Method and apparatus for NW-side setting information (NSI) configuration and indication in wireless communication networks

Abstract

The disclosure relates to a 5G or 6G communication system for supporting a higher data transmission rate. A method performed by a user equipment (UE) in a communication system is provided. The method comprises receiving, from a base station, first information for training data set, wherein the first information includes identification (ID) for a network-side setting; obtaining training data set based on the information; training an artificial intelligence (AI) / machine learning (ML) model based on the network-side setting using the training data set; receiving, from the base station, second information for inference, wherein the second information includes the ID for the network-side setting; performing an inference using the AI / ML model associated with the network-side setting; and transmitting, to the base station, a result of the inference.

Description

METHOD AND APPARATUS FOR NW-SIDE SETTING INFORMATION (NSI) CONFIGURATION AND INDICATION IN WIRELESS COMMUNICATION NETWORKS

[0001] The present disclosure relates to the field of 5G and beyond 5G communication networks and more particularly to life cycle management of AI / ML models in wireless communication networks.

[0002] 5G mobile communication technologies define broad frequency bands such that high transmission rates and new services are possible, and can be implemented not only in "Sub 6GHz" bands such as 3.5GHz, but also in "Above 6GHz" bands referred to as mmWave including 28GHz and 39GHz. In addition, it has been considered to implement 6G mobile communication technologies (referred to as Beyond 5G systems) in terahertz bands (for example, 95GHz to 3THz bands) in order to accomplish transmission rates fifty times faster than 5G mobile communication technologies and ultra-low latencies one-tenth of 5G mobile communication technologies.

[0003] At the beginning of the development of 5G mobile communication technologies, in order to support services and to satisfy performance requirements in connection with enhanced Mobile BroadBand (eMBB), Ultra Reliable Low Latency Communications (URLLC), and massive Machine-Type Communications (mMTC), there has been ongoing standardization regarding beamforming and massive MIMO for mitigating radio-wave path loss and increasing radio-wave transmission distances in mmWave, supporting numerologies (for example, operating multiple subcarrier spacings) for efficiently utilizing mmWave resources and dynamic operation of slot formats, initial access technologies for supporting multi-beam transmission and broadbands, definition and operation of BWP (BandWidth Part), new channel coding methods such as a LDPC (Low Density Parity Check) code for large amount of data transmission and a polar code for highly reliable transmission of control information, L2 pre-processing, and network slicing for providing a dedicated network specialized to a specific service.

[0004] Currently, there are ongoing discussions regarding improvement and performance enhancement of initial 5G mobile communication technologies in view of services to be supported by 5G mobile communication technologies, and there has been physical layer standardization regarding technologies such as V2X (Vehicle-to-everything) for aiding driving determination by autonomous vehicles based on information regarding positions and states of vehicles transmitted by the vehicles and for enhancing user convenience, NR-U (New Radio Unlicensed) aimed at system operations conforming to various regulation-related requirements in unlicensed bands, NR UE Power Saving, Non-Terrestrial Network (NTN) which is UE-satellite direct communication for providing coverage in an area in which communication with terrestrial networks is unavailable, and positioning.

[0005] Moreover, there has been ongoing standardization in air interface architecture / protocol regarding technologies such as Industrial Internet of Things (IIoT) for supporting new services through interworking and convergence with other industries, IAB (Integrated Access and Backhaul) for providing a node for network service area expansion by supporting a wireless backhaul link and an access link in an integrated manner, mobility enhancement including conditional handover and DAPS (Dual Active Protocol Stack) handover, and two-step random access for simplifying random access procedures (2-step RACH for NR). There also has been ongoing standardization in system architecture / service regarding a 5G baseline architecture (for example, service based architecture or service based interface) for combining Network Functions Virtualization (NFV) and Software-Defined Networking (SDN) technologies, and Mobile Edge Computing (MEC) for receiving services based on UE positions.

[0006] As 5G mobile communication systems are commercialized, connected devices that have been exponentially increasing will be connected to communication networks, and it is accordingly expected that enhanced functions and performances of 5G mobile communication systems and integrated operations of connected devices will be necessary. To this end, new research is scheduled in connection with eXtended Reality (XR) for efficiently supporting AR (Augmented Reality), VR (Virtual Reality), MR (Mixed Reality) and the like, 5G performance improvement and complexity reduction by utilizing Artificial Intelligence (AI) and Machine Learning (ML), AI service support, metaverse service support, and drone communication.

[0007] Furthermore, such development of 5G mobile communication systems will serve as a basis for developing not only new waveforms for providing coverage in terahertz bands of 6G mobile communication technologies, multi-antenna transmission technologies such as Full Dimensional MIMO (FD-MIMO), array antennas and large-scale antennas, metamaterial-based lenses and antennas for improving coverage of terahertz band signals, high-dimensional space multiplexing technology using OAM (Orbital Angular Momentum), and RIS (Reconfigurable Intelligent Surface), but also full-duplex technology for increasing frequency efficiency of 6G mobile communication technologies and improving system networks, AI-based communication technology for implementing system optimization by utilizing satellites and AI (Artificial Intelligence) from the design stage and internalizing end-to-end AI support functions, and next-generation distributed computing technology for implementing services at levels of complexity exceeding the limit of UE operation capability by utilizing ultra-high-performance communication and computing resources.

[0008] The above information is presented as background information only to assist with an understanding of the disclosure. No determination has been made, and no assertion is made, as to whether any of the above might be applicable as prior art with regard to the disclosure.

[0009] The principal object of the disclosure herein is to disclose methods and apparatus for the network to configure and indicate network-side setting information for AI / ML operations in wireless communication networks. For the AI / ML models, which are fully or partly deployed at a terminal, based on the various embodiments of this disclosure, the network may assist AI / ML operations management by provisioning alignment of network-side setting information.

[0010] As specific object of the disclosure herein is to disclose methods and systems for a UE to reports its capability by including information pertaining to the available network-side setting information indicated by the network.

[0011] As a yet another specific object of the disclosure herein is to disclose methods and systems for the gNB to receive AI / ML related capability reports from the UE and to configure the UE with AI / ML operations accordingly.

[0012] As a yet another specific object of the disclosure herein is to disclose methods and systems for a network to indicate the network-side setting information for training dataset collection.

[0013] As a yet another specific object of the disclosure herein is to disclose methods and systems for the UE to receive configuration or indication from the network pertaining to the network-side setting information for training dataset collection and categorize the training dataset accordingly.

[0014] As a yet another specific object of the disclosure herein is to disclose methods and systems for a network to indicate the network-side setting information for AI / ML operations at the UE including AI / ML model inference and monitoring.

[0015] As a yet another specific object of the disclosure herein is to disclose methods and systems for the UE to receive configuration or indication from the network pertaining to the network-side setting information for AI / ML operations at the UE including AI / ML model inference and monitoring. The UE then may select the appropriate model compatible to the configured or indicated network-side setting information.

[0016] The present disclosure has been made to address the above-mentioned problems and disadvantages, and to provide at least the advantages described below.

[0017] In accordance with an aspect of the present disclosure, a method performed by a user equipment (UE) in a communication system is provided. The method comprises receiving, from a network node, first information for training data set, wherein the first information includes identification (ID) for a network-side setting; obtaining training data set based on the information; training an artificial intelligence (AI) / machine learning (ML) model based on the network-side setting using the training data set; receiving, from the network node, second information for inference, wherein the second information includes the ID for the network-side setting; performing an inference using the AI / ML model associated with the network-side setting; and transmitting, to the network node, a result of the inference.

[0018] In accordance with an aspect of the present disclosure, a method performed by a base station in a communication system is provided. The method comprises transmitting, to a user equipment (UE), first information for training data set used for training an artificial intelligence (AI) / machine learning (ML) model based on a network-side setting, wherein the first information includes identification (ID) for a network-side setting; transmitting, to the UE, second information for inference, wherein the second information includes the ID for the network-side setting; and receiving, from the UE, a result of an inference performed using the AI / ML model associated with the network-side setting.

[0019] In accordance with an aspect of the present disclosure, a user equipment (UE) in a communication system is provided. The UE comprises a transceiver; and at least one processor configured to receive, from a base station, first information for training data set, wherein the first information includes identification (ID) for a network-side setting, obtain training data set based on the information, train an artificial intelligence (AI) / machine learning (ML) model based on the network-side setting using the training data set, receive, from the base station, second information for inference, wherein the second information includes the ID for the network-side setting, perform an inference using the AI / ML model associated with the network-side setting, and transmit, to the base station, a result of the inference.

