared channel (PDSCH) can be transmitted in the DL data region. A physical uplink control channel (PUCCH) can be transmitted in the UL control region, and a physical uplink shared channel (PUSCH) can be transmitted in the UL data region.

[0093] Using this subframe (or slot) structure has the advantage of minimizing the final data transmission latency by reducing the time required to retransmit data that has experienced reception errors. In this self-contained subframe (or slot) structure, a time gap may be required during the transition from transmit mode to receive mode or from receive mode to transmit mode. To this end, some OFDM symbols during the transition from DL to UL in the subframe structure can be designated as a guard period (GP).

[0094] Figure 6 illustrates the structure of a self-contained slot.

[0095] In an NR system, a frame is characterized by a self-contained structure in which a DL control channel, DL or UL data, and a UL control channel can all be included within a single slot. For example, the first N symbols within a slot can be used to transmit a DL control channel (hereinafter, referred to as a DL control region), and the last M symbols within a slot can be used to transmit a UL control channel (hereinafter, referred to as a UL control region). N and M are each integers greater than or equal to 0. The resource region (hereinafter, referred to as a data region) between the DL control region and the UL control region can be used for DL ​​data transmission or UL data transmission. As an example, the following configuration can be considered. Each section is listed in chronological order.

[0096] 1. DL only configuration

[0097] 2. UL only configuration

[0098] 3. Mixed UL-DL configuration

[0099] - DL area + GP (Guard Period) + UL control area

[0100] - DL control area + GP + UL area

[0101] DL area: (i) DL data area, (ii) DL control area + DL data area

[0102] UL domain: (i) UL data domain, (ii) UL data domain + UL control domain

[0103] In the DL control region, a PDCCH can be transmitted, and in the DL data region, a PDSCH can be transmitted. In the UL control region, a PUCCH can be transmitted, and in the UL data region, a PUSCH can be transmitted. In the PDCCH, downlink control information (DCI), for example, DL data scheduling information, UL data scheduling information, etc., can be transmitted. In the PUCCH, uplink control information (UCI), for example, ACK / NACK (Positive Acknowledgement / Negative Acknowledgement) information for DL ​​data, CSI (Channel State Information) information, SR (Scheduling Request), etc., can be transmitted. GP provides a time gap when a base station and a terminal switch from transmission mode to reception mode or when switching from reception mode to transmission mode. Some symbols at the time of switching from DL to UL within a subframe can be set as GP.

[0104] <NR에서 빔 관리(beam management)>

[0105] Current 3GPP NR beam management methods can be divided into the initial access phase and the cell connection establishment phase. A terminal performing the initial access procedure establishes its initial transmit / receive (Tx / Rx) beam through a random access procedure, i.e., the random access channel (RACH) procedure.

[0106] Figure 7 shows an example of initial beam measurement and selection in NR.

[0107] Referring to Fig. 7, in order to provide base station transmission beam (gNB Tx beam) settings to terminals (UE1 / UE2) without cell connection, the base station periodically and repeatedly transmits SSBs (synchronization signal blocks) with beams in different directions mapped. In addition, SSBs can be transmitted at 20ms cycles within 5ms. Specifically, the default value for initial cell selection can be 20ms.

[0108] A terminal can select a qualified SSB through signal measurement for periodically transmitted SSBs and transmit a PRACH (physical random access channel) preamble mapped to the selected SSB, thereby informing the base station of information about the selected Tx beam. For example, based on signal strength measurements, terminals at different locations, i.e., UE1, can select an SSB with an SSB index of 3, and UE2, can select an SSB with an SSB index of 9, and UE1 and UE2 can each transmit a corresponding PRACH preamble for the selected SSB. Here, it is assumed that each SSB is beamformed in a specific direction.

[0109] Figure 8 shows an example of an initial connection procedure between a terminal and a base station in NR.

[0110] Referring to Fig. 8, after the terminal (UE) is powered on (S801), the UE receives cell-related parameter information (e.g., PRACH information corresponding to each SSB) required in the initial access phase through a system information message transmitted by the base station (gNB) (S802). Here, the system information message includes a master information block (MIB) and a system information block 1 (SIB1) including cell common information.

[0111] After acquiring the system information message, the terminal receives SSBs periodically transmitted from the base station (S803). Then, the terminal measures the RSRP (reference signal received power) for the received SSBs. Among the N SSBs, i.e., beams, the one SSB (beam) with the highest / qualified value is selected (S804).

[0112] Thereafter, the terminal transmits an RA (random access) preamble belonging to the PRACH resource corresponding to the selected SSB (beam) to the base station (S805). Through this, the terminal can inform the base station of the selected initial beam information.

[0113] The base station receives a RA (random access) preamble belonging to a PRACH resource corresponding to the SSB (beam) selected from the terminal, and in response transmits a RAR (random access response) to the terminal using the selected SSB (beam) (S806).

[0114] Meanwhile, a base station that does not know the location / beam information of a terminal that first enters a cell, i.e. a terminal performing a CBRA (contention based random access) procedure, can set up to 64 beams in common with the cell to set the beam of a terminal that has no connection, and the terminal sequentially measures all beams to find the optimal beam at its location. This not only causes a time delay in beam selection and cell connection as the number of beams in the cell increases, but can also increase the power consumption of the terminal by requiring the terminal to measure a large number of beams.

[0115] To address the aforementioned issues, the base station can map a wide beam for SSB to determine the approximate location / beam of the initially connected terminal, and then perform beam refinement to configure a narrow beam after the terminal accesses the cell. However, while a narrow beam provides a high data rate to the terminal, it is sensitive to the terminal's movement or environmental changes, which can easily cause disconnections. To address this, the base station allocates a CSI resource (CSI-RS / SSB) with a candidate beam mapped to the terminal in a UE-specific manner, allowing the terminal to continuously measure the surrounding beam strength and report the measurement results to the base station. This can be configured by the base station through CSI resource configuration and CSI report configuration.

[0116] Figure 9 shows an example of candidate beam settings in NR.

[0117] A terminal that has been configured for beam reporting performs measurement of the reference signal (RS) assigned to it and reports the result based on the configuration of the base station. This follows the CSI framework defined by 3GPP. However, this UE-specific CSI configuration method has a problem in that as the number of terminals in the cell increases, the RS resources allocated to each terminal also increase rapidly. To alleviate this resource overhead problem, the base station can select a method of allocating the same candidate beam, i.e., CSI resources, to terminals in similar locations, as shown in FIG. 9. This can be referred to as UE group-specific CSI resource configuration. However, when terminals with different mobility share the same resource, an issue arises in which new candidate beam resources must be allocated to terminals that leave the corresponding resource area. If a minimum number of candidate beams is allocated to a UE with high / medium mobility to reduce resource overhead, the UE will experience frequent RRC reconfigurations. These candidate beam reconfigurations through RRC incur relatively large delays, potentially leading to beam dropouts. To mitigate this issue, the base station can operate candidate beams by appropriately increasing the number of beams within the CSI resource set. However, from the UE's perspective, this may present a trade-off: the increased number of beams increases the measurement burden.