[0020] In accordance with an aspect of the present disclosure, a base station in a communication system is provided. The base station comprises a transceiver; and at least one processor configured to transmit, to a user equipment (UE), first information for training data set used for training an artificial intelligence (AI) / machine learning (ML) model based on a network-side setting, wherein the first information includes identification (ID) for a network-side setting, transmit, to the UE, second information for inference, wherein the second information includes the ID for the network-side setting, and receive, from the UE, a result of an inference performed using the AI / ML model associated with the network-side setting.

[0021] The present disclosure provides methods and apparatus for the network to configure and indicate network-side setting information for AI / ML operations in wireless communication networks.

[0022] Embodiments herein are illustrated in the accompanying drawings, throughout which like reference letters indicate corresponding parts in the various figures. The embodiments herein will be better understood from the following description with reference to the drawings, in which:

[0023] FIGURE 1 illustrates an example wireless network according to embodiments of the present disclosure;

[0024] FIGURE 2a illustrates an example wireless transmit path according to embodiments of the present disclosure;

[0025] FIGURE 2b illustrates an example wireless receive path according to embodiments of the present disclosure;

[0026] FIGURE 3a illustrates an example UE according to embodiments of the present disclosure;

[0027] FIGURE 3b illustrates an example gNB according to embodiments of the present disclosure;

[0028] FIGURE4 illustrates exemplary cross-polarized MIMO antenna system according to embodiments of the present disclosure;

[0029] FIGURE 5 illustrates exemplary layout for channel state information reference signal (CSI-RS) resource mapping in an orthogonal frequency division multiple access (OFDM) time-frequency grid according to embodiments of the present disclosure;

[0030] FIGURE 6 illustrates an example of CSI report configurations and CSI measurement configurations according to embodiments of the present disclosure;

[0031] FIGURE 7 illustrates single-sided and two-sided models according to embodiments of the present disclosure;

[0032] FIGURE 8 illustrates two-side model based CSI compression according to embodiments of the present disclosure;

[0033] FIGURE 9 illustrates AI / ML based spatial-domain beam prediction according to embodiments of the present disclosure;

[0034] FIGURE 10 illustrates AI / ML based temporal-domain beam prediction according to embodiments of the present disclosure;

[0035] FIGURE 11 illustrates exemplary CSI report configuration for predictive beam reporting according to embodiments of the present disclosure;

[0036] FIGURE 12 illustrates exemplary cases for a network-side setting information varies in AI / ML based beam prediction according to embodiments of the present disclosure;

[0037] FIGURE 13 illustrates exemplary two-sided model training with training dataset provision from the network according to embodiments of the present disclosure;

[0038] FIGURE 14 illustrates alignment between training dataset collection and inference in terms of network-side setting information according to embodiments of the present disclosure;

[0039] FIGURE 15 illustrates exemplary procedure for UE capability signaling and inference operation according to embodiments of the present disclosure;

[0040] FIGURE 16 illustrates exemplary procedure for UE capability update according to embodiments of the present disclosure;

[0041] FIGURE 17 illustrates linkage between configuration for training / fine-tuning dataset and AI / ML inference according to embodiments of the present disclosure; and

[0042] FIGURE 18 illustrates exemplary categorization of configuration for AI / MLaccording to embodiments of the present disclosure.

[0043] Wireless communication has been one of the most successful innovations in modern history. Recently, the number of subscribers to wireless communication services exceeded five billion and continues to grow quickly. The demand of wireless data traffic is rapidly increasing due to the growing popularity among consumers and businesses of smart phones and other mobile data devices, such as tablets, "note pad" computers, net books, eBook readers, and machine type of devices. In order to meet the high growth in mobile data traffic and support new applications and deployments, improvements in radio interface efficiency and coverage is of paramount importance.

[0044] To meet the demand for wireless data traffic having increased since deployment of 4G communication systems, and to enable various vertical applications, 5G communication systems have been developed and are currently being deployed.

[0045] The 5G communication system is considered to be implemented to include higher frequency (mmWave) bands, such as 28 GHz or 60 GHz bands or, in general, above 6 GHz bands, so as to accomplish higher data rates, or in lower frequency bands, such as below 6 GHz, to enable robust coverage and mobility support. Aspects of the present disclosure may be applied to deployment of 5G communication systems, 6G or even later releases which may use THz bands. 　To decrease propagation loss of the radio waves and increase the transmission distance, the beamforming, massive multiple-input multiple-output (MIMO), Full Dimensional MIMO (FD-MIMO), array antenna, an analog beam forming, large-scale antenna techniques are discussed in 5G communication systems.

[0046] In addition, in 5G communication systems, development for system network improvement is under way based on advanced small cells, cloud Radio Access Networks (RANs), ultra-dense networks, device-to-device (D2D) communication, wireless backhaul, moving network, cooperative communication, Coordinated Multi-Points (CoMP), reception-end interference cancellation and the like.

[0047] FIGURE.1 illustrates an example wireless network 100 according to this disclosure. The embodiment of the wireless network 100 shown in FIGURE.1 is for illustration only. Other embodiments of the wireless network 100 can be used without departing from the scope of this disclosure.

[0048] The wireless network 100 includes an gNodeB (gNB) 101, an gNB 102, and an gNB 103. The gNB 101 communicates with the gNB 102 and the gNB 103. The gNB 101 also communicates with at least one Internet Protocol (IP) network 130, such as the Internet, a proprietary IP network, or other data network.

[0049] Depending on the network type, the term 'gNB' can refer to any component (or collection of components) configured to provide remote terminals with wireless access to a network, such as base transceiver station, a radio base station, transmit point (TP), transmit-receive point (TRP), a ground gateway, an airborne gNB, a satellite system, mobile base station, a macrocell, a femtocell, a WiFi access point (AP) and the like. Also, depending on the network type, other well-known terms may be used instead of "user equipment" or "UE," such as "mobile station," "subscriber station," "remote terminal," "wireless terminal," or "user device." For the sake of convenience, the terms "user equipment" and "UE" are used in this patent document to refer to equipment that wirelessly accesses a gNB. The UE could be a mobile device or a stationary device. For example, UE could be a mobile telephone, smartphone, monitoring device, alarm device, fleet management device, asset tracking device, automobile, desktop computer, entertainment device, infotainment device, vending machine, electricity meter, water meter, gas meter, security device, sensor device, appliance etc

[0050] The gNB 102 provides wireless broadband access to the network 130 for a first plurality of user equipments (UEs) within a coverage area 120 of the gNB 102. The first plurality of UEs includes a UE 111, which may be located in a small business (SB); a UE 112, which may be located in an enterprise (E); a UE 113, which may be located in a WiFi hotspot (HS); a UE 114, which may be located in a first residence (R); a UE 115, which may be located in a second residence (R); and a UE 116, which may be a mobile device (M) like a cell phone, a wireless laptop, a wireless PDA, or the like. The gNB 103 provides wireless broadband access to the network 130 for a second plurality of UEs within a coverage area 125 of the gNB 103. The second plurality of UEs includes the UE 115 and the UE 116. In some embodiments, one or more of the gNBs 101-103 may communicate with each other and with the UEs 111-116 using 5G, long-term evolution (LTE), LTE-A, WiMAX, or other advanced wireless communication techniques.

[0051] Dotted lines show the approximate extents of the coverage areas 120 and 125, which are shown as approximately circular for the purposes of illustration and explanation only. It should be clearly understood that the coverage areas associated with gNBs, such as the coverage areas 120 and 125, may have other shapes, including irregular shapes, depending upon the configuration of the gNBs and variations in the radio environment associated with natural and man-made obstructions.

[0052] As described in more detail below, one or more of BS　101, BS　102 and BS　103 include 2D antenna arrays as described in embodiments of the present disclosure. In some embodiments, one or more of BS　101, BS　102 and BS　103 support the codebook design and structure for systems having 2D antenna arrays.

[0053] Although FIGURE.1 illustrates one example of a wireless network 100, various changes may be made to FIGURE.1. For example, the wireless network 100 can include any number of gNBs and any number of UEs in any suitable arrangement. Also, the gNB 101 can communicate directly with any number of UEs and provide those UEs with wireless broadband access to the network 130. Similarly, each gNB 102-103 can communicate directly with the network 130 and provide UEs with direct wireless broadband access to the network 130. Further, the gNB 101, 102, and / or 103 can provide access to other or additional external networks, such as external telephone networks or other types of data networks.

[0054] Figures 2a and 2b illustrate example wireless transmit and receive paths according to this disclosure. In the following description, a transmit path 200 may be described as being implemented in an gNB (such as gNB 102), while a receive path 250 may be described as being implemented in a UE (such as UE 116). However, it will be understood that the receive path 250 can be implemented in an gNB and that the transmit path 200 can be implemented in a UE. In some embodiments, the receive path 250 is configured to support the codebook design and structure for systems having 2D antenna arrays as described in embodiments of the present disclosure.

[0055] The transmit path 200 includes a channel coding and modulation block 205, a serial-to-parallel (S-to-P) block 210, a size N Inverse Fast Fourier Transform (IFFT) block 215, a parallel-to-serial (P-to-S) block 220, an add cyclic prefix block 225, and an up-converter (UC) 230. The receive path 250 includes a down-converter (DC) 255, a remove cyclic prefix block 260, a serial-to-parallel (S-to-P) block 265, a size N Fast Fourier Transform (FFT) block 270, a parallel-to-serial (P-to-S) block 275, and a channel decoding and demodulation block 280.

[0056] In the transmit path 200, the channel coding and modulation block 205 receives a set of information bits, applies coding (such as a low-density parity check (LDPC) coding), and modulates the input bits (such as with Quadrature Phase Shift Keying (QPSK) or Quadrature Amplitude Modulation (QAM)) to generate a sequence of frequency-domain modulation symbols. The serial-to-parallel block 210 converts (such as de-multiplexes) the serial modulated symbols to parallel data in order to generate N parallel symbol streams, where N is the IFFT / FFT size used in the gNB 102 and the UE 116. The size N IFFT block 215 performs an IFFT operation on the N parallel symbol streams to generate time-domain output signals. The parallel-to-serial block 220 converts (such as multiplexes) the parallel time-domain output symbols from the size N IFFT block 215 in order to generate a serial time-domain signal. The add cyclic prefix block 225 inserts a cyclic prefix to the time-domain signal. The up-converter 230 modulates (such as up-converts) the output of the add cyclic prefix block 225 to an RF frequency for transmission via a wireless channel. The signal may also be filtered at baseband before conversion to the RF frequency.

[0057] A transmitted RF signal from the gNB 102 arrives at the UE 116 after passing through the wireless channel, and reverse operations to those at the gNB 102 are performed at the UE 116. The down-converter 255 down-converts the received signal to a baseband frequency, and the remove cyclic prefix block 260 removes the cyclic prefix to generate a serial time-domain baseband signal. The serial-to-parallel block 265 converts the time-domain baseband signal to parallel time domain signals. The size N FFT block 270 performs an FFT algorithm to generate N parallel frequency-domain signals. The parallel-to-serial block 275 converts the parallel frequency-domain signals to a sequence of modulated data symbols. The channel decoding and demodulation block 280 demodulates and decodes the modulated symbols to recover the original input data stream.

[0058] Each of the gNBs 101-103 may implement a transmit path 200 that is analogous to transmitting in the downlink to UEs 111-116 and may implement a receive path 250 that is analogous to receiving in the uplink from UEs 111-116. Similarly, each of UEs 111-116 may implement a transmit path 200 for transmitting in the uplink to gNBs 101-103 and may implement a receive path 250 for receiving in the downlink from gNBs 101-103.

[0059] Each of the components in Figures 2a and 2b can be implemented using only hardware or using a combination of hardware and software / firmware. As a particular example, at least some of the components in Figures 2a and 2b may be implemented in software, while other components may be implemented by configurable hardware or a mixture of software and configurable hardware. For instance, the FFT block 270 and the IFFT block 215 may be implemented as configurable software algorithms, where the value of size N may be modified according to the implementation.

[0060] Furthermore, although described as using FFT and IFFT, this is by way of illustration only and should not be construed to limit the scope of this disclosure. Other types of transforms, such as Discrete Fourier Transform (DFT) and Inverse Discrete Fourier Transform (IDFT) functions, can be used. It will be appreciated that the value of the variable N may be any integer number (such as 1, 2, 3, 4, or the like) for DFT and IDFT functions, while the value of the variable N may be any integer number that is a power of two (such as 1, 2, 4, 8, 16, or the like) for FFT and IFFT functions.

[0061] Although Figures 2a and 2b illustrate examples of wireless transmit and receive paths, various changes may be made to Figures 2a and 2b. For example, various components in Figures 2a and 2b can be combined, further subdivided, or omitted and additional components can be added according to particular needs. Also, Figures 2a and 2b are meant to illustrate examples of the types of transmit and receive paths that can be used in a wireless network. Any other suitable architectures can be used to support wireless communications in a wireless network.

[0062] Figure 3a illustrates an example UE 116 according to this disclosure. The embodiment of the UE 116 illustrated in Figure 3a is for illustration only, and the UEs 111-115 of Figure 1 can have the same or similar configuration. However, UEs come in a wide variety of configurations, and Figures 3a does not limit the scope of this disclosure to any particular implementation of a UE.

[0063] The UE　116 includes an antenna 305, a radio frequency (RF) transceiver 310, transmit (TX) processing circuitry 315, a microphone 320, and receive (RX) processing circuitry 325. The UE　116 also includes a speaker 330, a main processor 340, an input / output (I / O) interface (IF) 345, a keypad 350, a display 355, and a memory 360. The memory 360 includes a basic operating system (OS) program 361 and one or more applications 362.

[0064] The RF transceiver 310 receives, from the antenna 305, an incoming RF signal transmitted by an gNB of the network 100. The RF transceiver 310 down-converts the incoming RF signal to generate an intermediate frequency (IF) or baseband signal. The IF or baseband signal is sent to the RX processing circuitry 325, which generates a processed baseband signal by filtering, decoding, and / or digitizing the baseband or IF signal. The RX processing circuitry 325 transmits the processed baseband signal to the speaker 330 (such as for voice data) or to the main processor 340 for further processing (such as for web browsing data).

[0065] The TX processing circuitry 315 receives analog or digital voice data from the microphone 320 or other outgoing baseband data (such as web data, e-mail, or interactive video game data) from the main processor 340. The TX processing circuitry 315 encodes, multiplexes, and / or digitizes the outgoing baseband data to generate a processed baseband or IF signal. The RF transceiver 310 receives the outgoing processed baseband or IF signal from the TX processing circuitry 315 and up-converts the baseband or IF signal to an RF signal that is transmitted via the antenna 305.

[0066] The main processor 340 can include one or more processors or other processing devices and execute the basic OS program 361 stored in the memory 360 in order to control the overall operation of the UE 116. For example, the main processor 340 can control the reception of forward channel signals and the transmission of reverse channel signals by the RF transceiver 310, the RX processing circuitry 325, and the TX processing circuitry 315 in accordance with well-known principles. In some embodiments, the main processor 340 includes at least one microprocessor or microcontroller.

[0067] The main processor 340 is also capable of executing other processes and programs resident in the memory 360, such as operations for channel quality measurement and reporting for systems having 2D antenna arrays as described in embodiments of the present disclosure as described in embodiments of the present disclosure. The main processor 340 can move data into or out of the memory 360 as required by an executing process. In some embodiments, the main processor 340 is configured to execute the applications 362 based on the OS program 361 or in response to signals received from gNBs or an operator. The main processor 340 is also coupled to the I / O interface 345, which provides the UE 116 with the ability to connect to other devices such as laptop computers and handheld computers. The I / O interface 345 is the communication path between these accessories and the main controller 340.

[0068] The main processor 340 is also coupled to the keypad 350 and the display unit 355. The operator of the UE 116 can use the keypad 350 to enter data into the UE 116. The display 355 may be a liquid crystal display or other display capable of rendering text and / or at least limited graphics, such as from web sites. The memory 360 is coupled to the main processor 340. Part of the memory 360 can include a random access memory (RAM), and another part of the memory 360 can include a Flash memory or other read-only memory (ROM).

[0069] Although Figures 3a illustrates one example of UE 116, various changes may be made to Figures 3a. For example, various components in Figures 3a can be combined, further subdivided, or omitted and additional components can be added according to particular needs. As a particular example, the main processor 340 can be divided into multiple processors, such as one or more central processing units (CPUs) and one or more graphics processing units (GPUs). Also, while Figures 3a illustrates the UE 116 configured as a mobile telephone or smartphone, UEs can be configured to operate as other types of mobile or stationary devices.