[0118] Figures 10a to 10c illustrate three procedures for beam management in NR.

[0119] Beam management in NR can be defined by dividing it into three procedures in terms of the procedures defined in the physical layer. Fig. 10a shows Procedure 1 (P1), Fig. 10b shows Procedure 2 (P2), and Fig. 10c shows Procedure 3 (P3), respectively. P1 is an operation to find a transmit / receive beam pair while simultaneously performing TRP (transmission reception point) beam sweeping and UE beam sweeping, similar to the beam setting method of a terminal performing the initial access procedure described above. A terminal that enters connected mode recognizes that the beams set by the base station through candidate beam (i.e., CSI resource set) setting will be swept, and first performs signal strength measurement for the TRP beam. When the terminal's TRP beam is selected through P2, the base station repeatedly transmits the selected one beam through P3. A terminal can select a UE beam while performing UE beam sweeping. The UE's choice of beam during this operation is left to the terminal implementation. The aforementioned operation can be applied to both downlink (DL) and uplink (UL).

[0120] Figures 11a to 11c illustrate examples of beam reporting procedures in NR.

[0121] Beam sweeping uses a method in which the base station notifies the terminal of reference signal (RS) resource information by configuring a specific candidate beam, i.e., a CSI resource set, so that information about the beam is implicitly provided by being mapped to the RS resource information. In other words, rather than notifying the terminal of the actual beam index, the base station recognizes the information about the mapped beam through the index information implicitly mapped to the RS information using the RS resource indicator (RI). This is configured using the 3GPP CSI framework, and the terminal implicitly reports RSRP information for the best four beams (RI) to the base station by measuring the RS strength for the resources configured by the base station. The method for reporting the measurement results also depends on the RRC configuration of the base station, and 3GPP defines one of the following three methods to be configured.

[0122] - Periodic reporting

[0123] - Aperiodic reporting

[0124] - Semi-persistent reporting

[0125] Figure 11a illustrates a periodic CSI reporting method, which is triggered through RRC configuration. That is, the terminal receives an RRC configuration message from the base station, and the RRC configuration message includes settings for CSI-related RS resources and reporting methods, i.e., CSI resource set information, and information that CSI reporting is periodic (S1101a). Thereafter, the terminal receives RSs periodically transmitted based on the received RRC configuration message (S1102a and S1105a), and measures the signal strength for a beam based on the received RSs (S1103a and S1106a). Then, the terminal periodically reports the measured results (values) to the base station (S1104a and S1107a).

[0126] Figure 11b illustrates an aperiodic CSI reporting method. Even if CSI-related RS resources and a reporting method are configured through an RRC configuration message, beam measurement through RS is not performed without a trigger message (or information) from a lower layer. That is, the terminal receives an RRC configuration message including CSI resource set information and information that CSI reporting is aperiodic from the base station regarding configuration of CSI-related RS resources and a reporting method (i.e., CSI resource set information) and information that CSI reporting is aperiodic (S1101b), and the CSI report trigger is performed through a medium access control (MAC) control element (CE) or downlink control information (DCI). The terminal receives CSI report trigger information including a trigger indication from the base station through the MAC CE or DCI (S1102b), and receives RSs transmitted once based on the received trigger indication (S1103b). Here, the transmission of RSs for the CSI resource set can be transmitted after a specific time (e.g., X slots) at which the CSI report trigger information is transmitted. Thereafter, the terminal measures the signal strength for the beam based on the received RSs (S1104b). Then, the terminal reports the measured result (value) to the base station once (S1105b). Here, the CSI report can be transmitted after a specific time (e.g., Y slots) at which the CSI report trigger information is received.

[0127] Fig. 11c shows a semi-persistent reporting method, which is an intermediate method between the periodic reporting method and the aperiodic reporting method. Upon receiving a configuration for CSI-related RS resources and a reporting method through an RRC configuration message, the terminal performs CSI reporting periodically until receiving a deactivation message (or information) only when activated by MAC CE. That is, the terminal receives an RRC configuration message from the base station that includes a configuration for CSI-related RS resources and a reporting method, that is, CSI resource set information and information that CSI reporting is semi-persistent (S1101c), and CSI report activation is performed through MAC CE. The terminal receives CSI report activation information including an activation indication from the base station via MAC CE (S1102c and S1110c), receives RSs periodically transmitted based on the received activation indication (S1103c, S1106c, S1111c and S1114c), and measures signal strength for a beam based on the received RSs (S1104c, S1107c, S1112c and S1115c). Then, the terminal periodically reports the measured result (value) to the base station (S1105c, S1108c, S1113c and S1116c). After CSI reporting is activated, if CSI report deactivation information including a deactivation indication is received from the base station via MAC CE (S1109c), the terminal stops CSI reporting.

[0128] 3GPP is conducting research on technology that applies AI / ML models to improve delay and terminal power consumption in the aforementioned beam search (measurement).

[0129] The list of terms applied to AI / ML is discussed as shown in Table 5 below.

[0130] Terminology Description Data collection A process of collecting data by the network nodes, management entity, or UE for the purpose of AI / ML model training, data analytics and inference AI / ML Model A data driven algorithm that applies AI / ML techniques to generate a set of outputs based on a set of inputs.)AI / ML model trainingA process to train an AI / ML Model [by learning the input / output relationship] in a data driven manner and obtain the trained AI / ML Model for inference. AI / ML model inferenceA process of using a trained AI / ML model to produce a set of outputs based on a set of inputs. AI / ML model validationA subprocess of training to evaluate the quality of an AI / ML model using a dataset different from the one used for model training, that helps selecting model parameters that generalize beyond the dataset used for model training. AI / ML model testing Testing is a sub-process of training that evaluates the performance of the final AI / ML model using a dataset different from that used for model training and validation. Unlike AI / ML model validation, testing does not assume subsequent adjustments to the model.(A subprocess of training, to evaluate the performance of a final AI / ML model using a dataset different from one used for model training and validation. Differently from AI / ML model validation, testing does not assume subsequent tuning of the model.)UE-side (AI / ML) modelAn AI / ML Model whose inference is performed entirely at the UENetwork-side (AI / ML) modelAn AI / ML Model whose inference is performed entirely at the networkOne-sided (AI / ML) modelA UE-side (AI / ML) model or a Network-side (AI / ML) modelTwo-sided (AI / ML) modelA pair of AI / ML model(s) on which joint inference is performed. Here, joint inference is AI / ML inference where inference is performed jointly by the UE and the network, i.e., the first part of the inference is performed by the UE first and the remaining part is performed by the gNB, or vice versa.(A paired AI / ML Model(s) over which joint inference is performed, where joint inference comprises AI / ML Inference whose inference is performed jointly across the UE and the network, i.e., the first part of inference is firstly performed by the UE and then the remaining part is performed by gNB, or vice versa.)AI / ML model transferTransfer of an AI / ML model over a wireless interface with parameters of a model structure known to the receiving end or a new model having parameters. Delivery of an AI / ML model over the air interface, either parameters of a model structure known at the receiving end or a new model with parameters. Delivery may contain a full model or a partial model. Model download: Model transfer from the network to UE. Model upload: Model transfer from UE to the network. Federated learning / federated training: A machine learning technique for training an AI / ML model on multiple distributed edge nodes (e.g., UEs, gNBs), each performing local model training using local data samples. This technique requires multiple interactions of the model but does not require the exchange of local data samples.(A machine learning technique that trains an AI / ML model across multiple decentralized edge nodes (e.g., UEs, gNBs) each performing local model training using local data samples. The technique requires multiple interactions of the model, but no exchange of local data samples.) Offline field data: The data collected from the field and used for offline training of the AI / ML model. Online field data: The data collected from the field and used for online training of the AI / ML model. Model monitoring: A procedure that monitors the inference performance of the AI / ML model. Supervised learning: A process of training a model from input and its corresponding labels. Unsupervised learning: A process of training a model without labeled data.)Semi-supervised learning: A process of training a model with a mix of labeled and unlabeled data. Reinforcement Learning (RL): A process of training an AI / ML model from input (aka state) and a feedback signal (aka reward) resulting from the model's output (aka action) in an environment the model is interacting with.. Model activation: Enable an AI / ML model for a specific function. Model deactivation: Disable an AI / ML model for a specific function. Model switching: Deactivating a currently active AI / ML model and activating another AI / ML model for a specific function. a different AI / ML model for a specific function.