[0070] Figures 3b illustrates an example gNB 102 according to this disclosure. The embodiment of the gNB 102 shown in Figures 3b is for illustration only, and other gNBs of Figure 1 can have the same or similar configuration. However, gNBs come in a wide variety of configurations, and Figures 3b does not limit the scope of this disclosure to any particular implementation of an gNB. It is noted that gNB　101 and gNB　103 can include the same or similar structure as gNB　102.

[0071] As shown in Figures 3b, the gNB 102 includes multiple antennas 370a-370n, multiple RF transceivers 372a-372n, transmit (TX) processing circuitry 374, and receive (RX) processing circuitry 376. In certain embodiments, one or more of the multiple antennas 370a-370n include 2D antenna arrays. The gNB 102 also includes a controller / processor 378, a memory 380, and a backhaul or network interface 382.

[0072] The RF transceivers 372a-372n receive, from the antennas 370a-370n, incoming RF signals, such as signals transmitted by UEs or other gNBs. The RF transceivers 372a-372n down-convert the incoming RF signals to generate IF or baseband signals. The IF or baseband signals are sent to the RX processing circuitry 376, which generates processed baseband signals by filtering, decoding, and / or digitizing the baseband or IF signals. The RX processing circuitry 376 transmits the processed baseband signals to the controller / processor 378 for further processing.

[0073] The TX processing circuitry 374 receives analog or digital data (such as voice data, web data, e-mail, or interactive video game data) from the controller / processor 378. The TX processing circuitry 374 encodes, multiplexes, and / or digitizes the outgoing baseband data to generate processed baseband or IF signals. The RF transceivers 372a-372n receive the outgoing processed baseband or IF signals from the TX processing circuitry 374 and up-converts the baseband or IF signals to RF signals that are transmitted via the antennas 370a-370n.

[0074] The controller / processor 378 can include one or more processors or other processing devices that control the overall operation of the gNB 102. For example, the controller / processor 378 can control the reception of forward channel signals and the transmission of reverse channel signals by the RF transceivers 372a-372n, the RX processing circuitry 376, and the TX processing circuitry 374 in accordance with well-known principles. The controller / processor 378 can support additional functions as well, such as more advanced wireless communication functions. For instance, the controller / processor 378 can perform the blind interference sensing (BIS) process, such as performed by a BIS algorithm, and decodes the received signal subtracted by the interfering signals. Any of a wide variety of other functions can be supported in the gNB 102 by the controller / processor 378. In some embodiments, the controller / processor 378 includes at least one microprocessor or microcontroller.

[0075] The controller / processor 378 is also capable of executing programs and other processes resident in the memory 380, such as a basic OS. The controller / processor 378 is also capable of supporting channel quality measurement and reporting for systems having 2D antenna arrays as described in embodiments of the present disclosure. In some embodiments, the controller / processor 378 supports communications between entities, such as web RTC. The controller / processor 378 can move data into or out of the memory 380 as required by an executing process.

[0076] The controller / processor 378 is also coupled to the backhaul or network interface 382. The backhaul or network interface 382 allows the gNB 102 to communicate with other devices or systems over a backhaul connection or over a network. The interface 382 can support communications over any suitable wired or wireless connection(s). For example, when the gNB 102 is implemented as part of a cellular communication system (such as one supporting 5G, LTE, or LTE-A), the interface 382 can allow the gNB 102 to communicate with other gNBs over a wired or wireless backhaul connection. When the gNB 102 is implemented as an access point, the interface 382 can allow the gNB 102 to communicate over a wired or wireless local area network or over a wired or wireless connection to a larger network (such as the Internet). The interface 382 includes any suitable structure supporting communications over a wired or wireless connection, such as an Ethernet or RF transceiver.

[0077] The memory 380 is coupled to the controller / processor 378. Part of the memory 380 can include a RAM, and another part of the memory 380 can include a Flash memory or other ROM. In certain embodiments, a plurality of instructions, such as a BIS algorithm is stored in memory. The plurality of instructions are configured to cause the controller / processor 378 to perform the BIS process and to decode a received signal after subtracting out at least one interfering signal determined by the BIS algorithm.

[0078] As described in more detail below, the transmit and receive paths of the gNB 102 (implemented using the RF transceivers 372a-372n, TX processing circuitry 374, and / or RX processing circuitry 376) support communication with aggregation of FDD cells and TDD cells.

[0079] Although Figures 3b illustrates one example of a gNB 102, various changes may be made to Figures 3b. For example, the gNB 102 can include any number of each component shown in Figures 3b. As a particular example, an access point can include a number of interfaces 382, and the controller / processor 378 can support routing functions to route data between different network addresses. As another particular example, while shown as including a single instance of TX processing circuitry 374 and a single instance of RX processing circuitry 376, the gNB 102 can include multiple instances of each (such as one per RF transceiver).

[0080] Multiple input multiple output (MIMO) system wherein a BS and / or a UE is equipped with multiple antennas has been widely employed in wireless systems for its advantages in terms of spatial multiplexing, diversity gain and array gain.

[0081] FIGURE 4 illustrates exemplary cross-polarized MIMO antenna system.

[0082] FIGURE 4 illustrates an example of MIMO antenna configuration with 24 antenna elements. In the figure 4, 4 cross-polarized (401) antenna elements form a 4x1 subarray (402). 12 subarrays form a 2V3H MIMO antennas configuration consisting 2 subarrys (404) in vertical dimensions and 3 subarrays (403) in horizontal dimensions. Although FIGURE 4 illustrates one example of MIMO antenna configuration, the disclosed invention can be applied to various such configurations.

[0083] In MIMO systems, the channel state information (CSI) is required at the base station (BS) so that a signal from the BS is received at the UE with maximum possible received power and minimum possible interference. The acquisition of CSI at the BS can be via a measurement at the BS from an UL reference signal or via a measurement and feedback by the UE from a DL reference signal for time-domain duplexing (TDD) and frequency-domain duplexing (FDD) systems, respectively. In 5G FDD systems, the channel state information reference signal (CSI-RS) is the primary reference signal that is used by the UE to measure and report CSI.

[0084] Figure 5 illustrates exemplary layout for channel state information reference signal (CSI-RS) resource mapping in an orthogonal frequency division multiple access (OFDM) time-frequency grid.

[0085] In some embodiments, a UE may receive a configuration signaling from a BS for a CSI-RS that can be used for channel measurement. An example of such configuration is illustrated in FIGURE 5. In the FIGURE 5, 12 antenna ports (CSI-RS ports) are mapped to a CSI-RS with 3 code-domain multiplexing (CDM) groups, wherein each CDM group is mapped to 4 resource elements (REs) in OFDM time-frequency grid. The antenna ports that are mapped to the same CDM group can be orthogonalized in code-domain by employing orthogonal cover codes. The CSI-RS configuration in FIGURE 5 can be related to the MIMO antenna configuration in FIGURE 4, by mapping a CSI-RS port to one of the polarization of a subarray. In the 5G NR standards, three time-domain CSI-RS resources configurations, namely: periodic, semi-persistent and aperiodic are possible. In the figure, an illustrative example of periodic configuration is given with a period of 4 slots.

[0086] FIGURE 6 illustrates an example of CSI report configurations and CSI measurement configurations.

[0087] Moreover a UE can be configured to measure a CSI feedback with a CSI report configuration. A CSI report configuration can be periodic, semi-persistent or aperiodic manner. Figure 6 depicts the CSI report configuration and CSI measurement configurations that is supported in 5G NR system. A CSI report configuration (602) can be linked to a CSI resource configuration (603). The CSI resource configuration (602) may contain one or more CSI resource sets (604) for channel measurement (CMR) or inference measurement (IMR).

[0088] In the case of periodic (P) and semi-persistent (SP) CSI report setting, the CSI resource configuration contains a single CSI resource set. In case of aperiodic (AP) CSI report, a UE can be configured with multiple CSI report triggering states (600). Each CSI report triggering state is associated with one or more CSI associated report configuration information (601). A downlink control information (DCI) may include CSI request which indicates one of the configured triggering states. Moreover, the DCI with CSI request may also contain a resource set selection field (605) to select one of the resources sets (604).