[0131] 3GPP RAN1 has been discussing model monitoring techniques for each use case and is considering it as one of the key factors for maintaining the performance of AI / ML models.

[0132] Current AI / ML models aim to improve overall system performance by applying model inference output to communication techniques to reduce DL (downlink) RS (reference signal) overhead and terminal measurement burden. Furthermore, they utilize models to obtain more accurate results, thereby improving overall system performance. To maintain consistent performance, these AI / ML models require continuous monitoring. Various metrics were identified as candidates for model monitoring during the previous study period, and the information reported from terminals to base stations can also vary depending on which node performs this performance monitoring.

[0133] As more detailed standardization discussions begin in the future, procedures related to AI / ML model monitoring on terminals are expected to be defined. The most intuitive way to evaluate model performance is through model accuracy. This requires comparing the output derived from model inference with the results of actual communication environments. Additional RS transmissions, which serve as benchmarks / references for model performance measurement, need to be defined in relation to the model inference time, but there is currently no specific discussion or definition for this. Furthermore, if benchmark / reference RSs are required for monitoring, excessively frequent transmission or measurement of these RSs can incur additional overhead for base stations and terminals, and if not monitored in a timely manner, this can degrade system performance. To address this trade-off, a method for efficiently performing monitoring is needed.

[0134] As a solution to the above, this specification proposes that, when a benchmark / reference RS resource for measuring the performance of an operating model is periodically transmitted at a terminal or base station performing wireless communication using an AI / ML model, the cycle for the resource can be set to at least one cycle based on the DL RS cycle set for model inference. Additionally, when more than one monitoring cycle is set for a terminal, a method is proposed to adaptively change the model monitoring cycle based on model performance results and arbitrary settings / conditions.

[0135] More specifically, when the base station configures DL RS resources for model inference for a terminal, it configures additional DL RS resources associated with the DL RS resources for model inference for the purpose of monitoring, and the DL RS resources for monitoring can have at least one or more periods. Here, the transmission period for the DL RS resources for monitoring can be set to a period(s) that is a multiple of n (where n is a natural number greater than 1) based on the transmission period of the DL RS resources for model inference.

[0136] Fig. 12 is a flowchart illustrating a method of operating a terminal according to one embodiment of the present specification.

[0137] When one or more model monitoring periods are set for a terminal, the base station clearly notifies the terminal of a change in the monitoring DL RS transmission period based on information received from the terminal (e.g., monitoring result or assistance information) or internal information of the base station, or defines that the terminal and the base station recognize the change in the monitoring DL RS transmission period based on an arbitrary condition (e.g., timer or count). If the change in the monitoring period is performed based on signaling, the configuration information for one or more monitoring periods proposed in the present invention is proposed to be transmitted through radio resource control (RRC) control information (or message), and an instruction for changing the monitoring period is proposed to be performed through signaling such as downlink control information (DCI) / uplink control information (UCI) or medium access control (MAC) control element (CE). Additionally, if a condition for changing the monitoring period and a parameter related thereto are set together, it is proposed to change the monitoring period if the set condition is satisfied.

[0138] Referring to FIG. 12, the terminal receives an RRC message from the base station containing DL RS resource information having at least one cycle for model monitoring (S1201). Based on the received information, the terminal performs model monitoring every first cycle (S1202). The terminal confirms that a condition set for changing the cycle is satisfied (S1203). Here, satisfaction of the condition may be the expiration of a timer or reaching a maximum count value. Thereafter, the terminal performs model monitoring every second cycle (S1204).

[0139] FIG. 13 is an example showing multi-level model monitoring according to one embodiment of the present specification.

[0140] The monitoring cycle proposed in this specification may be set to at least one or more cycles (i.e., the 1st to Nth cycles) for each terminal / model / function as a multiple (N*x) of the inference cycle (x) for the corresponding model, as shown in Fig. 13. The terminal performs inference using the DL RS for model inference transmitted from the base station based on the inference cycle (x), and performs model monitoring using the DL RS for model monitoring transmitted from the base station based on the monitoring cycle (e.g., N*x).

[0141] If the terminal is configured with one or more monitoring periods via RRC, the terminal may start monitoring by setting the shortest period (e.g., 1st period) as the default model monitoring period or by explicitly receiving a default monitoring period instruction from the base station.

[0142] Referring to FIG. 13, if a terminal is set with n model monitoring periods, the terminal starts model monitoring with the indicated or default model monitoring period. Here, it is assumed that the 1st to Nth model monitoring periods are set in the order of the length of the period from the shortest period. When the terminal performs model monitoring from the first period, the base station and the terminal may want to change the model monitoring to a longer period depending on the model performance result value or any set condition. That is, when the model accuracy is sufficiently high according to the model monitoring result, or when the terminal's battery is low, or when the terminal's movement speed is low, and conditions do not require frequent monitoring, the terminal may change the monitoring period of the terminal to a longer period so that the terminal performs model monitoring intermittently. Conversely, when the terminal's movement speed is high, or when the terminal's battery is sufficient, or when the model's accuracy is low, and conditions require frequent model monitoring, the terminal may be made to perform model monitoring with a shorter period.

[0143] Below, specific methods for changing one or more set model monitoring cycles are described.

[0144] As a method for changing model monitoring based on set conditions, the following can be configured to implicitly change the model monitoring cycle of the terminal through parameter settings such as a timer or COUNT. This is preferably applied when the terminal calculates a monitoring performance metric. That is, when the terminal calculates a metric and reports the result value to the base station only under certain conditions (e.g., when the performance result is below a set threshold), the terminal and base station can mutually manage the timer or COUNT operation based on the model performance result value, and change the monitoring cycle according to the expiration of the timer or the arrival of the set COUNT value.

[0145] Option 1: Change the timer-based model monitoring.

[0146] First, we describe a method for changing the model monitoring cycle using a timer. A terminal can receive monitoring-related DL RSs with one or more cycles from a base station and be configured to report model monitoring results periodically or aperiodically. Furthermore, the configuration message can include a value for the timer proposed in this specification.

[0147] FIG. 14 is an example showing a timer-based model monitoring cycle change according to one embodiment of the present specification.

[0148] If a terminal is configured with a timer associated with both an aperiodic report and a monitoring cycle, the terminal starts the configured timer at the time defined below.

[0149] - Upon receiving an RRC message containing settings for model monitoring (DL RS / reporting); or

[0150] - As soon as the first DL RS resource for model inference set by the RRC message is received / measured; or

[0151] - Upon receiving / measuring the first DL RS resource for model monitoring set by the RRC message; or

[0152] - as soon as the report message associated with the first DL RS resource for model inference established by the RRC message is transmitted; or

[0153] - As soon as the first monitoring result for model monitoring set by RRC message is derived.

[0154] If, while the timer is running, the terminal reports to the base station a result indicating poor model performance as a result of model monitoring (e.g., a model performance result value below a threshold set by the base station or an indicator indicating the same meaning), the timer is restarted and the terminal changes the monitoring cycle to a cycle shorter than the current cycle.

[0155] If the timer expires, the terminal changes the model monitoring cycle to a cycle longer than the current cycle and performs monitoring based on the changed cycle. After the cycle is changed, the timer can be restarted according to the start condition of the timer. However, since the cycle will be changed without receiving an RRC message, it is preferable that the timer be (re)started at the time of reception / measurement of the first DL RS resource after the expiration of the timer for model inference or model monitoring set by the RRC message, transmission of a related report message, or derivation of a monitoring result.

[0156] Alternatively, the timer expiration may be defined to trigger signaling generation by the terminal or base station to request / indicate a change in the model monitoring period. That is, when the timer expires, the terminal may request / indicate a change in the monitoring period (e.g., to a longer period) to the base station via transmission of uplink control information (UCI) or medium access control (MAC) control element (CE), or the base station may instruct the terminal to change the monitoring period via transmission of downlink control information (DCI) or MAC CE.

[0157] The timer can be set to a value corresponding to the monitoring period, either as the number of monitoring periods (n) or as a number one less than the number of configured periods (n-1), or as a single value for the terminal. If it is set to a number one less than the number of configured periods or if one timer is set for the terminal, when the terminal performs model monitoring using the longest period (Nth period), the timer does not run, and the timer is restarted by the monitoring result report, and the monitoring period can be changed to a period shorter than the current period. When operating with the shortest period (1st period) among the configured periods, the timer is restarted without changing the period even if the model monitoring result is reported. In other words, this means that the timer is restarted while performing a period change to a period shorter than the current period only when model monitoring is performed based on a period other than the shortest period (1st period).

[0158] If a life cycle management (LCM) related signal is sent or received, such as deactivating the currently active model, switching to a new model, or falling back to legacy operation, the timer stops.

[0159] FIG. 15 is an example showing a timer-based model monitoring cycle change according to another embodiment of the present specification.

[0160] When a terminal is configured to receive monitoring-related DL RSs with one or more cycles from a base station and to periodically report model monitoring results, the terminal starts the corresponding timer at the time defined below.

[0161] - Upon receiving an RRC message containing settings for model monitoring (DL RS / reporting); or

[0162] - As soon as the first DL RS resource for model inference set by the RRC message is received / measured; or

[0163] - Upon receiving / measuring the first DL RS resource for model monitoring set by the RRC message; or

[0164] - as soon as the report message associated with the first DL RS resource for model inference established by the RRC message is transmitted; or

[0165] - As soon as the first monitoring result for model monitoring set by RRC message is derived.

[0166] In the case where the terminal periodically reports the model monitoring results, the timer can be defined to restart when the terminal reports to the base station a result indicating poor model performance (e.g., a model performance result value below a threshold set by the base station or an indicator indicating the same meaning) in the same manner as the aperiodic reporting method described above. As a result, the monitoring cycle can be defined to change at the same time as the aperiodic reporting method. In other words, the report of the model monitoring result indicating good model performance does not affect the timer operation.

[0167] Hereinafter, the terminal operation is described in detail with reference to FIGS. 14 and 15.

[0168] The terminal receives an RRC message from the base station containing configuration information for DL ​​RS resources and reporting related to the AI / ML model. The message may include at least one of the following information:

[0169] - DL RS resource and reporting method configuration information for AI / ML model inference. This may include inference cycle (e.g., x slots) configuration information.

[0170] - DL RS resource and reporting method configuration information for AI / ML model monitoring associated with AI / ML model inference. Here, i) at least one monitoring cycle configuration information (e.g., x slots and 2x slots), ii) timer value configuration information associated with monitoring, and / or iii) aperiodic reporting configuration information and reporting condition configuration information (e.g., when the performance result is below a threshold and indicates poor performance) may be included.

[0171] The terminal performs monitoring based on the period according to the received RRC message. Here, the period may be the first period according to the RRC message or a specific period (this period).

[0172] If the monitoring period is not the longest period (Nth period) among the set periods and a DL RS for model monitoring is received, the set timer is started.

[0173] If the model monitoring result satisfies the conditions for model monitoring reporting, i) the terminal reports the model monitoring result to the base station, ii) restarts the timer, and iii) if a shorter monitoring period than the current period exists, changes the monitoring period to a shorter period.

[0174] If the timer expires, the terminal decides to change the monitoring period to a period ((i+1)th period) that is longer than the current period (ith period), and performs monitoring based on the changed period thereafter.

[0175] If a life cycle management (LCM) instruction (e.g., model deactivation / switching / fallback) signaling is received for the terminal model, the timer is stopped.

[0176] Hereinafter, the operation of the base station will be described in detail with reference to FIGS. 14 and 15.

[0177] The base station transmits an RRC message to the terminal containing configuration information for DL ​​RS resources and reporting related to the AI / ML model. The message may include at least one of the following information:

[0178] - DL RS resource and reporting method configuration information for AI / ML model inference. This may include inference cycle (e.g., x slots) configuration information.

[0179] - DL RS resource and reporting method configuration information for AI / ML model monitoring associated with AI / ML model inference, which may include: i) information for configuring at least one monitoring cycle (e.g., x slots and 2x slots); ii) information for configuring a timer value associated with monitoring; and / or iii) information for configuring aperiodic reporting and conditions for reporting (e.g., when the performance result is below a threshold).

[0180] The base station transmits DL RS for monitoring based on the period according to the transmitted RRC message. Here, the period may be the first period according to the RRC message or a specific period (this period).

[0181] If the monitoring period is not the longest period (Nth period) among the set periods and a DL RS for model monitoring is transmitted, the set timer is started.