[0089] Moreover, a CSI report can be configured with one of the CSI reporting quantities. This may include CSI resource indicator (CRI), the rank indicator (RI), precoding matrix indicator (PMI) , channel quality indicator (CQI), layer indicator (LI), SINR, RSRP. In 5G NR, various CSI reporting quantiles are adopted. In particular, an RRC parameter reportQuantity set to either 'none', 'cri-RI-PMI-CQI ', 'cri-RI-i1', 'cri-RI-i1-CQI', 'cri-RI-CQI', 'cri-RSRP', 'cri-SINR', 'ssb-Index-RSRP', 'ssb-Index-SINR', 'cri-RI-LI-PMI-CQI', 'cri-RSRP- Index', 'ssb-Index-RSRP- Index', 'cri-SINR- Index' or 'ssb-Index-SINR- Index'.

[0090] The CSI reporting can be used for transmission beam management (BM), specifically, in higher frequency bands, e.g., in frequency range 2 (FR2). In this case, the gNB may configure the UE to report one of the following quantities including, 'cri-RSRP', 'cri-SINR', 'ssb-Index-RSRP', 'ssb-Index-SINR', 'cri-RSRP- Index', 'ssb-Index-RSRP- Index', 'cri-SINR- Index' or 'ssb-Index-SINR- Index'.

[0091] For a yet another purpose, the CSI report can be used for the downlink transmission CSI including 'cri-RI-PMI-CQI ', 'cri-RI-i1', 'cri-RI-i1-CQI', 'cri-RI-CQI'.

[0092] Recently, data-driven algorithms, also known as artificial-intelligence or machine-learning (AI / ML), have gained considerable attentions. Main application areas include solving non-linear optimization problems that cannot be directly solved by convention solutions. Use cases that have recently been highlighted include CSI compression, CSI prediction, beam prediction, positioning, channel estimation and interpolation, MU-MIMO scheduling, etc.

[0093] FIGURE 7 illustrates single-sided and two-sided models according to embodiments of the present disclosure.

[0094] In this disclosure, any data-driven algorithm and its parts are referred as AI / ML model. As shown in the FIGURE 7, such AI / ML model can be located at the UE (701), network (702) or at both UE and network (703). An AI / ML model may require to be trained with a training dataset before it is used for inference (to produce a set of prediction output from set of inputs). When the AI / ML model inference is performed in one node, i.e., UE or network node, such model is referred as single-sided model or, in particular, UE-sided model or network-sided model. When the inference is jointly made at two nodes, i.e., the first part of the inference at UE and the second part at the network, the AI / ML model is said to be two-sided model.

[0095] A description of example embodiments is provided on the following pages.

[0096] The text and figures are provided solely as examples to aid the reader in understanding the invention. They are not intended and are not to be construed as limiting the scope of this invention in any manner. Although certain embodiments and examples have been provided, it will be apparent to those skilled in the art based on the disclosures herein that changes in the embodiments and examples shown may be made without departing from the scope of this invention.

[0097] The below flowcharts illustrate example methods that can be implemented in accordance with the principles of the present disclosure and various changes could be made to the methods illustrated in the flowcharts herein.　 For example, while shown as a series of steps, various steps in each figure could overlap, occur in parallel, occur in a different order, or occur multiple times. In another example, steps may be omitted or replaced by other steps.

[0098] In the below various mechanisms for full channel matrices, i.e., full CSI, reporting are provided.

[0099] In the below detailed description of the disclosure, the terms "AI / ML model", "model" "AI model" are used interchangeably to refer to a data-driven algorithm that takes a certain set of inputs and produces a certain set of outputs . An AI / ML model may require to be trained with a training dataset before it is used for inference (to produce a set of prediction output from set of inputs).

[0100] The AI / ML model can be neural network (NN)-based which is composed of a large number of interconnected neurons. The neurons can be described by parameters which may consist of weights and biases. The interconnection between neural networks may have structure. A typical form of structure is assortments of neurons into multiple layers. If the number of layers in AI / ML model is relatively large, the model can be referred as deep neural network (DNN). Then, the layers could be interconnected with dense or sparse connections.

[0101] The AI / ML model can take various backbone structures, e.g., dense neural networks (DNN) convolutional neural network (CNN), Long-short term memory (LSTM), transformer (TF), etc.

[0102] An AI / ML model can be scenario-specific or configuration-specific, i.e., it provides the desired performance only in a set of scenarios or set of configurations. These models are typically trained by a dataset collected from a certain set of scenarios and configurations. For example, an AI / ML CSI compression model may perform as desired only when it is applied to a set of CSI ports (antenna ports) configuration or CSI payload size configurations. In another exemplary case, an AI / ML CSI compression model may work only under certain set of scenarios, e.g., UE speed.

[0103] In some embodiments, the UE or network may have to keep multiple scenario / configurations specific AI / ML models for different sets of scenarios or configurations. Thus, when a certain set of configurations is applied or a certain scenarios is detected, the UE or network may select the appropriate model, i.e., model selection.

[0104] In some embodiments, the UE or network may have to activate the appropriate AI / ML model for inference. This activation process may require the UE or network to load the model to the processing unit, e.g., central processing unit (CPU), graphical processing unit (GPU), neural processing unit (NPU), etc.

[0105] In some embodiments, the UE or network may have to deactivate an AI / ML model. This deactivation process may include unloading the model from the processing unit (freeing up the processing unit), e.g., central processing unit (CPU), graphical processing unit (GPU), and neural processing unit (NPU).

[0106] In some embodiments, the UE or network may have to switch through AI / ML models depending on the scenarios and configurations. The switching process may include deactivation, selection and activation of AI / ML models.

[0107] In some embodiments, the UE or network may have to update an AI / ML models based dataset for a set of scenarios and configurations. The model update process may include at least updating the model parameters based on training dataset.

[0108] In some embodiments, the UE or network may have to collect training dataset for a given scenarios or configurations. The training data collected can then be applied to train a new model or update an existing one.

[0109] In some embodiments, the UE or network may have to monitor the performance of AI / ML model. The model monitoring process may include comparison of the output from AI / ML model to the ground truth. In some cases, one node makes measurement of the ground truth and one node makes AI / ML model inference. In such cases, it may be necessary to exchange monitoring dataset, e.g., ground truth, AI / ML model inference output, from one node to the other.

[0110] In some embodiments, one node, e.g., network node, UE, may train a model and transfer to the other node. The model can be compiled for execution before or after the model transfer. This may be beneficial as it allows to train the model in the environment it is going to be used (for inference).

[0111] The process of managing the different aspects mentioned above, including: data collection, model training, model selection, model activation, model inference, model deactivation, model switching, model updating, model monitoring, etc., can be referred as model life cycle management (LCM).

[0112] In some embodiments a node can give assistance or control the LCM of a model in another node. As a typical example, the network may assist / control a model in the UE side for UE-side or UE part of two-sided model. ).

[0113] In some consideration, the network may provide the LCM assistance to the UE by being specific to a particular model. Thus, the network may be required to identify the model in UE side unambiguously. For this purpose a model ID can be used. This type of model LCM assistance can be termed as model-ID based LCM.

[0114] The model ID for model-ID based LCM can be associated to an implementation of AI / ML model, e.g., certain model structure, model parameter values and model quantization, etc. Thus, model ID can identify certain implementation of an AI / ML model unambiguously. If a model identified in this manner are deployed in more than one UEs, the same set of outputs is expected from the deployed models if the same set of inputs are fed to them.

[0115] The assignment of a model ID to a model Model-ID-based LCM can be performed in a model registration process. A model can be trained, tested at training server. It then gets registered to a registration server. The registration process may include assignment of model ID and association of meta-information. The meta-information may include all or subset of the following, model input information, model output information, model application scope (applicable scenarios, configurations), model size, model computational complexity information, model inference delay information, etc. The UE then may report it capability by indicating the supported AI / ML models via indication of model ID. The network then provides LCM assistance to the UE based on the reported models UE supports.

[0116] One practical limitation of model-ID based LCM is scalability. Model-ID based LCM may require assignment of a model ID for each possible implementations of AI / ML model. However, as the number of implementations is expected to be very large, such implementation dependent model identification is not scalable.

[0117] Another issue with model-ID-based LCM is its flexibility with respect to model update. In a model-ID based LCM, model update whether major or minor, may require model re-registration and model-ID reassignment. This process may incur delay between the model update and use (inference). Thus, it may be less flexible to update a model.