[0182] If the model monitoring result received from the terminal satisfies the conditions for model monitoring report, i) the timer is restarted, and iii) if there is a monitoring cycle shorter than the current cycle, DL RS is transmitted based on the shorter cycle.

[0183] If the timer expires, the base station decides to change the monitoring period to a longer period ((i+1)th period) than the current period (ith period), and then transmits DL RS based on the changed period.

[0184] If the terminal transmits a life cycle management (LCM) instruction (e.g., model deactivation / switching / fallback) signal for the model, the timer is stopped.

[0185] Option 2: Modifying COUNT-Based Model Monitoring

[0186] Below, a method utilizing counts is described. A terminal can be configured to receive monitoring-related DL RSs with one or more cycles from a base station and report model monitoring results periodically or aperiodically. Furthermore, the configuration message can include the maximum value (MAX value) for the count proposed in this specification.

[0187] FIG. 16 is an example showing a COUNT-based model monitoring cycle change according to one embodiment of the present specification.

[0188] When the terminal receives a monitoring-related DL RS having one or more periods from the base station and is set to report model monitoring results aperiodically, the terminal sets the COUNT parameter to 0.

[0189] If the terminal measures DL RS resources for model monitoring and calculates the monitoring result, but does not satisfy the condition for reporting the result to the base station (for example, a result indicating poor performance result), and the current cycle is not the longest cycle among the set cycles, then COUNT is increased by 1.

[0190] If the terminal measures DL RS resources for model monitoring and reports the monitoring results to the base station (e.g., results indicating poor performance results), COUNT is reset to 0, and if a shorter period than the current period exists, the monitoring period is changed to a shorter period.

[0191] If COUNT reaches the MAX value, the terminal changes the cycle to a longer cycle than the current cycle and performs monitoring based on the changed cycle. When the cycle is changed, COUNT is reset to 0.

[0192] Alternatively, when COUNT reaches MAX, the terminal or base station may trigger signaling generation to request / indicate a change in the model monitoring period. This means that when the timer expires, the terminal may request / indicate a change in the monitoring period (e.g., to a longer period) to the base station via UCI or MAC CE transmission, or the base station may instruct the terminal to change the monitoring period via DCI or MAC CE transmission.

[0193] The COUNT can be set to different values ​​for each terminal cycle or for each terminal. If set to a terminal cycle, it is preferable to set the COUNT value corresponding to the monitoring cycle to a number (n-1) less than the number of monitoring cycles.

[0194] Hereinafter, terminal operation is described in detail with reference to FIG. 16.

[0195] The terminal receives an RRC message from the base station containing configuration information for DL ​​RS resources and reporting related to the AI / ML model. The message may include at least one of the following information:

[0196] - DL RS resource and reporting method configuration information for AI / ML model inference. This may include inference cycle (e.g., x slots) configuration information.

[0197] - DL RS resource and reporting method configuration information for AI / ML model monitoring associated with AI / ML model inference. Here, i) at least one monitoring cycle configuration information (e.g., x slots and 2x slots), ii) COUNT MAX value configuration information associated with monitoring, and / or iii) aperiodic reporting configuration information and conditions for reporting (e.g., when the performance result is below a threshold and indicates poor performance) may be included.

[0198] The terminal sets the COUNT parameter to 0.

[0199] The terminal performs monitoring based on the period according to the received RRC message. Here, the period may be the first period according to the RRC message or a specific period (this period).

[0200] If the monitoring period is not the longest period (Nth period) among the set periods, COUNT is increased by 1.

[0201] If the model monitoring result satisfies the conditions for model monitoring reporting, i) the terminal reports the model monitoring result to the base station, ii) resets COUNT, and iii) if the current cycle is not the shortest cycle among the set cycles, the terminal changes the monitoring cycle to a cycle shorter than the current cycle and performs monitoring.

[0202] If the above COUNT value reaches MAX, i) the terminal decides to change the monitoring cycle to a cycle longer than the current cycle, ii) performs model monitoring based on the changed cycle, and iii) resets COUNT.

[0203] If a life cycle management (LCM) instruction (e.g., model deactivation / switching / fallback) signaling is received for the terminal model, the corresponding COUNT is set to 0.

[0204] Hereinafter, the operation of the base station is described in detail with reference to FIG. 16.

[0205] The base station transmits an RRC message to the terminal containing configuration information for DL ​​RS resources and reporting related to the AI / ML model. The message may include at least one of the following information:

[0206] - DL RS resource and reporting method configuration information for AI / ML model inference. This may include inference cycle (e.g., x slots) configuration information.

[0207] - DL RS resource and reporting method configuration information for AI / ML model monitoring associated with AI / ML model inference, which may include: i) information for configuring at least one monitoring cycle (e.g., x slots and 2x slots); ii) information for configuring a COUNT MAX value associated with monitoring; and / or iii) information for configuring aperiodic reporting and conditions for reporting (e.g., when the performance result is below a threshold).

[0208] The base station sets the COUNT parameter to 0.

[0209] The base station transmits DL RS for monitoring based on the period according to the transmitted RRC message. Here, the period may be the first period according to the RRC message or a specific period (this period).

[0210] If the model monitoring result is not received from the terminal and the current period is not the longest period (Nth period), COUNT is increased by 1.

[0211] If the model monitoring result is received from the terminal, i) COUNT is reset, and iii) if the current cycle is not the shortest cycle among the set cycles, the monitoring cycle is changed to a cycle shorter than the current cycle and a DL RS for monitoring is transmitted.

[0212] If the above COUNT value reaches MAX, i) the base station decides to change the monitoring period to a period longer than the current period, ii) transmits a DL RS for model monitoring based on the changed period, and iii) resets the COUNT.

[0213] If the terminal transmits LCM (life cycle management) instructions (e.g., model deactivation / switching / fallback) signaling for the model, set the COUNT to 0.

[0214] If parameter settings such as the timer and COUNT proposed in this specification are defined in a higher layer (e.g., MAC layer), changes in the monitoring period can be notified to a lower layer (e.g., PHY layer).

[0215] Meanwhile, even if the conditions for setting and reporting the aperiodic reporting of this specification are defined in the opposite way to what was described above (i.e., reporting to the base station is performed only when the performance result is better than the threshold), the method using the timer and COUNT can be applied, and it is obvious that modified examples that produce the same effect based on the technical idea of ​​the invention can be implemented.

[0216] Fig. 17 is a flowchart illustrating a method of operating a terminal according to another embodiment of the present specification.

[0217] Referring to Figure 17, the terminal receives configuration information for performance monitoring of an AI / ML model from a base station (S1701). The configuration information may include information on at least one monitoring cycle and cycle change conditions associated with a reference signal (RS).

[0218] The terminal performs performance monitoring of the AI / ML model based on the first monitoring cycle among at least one monitoring cycle (S1702).

[0219] Thereafter, the terminal performs performance monitoring of the AI / ML model based on the second monitoring cycle among at least one monitoring cycle, depending on satisfaction of the cycle change condition (S1703).

[0220] The above periodic change conditions may include aperiodic reporting conditions of performance monitoring results of the AI / ML model.