[0118] In some considerations, the network may provide LCM assistance to the UE based on the identified AI / ML functionalies it supports.

[0119] In one consideration, an AI / ML functionality may mean some or all of the following: model purpose (use case), model input configuration, model output configuration, model scope, model application scenarios, etc. Then the network may provide LCM assistance to the UE based on the AI / ML functionalities the UE the supports. In the forthcoming description of the disclosure, such approaches are referred to asfunctionality-based LCM.

[0120] One exemplary procedure of functionality-based LCM. A model can be trained and tested at a model training server. The model can then be deployed to the UE with description on the associated functionalities. The associated functionalities could be from a set of specified AI / ML functionalities. In this case, the UE capability signaling informs the network which AI / ML functionalities the UE supports. Then the network may provide the LCM assistance to the UE based on the reported AI / ML functionalities.

[0121] FIGURE 8 illustrates two-side model based CSI compression according to embodiments of the present disclosure.

[0122] One use case of artificial intelligence (AI) is AI / ML based CSI feedback. In particular, an auto-encoder (AE) (800), which is a two-sided model as depicted in FIGURE 8, consisting of an encoder part (801) at the UE (803) which generates the CSI feedback and a decoder (802) at the gNB (804) which reconstructs the CSI feedback. The main aim of an AE-based CSI feedback is to find the best representation of a channel state information in terms of feedback overhead. In another words, AE compresses the CSI to reduce the CSI feedback overhead.

[0123] A yet another use case of AI / ML is beam prediction. In one aspect, this beam prediction can be performed in the spatial domain. In particular, the UE or the base station may predict the appropriate or the best beam among a large set of beams based on fewer measurements from a smaller set of beam.

[0124] FIGURE 9 illustrates AI / ML based spatial-domain beam prediction according to embodiments of the present disclosure.

[0125] FIGURE 9 depicts an exemplary case wherein a measurement set (901) is considered to predict, by an AI / ML model (900), the beams in prediction set (902). The beams corresponding to the solid circle (903) are measured and to predict the best beam among the beams corresponding to the hollow circles (905). In the FIGURE 9, therefore, 16 beams are measured to predict among the 32 beams in the prediction set. One exemplary way to implement the above is to consider wider and fewer beams (906) for measurement set (901) and narrow and a larger number of beams (907) for prediction set (902). A yet another exemplary way to implement the above is to consider fewer narrow beams for measurement set which are a subset of a larger number of beams for prediction set.

[0126]

[0127] In one aspect, the beam prediction can also be performed in the temporal domain too. In particular, a UE or the base station may predict the appropriate or the best beam among a large set of beams for the future based on fewer historical measurements on a smaller set of beams.

[0128] FIGURE 10 illustrates AI / ML based temporal-domain beam prediction according to embodiments of the present disclosure.

[0129] FIGURE 10 depicts an exemplary case wherein three measurements at t-2, t-1, and t, from a measurement set (1001) is considered to predict, by an AI / ML model (1000), the beams in prediction set (1002) for the future time instances t+1, t+2 and t+3.

[0130] In the forthcoming sections, the term 'measurement set' is interchangeably used with the term "Set B'. Similarly, the term 'prediction set' is interchangeably used with the term 'Set A'.

[0131] FIGURE 11 illustrates exemplary CSI report configuration for predictive beam reporting according to embodiments of the present disclosure.

[0132] When the beam prediction is done by an AI / ML model at the UE, i.e., UE-side AI / ML model, the Set B and Set A can be configured to UE as a set of CSI-RS resources. Such configuration can be under the CSI-ReportConfig (1101). FIGURE 11 shows configuration ofMandNCSI-RS resources for SetB (1102) and SetA (1103), respectively.

[0133] In general, the mapping between the CSI-RS resources in SetA and SetB to the physical transmission beams is up to gNB's implementation. As an example, the beam characteristics, including pointing angles, beam width, etc., for the beams corresponding to CSI-RS resources in setB can be different from one gNB to the other. Moreover, even in one gNB or cell, the pointing angles of such beam might vary in time.

[0134] FIGURE 12 illustrates exemplary cases for a network-side setting information varies in AI / ML based beam prediction according to embodiments of the present disclosure.

[0135] FIGURE 12 illustrates three exemplary cases, wherein the gNB-side setting would result in variation from the physical beam in SetA and SetB. In (1200), it is shown that due to the desired network-energy saving (NES) mode desired by the gNB, the number and beamwidth of the beams in SetA and SetB can be adjusted. On the other hand, in (1201), the number of TRPs corresponding to beams in SetA and SetB can be changed dynamically due to the desired mode of operation such as scheduling. Another example is provided in (1202), wherein the gNB antenna tilt angle can be changed depending on the geographic distribution of the users during the different times of the day. This again would result on variation on the physical characteristics of the beams in SetA and SetB.

[0136] In the aforementioned cases of AI / ML-based beam prediction, various aspects affect the performance. As an example, the AI / ML model for beam prediction must be trained with the same physical beam characteristics to measurement resources (CSI-RS) mapping during the training and inference stages. Therefore, a mechanism to secure consistency on the gNB Network-side setting during the training and inference stage is highly desired.

[0137] One way of developing two-sided model is by training sequentially the network-part and UE-parts of the two sides.

[0138] FIGURE 13 illustrates exemplary two-sided model training with training dataset provision from the network according to embodiments of the present disclosure.

[0139] In FIGURE 13, it is shown that the network (1301) trains its two-sided model (autoencoder) (1303) based on data collected from its coverage area. The network then provides a labeled dataset (1304) to the UE (1302). The UE (1302), then trains its UE-part of a two-sided model (1305) based on the dataset collected from the network.

[0140] Based on the gNB's implementation, the dataset for the training of the network-side model may vary. This variation may result from variation in network-side vendors, variation on the antenna setting, variation on the TRP which the decoder belongs to, etc. Thus, this variation would affect which models would pair with decoder at the network.

[0141] Similar to the aforementioned problem for beam prediction, the dataset sharing-based two-sided model training require consistency on the assumed network-side setting, including training dataset, for the model to pair at the inference stage.

[0142] Network-side Setting Information (NSI)

[0143] The aforementioned challenges for AI / ML operations at the UE-side, one of the key aspects is alignment of assumptions including network-side setting for model training and inference. In the following solutions to streamline assumption for UE-side or UE-part of two-sided model training is provided.

[0144] In one aspect of the present invention, the network configures the UE with measurement or report configuration for data collection wherein the configuration indicates the applied network-side setting information (NSI) via an identification which can be termed as NSI ID, or a model ID, or scenario ID or setting ID. In the following description of the invention, the term NSI ID is interchangeably used with model ID, scenario ID or setting ID, dataset ID or other ID that achieves alignment on network's setting for data collection or data provision for the training of UE-side or UE-part of two-side models.

[0145] In one aspect of the present invention, the network configures the UE with measurement or reporting configuration for inference wherein the configuration indicates the applied network-side setting information (NSI) via an identification which can be termed as NSI ID, or a model ID, or scenario ID or setting ID. In the following description of the invention, the term NSI ID is interchangeably used with model ID, scenario ID or setting ID, dataset ID or other ID that achieves alignment on network's setting for data collection or data provision for the inference of UE-side or UE-part of two-side models.

[0146] In a yet another aspect of the present invention, the UE upon the reception of the measurement or reporting configuration for inference which indicates the applied network-side setting information via an identification (ID), the UE selects the corresponding appropriate model for inference which matches for the indicated network side setting.

[0147] The network-side setting indicated by the ID may correspond to the network side antenna setting, the power setting, two-sided model training dataset setting, the physical beam to resource mapping, etc.

[0148] FIGURE 14 illustrates one exemplary case wherein the provision of network setting information achieves alignment between the model training and inference phases. In (1401) the gNB (1400) provides training dataset for the UE (1402). The provision of training dataset can be in terms of configuration for measurement and reporting or in the form of quantized dataset. In accordance with one aspect of this disclosure, the configuration for the provision of dataset includes the network-side setting information indication (ID). The collected training dataset in (1401) can be used inside the UE for on-device training or further transferred (1403) to an entity (1404) which carry out model training. The UE may use the ID to categorize the dataset and up on the transfer of the collected dataset to an external entity, the UE may also report the ID with collected dataset. This ID with additional set of information such us global cell id (GCI) can be used to develop a cell or site-specific model.