[0221] Meanwhile, the above-described configuration information further includes timer information, and a timer may be started based on the timer information. While the timer is running, the terminal may transmit the performance monitoring results of the AI / ML model to the base station based on the satisfaction of the aperiodic reporting conditions for the performance monitoring results of the AI / ML model. Furthermore, the terminal may restart the timer based on the transmission of the performance monitoring results of the AI / ML model. Here, the second monitoring period may be shorter than the first monitoring period.

[0222] Alternatively, the above-described configuration information may further include timer information, and when a timer started based on the timer information expires, performance monitoring of the AI / ML model may be performed based on the second monitoring cycle. Here, the second monitoring cycle may be longer than the first monitoring cycle.

[0223] On the other hand, after the COUNT value associated with the performance monitoring of the AI / ML model increases, and the aperiodic reporting condition of the performance monitoring result of the AI / ML model is satisfied, the terminal may transmit the performance monitoring result of the AI / ML model to the base station. Upon transmission of the performance monitoring result of the AI / ML model, the COUNT value may be reset. Here, the second monitoring period may be shorter than the first monitoring period.

[0224] On the other hand, the above setting information further includes information on a COUNT MAX value associated with performance monitoring of the AI / ML model, and when the increased COUNT value associated with performance monitoring of the AI / ML model reaches the COUNT MAX value, performance monitoring of the AI / ML model can be performed based on the second monitoring cycle. Here, the second monitoring cycle can be longer than the first monitoring cycle.

[0225] Fig. 18 is a flowchart illustrating an operation method of a base station according to one embodiment of the present specification.

[0226] Referring to Figure 18, the base station transmits configuration information for performance monitoring of the AI / ML model to the terminal (S1801). The configuration information may include information on at least one monitoring cycle and cycle change conditions associated with a reference signal (RS).

[0227] The base station transmits a reference signal based on a first monitoring period among at least one monitoring period (S1802).

[0228] Thereafter, the base station transmits a reference signal based on the second monitoring cycle among at least one monitoring cycle, depending on satisfaction of the cycle change condition (S1803).

[0229] The above periodic change conditions may include aperiodic reporting conditions of performance monitoring results of the AI / ML model.

[0230] Meanwhile, the above-described configuration information further includes timer information, and a timer may be started based on the timer information. While the timer is running, the base station may receive the performance monitoring results of the AI / ML model from the terminal based on the satisfaction of the aperiodic reporting conditions for the performance monitoring results of the AI / ML model. Furthermore, the base station may restart the timer based on the reception of the performance monitoring results of the AI / ML model. Here, the second monitoring period may be shorter than the first monitoring period.

[0231] On the other hand, the above setting information further includes timer information, and when a timer started based on the timer information expires, transmission of a reference signal based on the second monitoring period may be performed. Here, the second monitoring period may be longer than the first monitoring period.

[0232] On the other hand, after the COUNT value associated with the performance monitoring of the AI / ML model increases, and the aperiodic reporting condition of the performance monitoring result of the AI / ML model is satisfied, the base station may receive the performance monitoring result of the AI / ML model from the terminal. Upon receiving the performance monitoring result of the AI / ML model, the COUNT value may be reset. Here, the second monitoring period may be shorter than the first monitoring period.

[0233] On the other hand, the above setting information further includes information on a COUNT MAX value associated with performance monitoring of the AI / ML model, and when the increased COUNT value associated with performance monitoring of the AI / ML model reaches the COUNT MAX value, transmission of a reference signal based on the second monitoring period may be performed. Here, the second monitoring period may be longer than the first monitoring period.

[0234] The concepts disclosed in this specification may be applied independently or may be combined and operated in any form. Furthermore, while this specification is based on a 5G NR system, the scope of this specification encompasses all cases in which the concepts of this specification apply, regardless of the specific wireless communication technology.

[0235] Figure 19 illustrates a device according to one embodiment of the present specification.

[0236] Referring to FIG. 19, the wireless communication system may include a first device (100a) and a second device (100b).

[0237] The first device (100a) may be a base station, a network node, a transmitting terminal, a receiving terminal, a wireless device, a wireless communication device, a vehicle, a vehicle equipped with an autonomous driving function, a connected car, a drone (Unmanned Aerial Vehicle, UAV), an AI (Artificial Intelligence) module, a robot, an AR (Augmented Reality) device, a VR (Virtual Reality) device, an MR (Mixed Reality) device, a hologram device, a public safety device, an MTC device, an IoT device, a medical device, a fintech device (or a financial device), a security device, a climate / environmental device, a device related to 5G services, or any other device related to the 4th industrial revolution field.

[0238] The second device (100b) may be a base station, a network node, a transmitting terminal, a receiving terminal, a wireless device, a wireless communication device, a vehicle, a vehicle equipped with an autonomous driving function, a connected car, a drone (Unmanned Aerial Vehicle, UAV), an AI (Artificial Intelligence) module, a robot, an AR (Augmented Reality) device, a VR (Virtual Reality) device, an MR (Mixed Reality) device, a hologram device, a public safety device, an MTC device, an IoT device, a medical device, a fintech device (or a financial device), a security device, a climate / environmental device, a device related to 5G services, or any other device related to the 4th industrial revolution field.

[0239] The first device (100a) may include at least one processor, such as a processor (1020a), at least one memory, such as a memory (1010a), and at least one transceiver, such as a transceiver (1031a). The processor (1020a) may perform the functions, procedures, and / or methods described above. The processor (1020a) may perform one or more protocols. For example, the processor (1020a) may perform one or more layers of a wireless interface protocol. The memory (1010a) may be connected to the processor (1020a) and may store various types of information and / or commands. The transceiver (1031a) may be connected to the processor (1020a) and may be controlled to transmit and receive wireless signals.

[0240] The second device (100b) may include at least one processor, such as a processor (1020b), at least one memory device, such as a memory (1010b), and at least one transceiver, such as a transceiver (1031b). The processor (1020b) may perform the functions, procedures, and / or methods described above. The processor (1020b) may implement one or more protocols. For example, the processor (1020b) may implement one or more layers of a wireless interface protocol. The memory (1010b) may be connected to the processor (1020b) and may store various types of information and / or commands. The transceiver (1031b) may be connected to the processor (1020b) and may be controlled to transmit and receive wireless signals.

[0241] The memory (1010a) and / or the memory (1010b) may be connected internally or externally to the processor (1020a) and / or the processor (1020b), or may be connected to another processor via various technologies such as a wired or wireless connection.

[0242] The first device (100a) and / or the second device (100b) may have one or more antennas. For example, the antenna (1036a) and / or the antenna (1036b) may be configured to transmit and receive wireless signals.

[0243] Figure 20 is a block diagram showing the configuration of a terminal according to one embodiment of the present specification.

[0244] In particular, FIG. 20 is a drawing illustrating the device of FIG. 19 in more detail.

[0245] The device includes a memory (1010), a processor (1020), a transceiver (1031), a power management module (1091), a battery (1092), a display (1041), an input unit (1053), a speaker (1042), and a microphone (1052), a subscriber identification module (SIM) card, and one or more antennas.