[0149] If the model is trained in the external physical entity other than the UE, the model may be delivered to the UE (1405) via either over-the-air interface signaling or other mechanisms. In the inference stage, the network may configure the UE with configuration associated to AI / Ml inference (1406). Such configuration may be associated to a report configuration or a measurement configuration or both. In the inference configuration, the network may provide the UE with an identification or ID which indicates the network-side setting information. Such network-side setting information may include model ID, dataset ID, network antenna setting or RF configurations or other related network-side settings. In accordance to the one aspect of the present disclosure, for a given network-side setting information identified by an ID, the network ensures the same relevant settings are applied in both data collections for training and inference stages.

[0150] In accordance to one aspect of the current disclosure, the UE upon receiving a configuration for AI / ML-based inference, wherein the configuration contains an indication for network-side setting information, the UE selects the appropriate AI / ML model compatible with the indicated setting. If the AI / ML model inference configuration corresponds to a reporting (1407), the UE ensures the reporting is from an AI / ML model compatible with the indicated network-side setting information.

[0151] In one exemplary embodiment, the network provides the list of network-side setting information for an AI / ML feature. The list may containNnetwork-side setting information (NSI) where each NSI configuration is associated with an ID.

[0152] In one exemplary embodiment, for the beam prediction based on UE-side AI / ML model, the network-side setting information (NSI) indicated by the corresponding ID corresponds to the beam characteristics for beams in measurement (setB) and prediction (setA) sets. Such network-side setting information is indicated in csi-reportConfig for predictive beam reporting.

[0153] In a yet another exemplary embodiment, for two-sided models, the network-side setting information (NSI) indicated by the corresponding ID corresponds compatibility information or pairing information for UE-part and network-part of two-sided models.

[0154] In a yet another exemplary embodiment, for two-sided models, the network-side setting information (NSI) indicated by the corresponding ID corresponds to a training dataset that is being provisioned or has been provided to the UE side.

[0155] In one aspect of this disclosure, the UE indicates to the network the list of network-side setting information (NSI), it supports via capability reporting. The UE may report the supported network-side setting information for supported AI / ML features or feature groups (FG).

[0156] In some cases, due to UE's internal condition such as memory, battery, etc., the UE may require to update the list of AI / ML operations associated to the network-side setting information provided. This will allow the UE to manage the AI / ML operations it can perform within the constraints of its internal condition.

[0157] In a yet another aspect of this disclosure, the UE may update to the network the list of network-side setting information (NSI) it supports. Such update can be provided by capability report update via RRC, or semi-dynamically via MAC-CE signaling or dynamically via uplink control information (UCI).

[0158] FIGURE 15 illustrates exemplary procedure for UE capability signaling and inference operation according to embodiments of the present disclosure.

[0159] In step 1500, the network provides a list of network-side settings per AI / ML feature or feature group (FG) to the UE. The UE receives the list of network-side settings per AI / ML feature or feature group (FG) from the network. In step 1501, the UE reports list of supported NSI to the network via capability report. The network receives the list of UE's supported NSI via the capability report from the UE. In step 1502, the network provides AI / ML functionality configuration with NSI to the UE. The UE receives AI / ML functionality configuration with NSI from the network. In step 1503, the UE reports AI / ML inference to the network. The network receives AI / ML inference reported from the UE. In step 1504, the UE reports its updated list of NSI to the network. The network receives updated list of NSI from the UE.

[0160] On-device Model Update and Fine-tuning.

[0161] In some case, the UE visits a certain area / site / cell often and stays there for a considerable amount of time. This is typically true if one considers the semi-deterministic mobility pattern human beings and machines follow. For example, employees commute to their offices several days a week and stay in the same neighborhood for several hours per day. In this case, if the UE has a capability to update its AI / ML model on-device, it may be beneficial to request a training dataset from the network to update its model.

[0162] When a UE has an older version of the AI / ML model which has not been updated via a new dataset, the UE may request the network for the provision / collection of fine-tuning dataset, or measurement configuration which can be used to collect fine-tuning dataset. Up on requesting for the fine-tuning dataset, the UE may indicate to the network the network-side setting information associated to the model to be updated.

[0163] FIGURE 16 illustrates exemplary procedure for UE capability update according to embodiments of the present disclosure.

[0164] In step 1600, the UE requests for data collection for fine-tuning corresponding to NSI to the network. In step 1601, the UE transmits configuration and / or activation of data collection for fine-tuning to the network. In step 1602, the network transmits measurement configuration for data collection with NSI indication to the UE. In step 1603, the network transmits a configuration for data collection to the UE. The UE receives the configuration for data collection from the network. In step 1604, the UE reports its updated list of NSI to the network. The network receives updated list of NSI from the UE.

[0165] In some cases, it may be beneficial if the gNB indicates the network-side information with additional area specific information. This allows the AI / ML model to be trained to capture localized / area / site-specific channel characteristics. Such area specific information could be indicated in the form of cell global identity, applicable area in terms of absolute or relative area coordinates, etc.

[0166] As another aspect of this disclosure, the network indicates to the UE the network-side setting information with applicable area / site information. The applicable area / site information could be indicated based on one or multiple of cell global identity, physical cell ID, absolute or relative range of position or area coordinates, etc.

[0167] In accordance with the present disclosure, the UE receives from the network one or multiple of network-side setting information and applicable area / site information. The UE then applies the applicable area / site information to select the appropriate / suitable AI / ML model. Moreover, the UE may use the network-side information and the applicable area / site information to categorize the collected training data. The UE may then train / fine-tune the AI / ML model based on the categorized training dataset.

[0168] FIGURE 17 illustrates the linkage between data collection (1705) for AI / ML model training / finetuning and AI / ML-based inference (1702) via a provision of the same applicable / cell / site / location information (1701, 1704) and network-side setting information (1703, 1706).

[0169] On Inference Timeline Management and AI / ML Processing Unit

[0170] In some cases, it is beneficial to group the configurations of AI / ML enabled feature. The set of configuration for AI / ML-enabled feature can then be referred as AI / ML-enabled functionality. The AI / ML functionality can in turn serve as a unit to switch AI / ML functions and manage inference timeline and AI / ML processing unit. As an example, the network may indicate to switch through configurations of the same functionality without additional delay between them. Conversely, the network may be required to provide sufficient time for the UE to switch through configurations belonging to the different network-side information ID.

[0171] In one aspect of this disclosure, the network configures the AI / ML enabled functionality as a set of configurations for an AI / ML-enabled feature or feature group wherein the configurations are associated with the same network-side information NSI ID.

[0172] In one aspect of this disclosure, the network activates the AI / ML functionality by activating one of the associated configurations. Upon activation of one of the configuration, the UE and network assume the associated functionality is activated.

[0173] In a yet another aspect of this disclosure, the network provides an activation control information to the configured functionality. The activation control information may indicated the functionality ID or the network-setting information ID and the corresponding activation message.

[0174] In one aspect of the current disclosure, a network provides a deactivation control information to a configured functionality. The deactivation control information may indicate the functionality ID or the network-setting information ID and the corresponding deactivation message.

[0175] In accordance with the present disclosure, up on the reception of deactivation control information, the UE deactivates all the corresponding AI / ML configurations are deactivated.

[0176] In accordance to the present disclosure, the network and the UE apply different timeline for the activated and deactivated functionalities. In particular, the UE and the network apply timeline x and timeline y for activated and deactivated AI / ML functionality, respectively. The timeline may include the processing time for AI / ML based inference, data collection, monitoring, etc.

[0177] In accordance to the present disclosure, if the network activates / triggers a configuration for AI / ML-enabled feature or feature group without providing an activation control information and without activating at least one of the configurations associated with the same functionality, the UE and the network may apply timeline y. Up on the reception of such activation / triggering information from the network, the UE performs inference in accordance with timeline y. The UE is not expected to complete its inference operation including the reporting associated with the inference operation before the minimum delay requirement set by timeline y.

[0178] In some cases, it is beneficial if the network manages the number of AI / ML-based operations it requests the UE to perform. Thus, a mechanism for the network not to overload the UE beyond its capacity may be required. If such mechanism is not in place, the network may run the risk of overloading the UE beyond its capability and the UE may fail to perform the requested AI / ML operation within the expected timeline.

[0179] In the following, the present disclosure provides solutions for the aforementioned problem.

[0180] In one aspect of the current disclosure, the UE reports its AI / ML processing capability via capability report in terms of the aggregated or total AI / ML processing units.