[0246] The processor (1020) may be configured to implement the proposed functions, procedures, and / or methods described herein. Layers of a radio interface protocol may be implemented in the processor (1020). The processor (1020) may include an application-specific integrated circuit (ASIC), other chipsets, logic circuits, and / or data processing devices. The processor (1020) may be an application processor (AP). The processor (1020) may include at least one of a digital signal processor (DSP), a central processing unit (CPU), a graphics processing unit (GPU), and a modem (modulator and demodulator). Examples of the processor (1020) may be a SNAPDRAGON™ series processor manufactured by Qualcomm®, an EXYNOSTM series processor manufactured by Samsung®, an A series processor manufactured by Apple®, a HELIO™ series processor manufactured by MediaTek®, an ATOM™ series processor manufactured by INTEL®, a KIRINTM series processor manufactured by HiSilicon®, or a corresponding next-generation processor.

[0247] The power management module (1091) manages power to the processor (1020) and / or the transceiver (1031). The battery (1092) supplies power to the power management module (1091). The display (1041) outputs the results processed by the processor (1020). The input unit (1053) receives input to be used by the processor (1020). The input unit (1053) can be displayed on the display (1041). A SIM card is an integrated circuit used to securely store an international mobile subscriber identity (IMSI) and its associated keys, which are used to identify and authenticate subscribers in mobile devices such as mobile phones and computers. Contact information can also be stored on many SIM cards.

[0248] The memory (1010) is operably coupled to the processor (1020) and stores various information for operating the processor (610). The memory (1010) may include a read-only memory (ROM), a random access memory (RAM), flash memory, a memory card, a storage medium, and / or other storage devices. When the embodiment is implemented in software, the techniques described herein may be implemented as modules (e.g., procedures, functions, etc.) that perform the functions described herein. The modules may be stored in the memory (1010) and executed by the processor (1020). The memory (1010) may be implemented within the processor (1020). Alternatively, the memory (1010) may be implemented external to the processor (1020) and communicatively connected to the processor (1020) via various means known in the art.

[0249] The transceiver (1031) is operably coupled to the processor (1020) and transmits and / or receives a radio signal. The transceiver (1031) includes a transmitter and a receiver. The transceiver (1031) may include baseband circuitry for processing a radio frequency signal. The transceiver controls one or more antennas to transmit and / or receive a radio signal. The processor (1020) transmits command information to the transceiver (1031) to initiate communication, for example, to transmit a radio signal constituting voice communication data. The antenna functions to transmit and receive radio signals. Upon receiving a radio signal, the transceiver (1031) may transmit the signal to the processor (1020) for processing and convert the signal to baseband. The processed signal may be converted into audible or readable information output through the speaker (1042).

[0250] The speaker (1042) outputs sound-related results processed by the processor (1020). The microphone (1052) receives sound-related input to be used by the processor (1020).

[0251] A user inputs command information, such as a phone number, for example, by pressing (or touching) a button on an input unit (1053) or by voice activation using a microphone (1052). The processor (1020) receives this command information and processes it to perform an appropriate function, such as dialing a phone number. Operational data can be extracted from a SIM card or memory (1010). In addition, the processor (1020) can display command information or operation information on a display (1041) for the user's recognition and convenience.

[0252] Figure 21 shows a block diagram of a processor in which the disclosure of this specification is implemented.

[0253] As can be seen from FIG. 21, the processor (1020) implementing the disclosure of the present specification may include multiple circuits to implement the proposed functions, procedures, and / or methods described herein. For example, the processor (1020) may include a first circuit (1020-1), a second circuit (1020-2), and a third circuit (1020-3). Furthermore, although not shown, the processor (1020) may include more circuits. Each circuit may include multiple transistors.

[0254] The above processor (1020) may be called an application-specific integrated circuit (ASIC) or an application processor (AP), and may include at least one of a digital signal processor (DSP), a central processing unit (CPU), and a graphics processing unit (GPU).

[0255] Fig. 22 is a block diagram showing in detail the transmitter / receiver of the first device illustrated in Fig. 19 or the transmitter / receiver unit of the device illustrated in Fig. 20.

[0256] Referring to FIG. 22, the transceiver unit (1031) includes a transmitter (1031-1) and a receiver (1031-2). The transmitter (1031-1) includes a Discrete Fourier Transform (DFT) unit (1031-11), a subcarrier mapper (1031-12), an IFFT unit (1031-13), a CP insertion unit (1031-14), and a wireless transmitter unit (1031-15). The transmitter (1031-1) may further include a modulator. In addition, for example, the transmitter may further include a scramble unit (not shown), a modulation mapper (not shown), a layer mapper (not shown), and a layer permutator (not shown), which may be arranged before the DFT unit (1031-11). That is, in order to prevent an increase in PAPR (peak-to-average power ratio), the transmitter (1031-1) first passes the information through a DFT (1031-11) before mapping the signal to a subcarrier. The signal spread (or precoded in the same sense) by the DFT unit (1031-11) is mapped to a subcarrier through a subcarrier mapper (1031-12) and then passes through an IFFT (Inverse Fast Fourier Transform) unit (1031-13) to be converted into a signal on the time axis.

[0257] The DFT unit (1031-11) performs DFT on the input symbols and outputs complex-valued symbols. For example, if Ntx symbols are input (where Ntx is a natural number), the DFT size is Ntx. The DFT unit (1031-11) may be called a transform precoder. The subcarrier mapper (1031-12) maps the complex symbols to each subcarrier in the frequency domain. The complex symbols may be mapped to resource elements corresponding to resource blocks allocated for data transmission. The subcarrier mapper (1031-12) may be called a resource element mapper. The IFFT unit (1031-13) performs IFFT on the input symbols and outputs a baseband signal for data, which is a time-domain signal. The CP insertion unit (1031-14) copies a portion of the rear portion of the baseband signal for data and inserts it into the front portion of the baseband signal for data. CP insertion prevents ISI (Inter-Symbol Interference) and ICI (Inter-Carrier Interference), thereby maintaining orthogonality even in multipath channels.

[0258] On the other hand, the receiver (1031-2) includes a wireless reception unit (1031-21), a CP removal unit (1031-22), an FFT unit (1031-23), and an equalization unit (1031-24). The wireless reception unit (1031-21), the CP removal unit (1031-22), and the FFT unit (1031-23) of the receiver (1031-2) perform the inverse functions of the wireless transmission unit (1031-15), the CP insertion unit (1031-14), and the IFF unit (1031-13) of the transmitter (1031-1). The receiver (1031-2) may further include a demodulator.

[0259] Although the preferred embodiments have been described above by way of example, the disclosure of this specification is not limited to these specific embodiments, and may be modified, changed, or improved in various forms within the scope of the spirit and claims of this specification.

[0260] In the exemplary system described above, the methods are described based on a flowchart as a series of steps or blocks. However, the order of the steps described is not limited, and some steps may occur in a different order or simultaneously with other steps described above. Furthermore, those skilled in the art will understand that the steps depicted in the flowchart are not exclusive, and other steps may be included, or one or more steps in the flowchart may be deleted without affecting the scope of the invention.