[0181] In a yet another aspect of the current disclosure, the UE reports its AI / ML processing capability via capability report in terms of AI / ML processing unit by being specific to a functionality. The capability report may contain information, which is AI / ML functionality-specific. The functionality-specific capability report indicates the processing requirement to perform various AI / ML operations over the functionality; this includes the processing requirement to perform a single inference, data collection measurement, data collection reporting, measurement for monitoring and reporting for monitoring.

[0182] Up one the reception of the capability report in terms of one or multiple of the aggregated / total AI / ML processing unit ( ) and functionality-specific AI / ML processing unit, the network tracks the remaining AI / ML processing resource headroom by counting the APUs associated to the overlapped (in time domain) AI / ML operations as follows:

[0183]

[0184] where, the APU requirement for configurationcof functionalityffor inference, monitoring and data collection are provided as , and , respectively. Additionally, and represent the number of configuration in functionfand the total number AI / ML functionalities.

[0185] However, it may be beneficial for the network if it is allowed to configure the UE with AIML operation that may cause overloading, e.g., for efficiency purposes. In one embodiment of the present disclosure, the UE the AIML operations, e.g., inference, monitoring, etc., if the occupied AIML processing units exceeds its reported capability (NAPU). The dropping could be based on the priority of AIML operations, the configuration order, etc.

[0186] In accordance with one aspect of the present disclosure, a method performed by a base station in a wireless communication system is provided, the method includes transmitting, to a terminal, configuration information about network-side setting information.

[0187] In accordance with an aspect of the present disclosure, a method performed by a user terminal in a wireless communication system is provided. The method includes receiving, from a base station, configuration information about network-side setting information.

[0188] In accordance with an aspect of the present disclosure, a method performed by a user terminal in a wireless communication system is provided. The method includes transmitting, by a terminal, capability information pertaining to network-side setting information.

[0189] In accordance with one aspect of the present disclosure, a method performed by a base station in a wireless communication system is provided, the method includes transmitting, to a terminal, information to activate / deactivate / switch / monitor AIML functionalities, models.

[0190] In accordance with an aspect of the present disclosure, a method performed by a user terminal in a wireless communication system is provided. The method includes receiving, from a base station, information to activate / deactivate / switch / monitor AIML functionalities, models.

[0191] Abbreviations

[0192] 2D Two-dimensional

[0193] ACK Acknowledgement

[0194] AoA Angle of arrival

[0195] AoD Angle of departure

[0196] ARQ Automatic Repeat Request

[0197] BW Bandwidth

[0198] CDM Code Division Multiplexing

[0199] CP Cyclic Prefix

[0200] C-RNTI Cell RNTI

[0201] CRS Common Reference Signal

[0202] CRI CSI-RS resource indicator

[0203] CSI Channel State Information

[0204] CSI-RS Channel State Information Reference Signal

[0205] CQI Channel Quality Indicator

[0206] DCI Downlink Control Information

[0207] dB deciBell

[0208] DL Downlink

[0209] DL-SCH DL Shared Channel

[0210] DMRS Demodulation Reference Signal

[0211] eMBB Enhanced mobile broadband

[0212] eNB eNodeB (base station)

[0213] FDD Frequency Division Duplexing

[0214] FDM Frequency Division Multiplexing

[0215] FFT Fast Fourier Transform

[0216] HARQ Hybrid ARQ

[0217] IFFT Inverse Fast Fourier Transform

[0218] LAA License assisted access

[0219] LBT Listen before talk

[0220] LTE Long-term Evolution

[0221] MIMO Multi-input multi-output

[0222] mMTC massive Machine Type Communications

[0223] MTC Machine Type Communications

[0224] MU-MIMO Multi-user MIMO

[0225] NACK Negative ACKnowledgement

[0226] NW Network

[0227] OFDM Orthogonal Frequency Division Multiplexing

[0228] PBCH Physical Broadcast Channel

[0229] PDCCH Physical Downlink Control Channel

[0230] PDSCH Physical Downlink Shared Channel

[0231] PHY Physical layer

[0232] PRB Physical Resource Block

[0233] PMI Precoding Matrix Indicator

[0234] PSS Primary Synchronization Signal

[0235] PUCCH Physical Uplink Control Channel

[0236] PUSCH Physical Uplink Shared Channel

[0237] QoS Quality of service

[0238] RAN Radio access network

[0239] RAT Radio access technology

[0240] RB Resource Block

[0241] RE Resource Element

[0242] RI Rank Indicator

[0243] RRC Radio Resource Control

[0244] RS Reference Signals

[0245] RSRP Reference Signal Received Power

[0246] SDM Space Division Multiplexing

[0247] SINR Signal to Interference and Noise Ratio

[0248] SPS Semi-Persistent Scheduling

[0249] SRS Sounding RS

[0250] SF Subframe

[0251] SSS Secondary Synchronization Signal

[0252] SU-MIMO Single-user MIMO

[0253] TDD Time Division Duplexing

[0254] TDM Time Division Multiplexing

[0255] TB Transport Block

[0256] TP Transmission point

[0257] TRP Transmission reception point

[0258] TTI Transmission time interval

[0259] UCI Uplink Control Information

[0260] UE User Equipment

[0261] UL Uplink

[0262] UL-SCH UL Shared Channel

[0263] URLLC Ultra-reliable low-latency communication

Claims

1.A method performed by a user equipment (UE) in a communication system, the method comprising:receiving, from a base station, first information for training data set, wherein the first information includes identification (ID) for a network-side setting;obtaining training data set based on the information;training an artificial intelligence (AI) / machine learning (ML) model based on the network-side setting using the training data set;receiving, from the base station, second information for inference, wherein the second information includes the ID for the network-side setting;performing an inference using the AI / ML model associated with the network-side setting; andtransmitting, to the base station, a result of the inference.2.The method of claim 1, wherein the first information includes:configuration for data collection for obtaining the training data set; orquantized data set corresponding to the training data set.3.The method of claim 1, wherein the second information includes:a first set of reference signals for prediction; anda second set of reference signals for measurement.4.The method of claim 1, wherein the second information is channel state information (CSI) report configuration information.5.A method performed by a base station in a communication system, the method comprising:transmitting, to a user equipment (UE), first information for training data set used for training an artificial intelligence (AI) / machine learning (ML) model based on a network-side setting, wherein the first information includes identification (ID) for a network-side setting;transmitting, to the UE, second information for inference, wherein the second information includes the ID for the network-side setting; andreceiving, from the UE, a result of an inference performed using the AI / ML model associated with the network-side setting.6.The method of claim 5, wherein the first information includes:configuration for data collection for obtaining the training data set; orquantized data set corresponding to the training data set.7.The method of claim 5, wherein the second information includes:a first set of reference signals for prediction; anda second set of reference signals for measurement.8.The method of claim 5, wherein the second information is channel state information (CSI) report configuration information.9.A user equipment (UE) in a communication system, the UE comprising:a transceiver; andat least one processor configured to:receive, from a base station, first information for training data set, wherein the first information includes identification (ID) for a network-side setting,obtain training data set based on the information,train an artificial intelligence (AI) / machine learning (ML) model based on the network-side setting using the training data set,receive, from the base station, second information for inference, wherein the second information includes the ID for the network-side setting,perform an inference using the AI / ML model associated with the network-side setting, andtransmit, to the base station, a result of the inference.10.The UE of claim 9, wherein the first information includes:configuration for data collection for obtaining the training data set; orquantized data set corresponding to the training data set.11.The UE of claim 9, wherein the second information includes:a first set of reference signals for prediction; anda second set of reference signals for measurement.12.The UE of claim 9, wherein the second information is channel state information (CSI) report configuration information.13.A base station in a communication system, the base station comprising:a transceiver; andat least one processor configured to:transmit, to a user equipment (UE), first information for training data set used for training an artificial intelligence (AI) / machine learning (ML) model based on a network-side setting, wherein the first information includes identification (ID) for a network-side setting,transmit, to the UE, second information for inference, wherein the second information includes the ID for the network-side setting, andreceive, from the UE, a result of an inference performed using the AI / ML model associated with the network-side setting.14.The base station of claim 13, wherein the first information includes:configuration for data collection for obtaining the training data set; orquantized data set corresponding to the training data set.15.The base station of claim 13, wherein the second information includes:a first set of reference signals for prediction; anda second set of reference signals for measurement.

Images