[0261] The claims set forth in this specification may be combined in various ways. For example, the technical features of the method claims of this specification may be combined and implemented as a device, and the technical features of the device claims of this specification may be combined and implemented as a method. Furthermore, the technical features of the method claims and the technical features of the device claims of this specification may be combined and implemented as a device, and the technical features of the method claims and the technical features of the device claims of this specification may be combined and implemented as a method.

Claims

1. In a method of operating a terminal in a wireless communication system, A step of receiving configuration information for performance monitoring of an AI (artificial intelligence) / ML (machine learning) model, wherein the configuration information includes information on at least one monitoring period and period change condition associated with a reference signal (RS); A step of performing performance monitoring of the AI / ML model based on a first monitoring cycle among at least one of the above monitoring cycles; and A method comprising the step of performing performance monitoring of the AI / ML model based on a second monitoring cycle among the at least one monitoring cycle, based on satisfaction of the above cycle change condition.

2. In paragraph 1, A method wherein the above periodic change condition includes an aperiodic reporting condition of the performance monitoring results of the AI / ML model.

3. In paragraph 2, The above setting information further includes timer information, and a step of starting a timer based on the timer information; A step of transmitting the performance monitoring result of the AI / ML model according to satisfaction of the aperiodic reporting condition of the performance monitoring result of the AI / ML model while the timer is running; and Including a step of restarting the timer according to the transmission of the performance monitoring result of the above AI / ML model, A method wherein the second monitoring period is shorter than the first monitoring period.

4. In paragraph 1, The above setting information further includes timer information, When the timer started based on the above timer information expires, performance monitoring of the AI / ML model based on the second monitoring cycle is performed. A method wherein the second monitoring period is longer than the first monitoring period.

5. In paragraph 2, A step of increasing the COUNT value associated with performance monitoring of the above AI / ML model; After increasing the COUNT value, a step of transmitting the performance monitoring result of the AI / ML model according to satisfaction of the aperiodic reporting condition of the performance monitoring result of the AI / ML model; and Including a step of resetting the COUNT value according to the transmission of the performance monitoring result of the AI / ML model, A method wherein the second monitoring period is shorter than the first monitoring period.

6. In paragraph 1, The above configuration information further includes information on the COUNT MAX value associated with performance monitoring of the AI / ML model, When the increased COUNT value associated with the performance monitoring of the AI / ML model reaches the COUNT MAX value, performance monitoring of the AI / ML model based on the second monitoring cycle is performed, A method wherein the second monitoring period is longer than the first monitoring period.

7. In a method of operating a base station in a wireless communication system, A step of transmitting configuration information for performance monitoring of an AI (artificial intelligence) / ML (machine learning) model, wherein the configuration information includes information on at least one monitoring period and period change condition associated with a reference signal (RS); A step of transmitting the reference signal based on a first monitoring period among at least one of the monitoring periods; and A method comprising the step of transmitting the reference signal based on a second monitoring period among the at least one monitoring period, based on satisfaction of the above period change condition.

8. In paragraph 7, A method wherein the above periodic change condition includes an aperiodic reporting condition of the performance monitoring results of the AI / ML model.

9. In paragraph 8, The above setting information further includes timer information, and a step of starting a timer based on the timer information; A step of receiving a performance monitoring result of the AI / ML model according to satisfaction of an aperiodic reporting condition of the performance monitoring result of the AI / ML model while the timer is running; and Including a step of restarting the timer according to the reception of the performance monitoring result of the AI / ML model, A method wherein the second monitoring period is shorter than the first monitoring period.

10. In paragraph 7, The above setting information further includes timer information, When the timer started based on the above timer information expires, transmission of the reference signal based on the second monitoring period is performed, A method wherein the second monitoring period is longer than the first monitoring period.

11. In paragraph 8, A step of increasing the COUNT value associated with performance monitoring of the above AI / ML model; After increasing the COUNT value, a step of receiving the performance monitoring result of the AI / ML model according to satisfaction of the aperiodic reporting condition of the performance monitoring result of the AI / ML model; and Including a step of resetting the COUNT value according to the reception of the performance monitoring result of the AI / ML model, A method wherein the second monitoring period is shorter than the first monitoring period.

12. In paragraph 7, The above configuration information further includes information on the COUNT MAX value associated with performance monitoring of the AI / ML model, When the increased COUNT value associated with the performance monitoring of the AI / ML model reaches the COUNT MAX value, the reference signal based on the second monitoring cycle is transmitted, A method wherein the second monitoring period is longer than the first monitoring period.

13. As a terminal in a wireless communication system, at least one processor; and At least one memory storing instructions and being operably electrically connected to said at least one processor, wherein the operations performed based on the instructions being executed by said at least one processor are: A step of receiving configuration information for performance monitoring of an AI (artificial intelligence) / ML (machine learning) model, wherein the configuration information includes information on at least one monitoring period and period change condition associated with a reference signal (RS). A step of performing performance monitoring of the AI / ML model based on a first monitoring cycle among at least one of the above monitoring cycles, and A terminal comprising a step of performing performance monitoring of the AI / ML model based on a second monitoring cycle among the at least one monitoring cycle, based on satisfaction of the above cycle change condition.

14. In paragraph 13, The above periodic change condition is a terminal that includes an aperiodic reporting condition of the performance monitoring result of the AI / ML model.

15. In paragraph 14, Based on the above instruction being executed by the at least one processor, the operations performed are: The above setting information further includes timer information, and a step of starting a timer based on the timer information, A step of transmitting the performance monitoring result of the AI / ML model according to satisfaction of the aperiodic reporting condition of the performance monitoring result of the AI / ML model while the timer is running, and Including a step of restarting the timer according to the transmission of the performance monitoring result of the above AI / ML model, The second monitoring period is shorter than the first monitoring period, the terminal.

16. In paragraph 13, The above setting information further includes timer information, When the timer started based on the above timer information expires, performance monitoring of the AI / ML model based on the second monitoring cycle is performed. The second monitoring period is longer than the first monitoring period, the terminal.

17. In paragraph 14, Based on the above instruction being executed by the at least one processor, the operations performed are: A step of increasing the COUNT value associated with the performance monitoring of the above AI / ML model, After increasing the COUNT value, a step of transmitting the performance monitoring result of the AI / ML model according to satisfaction of the aperiodic reporting condition of the performance monitoring result of the AI / ML model, and A step of resetting the COUNT value based on the transmission of the performance monitoring result of the AI / ML model, The second monitoring period is shorter than the first monitoring period, the terminal.

18. In paragraph 13, The above configuration information further includes information on the COUNT MAX value associated with performance monitoring of the AI / ML model, When the increased COUNT value associated with the performance monitoring of the AI / ML model reaches the COUNT MAX value, performance monitoring of the AI / ML model based on the second monitoring cycle is performed, The second monitoring period is longer than the first monitoring period, the terminal.

Images