Verification of artificial intelligence (AI) / machine learning (ML) in beam management and hierarchical beam prediction

Abstract

A wireless transmit / receive unit (WTRU) may perform measurements on a first set of beam resources. The WTRU may then predict a beam resource in a second set of beam resources based on the measurement of the first set of beam resources. In addition, the WTRU may report the predicted beam resources. In addition, the WTRU may receive one or more first signals using the first beam. In one example, the first beam may use beam resources in the second set of beam resources. In addition, the WTRU may perform measurements on one or more accuracy parameters of the received one or more first signals. Further, the WTRU may transmit one or more second signals using the first beam under conditions that the measured one or more accuracy parameters of the received one or more first signals are acceptable.

Description

[0001] This application is a divisional application. The parent application is entitled "Verification of Artificial Intelligence (AI) / Machine Learning (ML) in Beam Management and Hierarchical Beam Prediction," filed on August 4, 2023, with application number [Application Number Missing]. .

[0002] Cross-references to related applications This application claims the benefit of U.S. Provisional Application No. 63 / 395,587, filed August 5, 2022, the contents of which are incorporated herein by reference. Background Technology

[0003] Beam management is a target use case for artificial intelligence (AI) / machine learning (ML) in air interfaces for wireless communications. This technology can lay a solid foundation for improving the performance and complexity of conventional beam management, including beam prediction in the time and / or spatial domains to reduce overhead and latency, as well as improve beam selection accuracy.

[0004] In wireless communication, conventional beam selection is based on beam sweeping at the gNode B (gNB) side or base station side, and at the Wireless Transmit / Receive Unit (WTRU) side or mobile phone side. Within frequency range 2 (FR2), conventional beam management can result in beam sweeping and measurement of numerous antennas at both the gNB and WTRU sides. When selecting the optimal beam, the WTRU can report up to four beams during beam management. In one example, the WTRU can report a beam based on the Reference Signal Received Power (RSRP).

[0005] Using AI / ML models, FR2 beam selection / prediction can be performed based on Frequency Range 1 (FR1) Channel State Information (CSI) measurements. However, in scenarios with hierarchical spatial relationships and associations between beam resources across different frequency ranges, the implementation of such a framework is constrained by key challenges in addressing beam measurement and reporting, as well as the training and validation of AI / ML models. Furthermore, using AI / ML-based beam prediction may not always be beneficial. As an example, in the case of non-line-of-sight (NLOS) communication, AI / ML-based beam prediction may be inaccurate, and conventional beam management procedures would be more advantageous. Summary of the Invention

[0006] A Wireless Transmit / Receive Unit (WTRU) can determine one or more beam resources based on measurements of other beam resources. The measured beam resources can be frequency range 1 (FR1) beam resources, and the determined beam resources can be frequency range 2 (FR2) beam resources. This determination can be based on an artificial intelligence (AI) / machine learning (ML) model. The WTRU can use one or more determined FR2 beam resources to receive signals. Furthermore, the WTRU can perform a verification process based on one or more accuracy parameters.

[0007] In one example, the WTRU can perform measurements on a first set of beam resources. Then, based on these measurements of the first set of beam resources, the WTRU can predict beam resources in a second set of beam resources. Furthermore, the WTRU can report the predicted beam resources. Additionally, the WTRU can use the first beam to receive one or more first signals. In one example, the first beam can use beam resources from the second set of beam resources. Furthermore, the WTRU can perform measurements on one or more accuracy parameters of the received one or more first signals. Furthermore, the WTRU can use the first beam to transmit one or more second signals, provided that the measured accuracy parameters of the received one or more first signals are acceptable. In one example, these accuracy parameters may be acceptable when the measured LOS is higher than a LOS threshold and the CQI is higher than a CQI threshold.

[0008] In another example, provided that one or more accuracy parameters of one or more received first signals are acceptable, the WTRU may use the first beam to receive one or more third signals.

[0009] In one example, one or more received first signals may be Physical Downlink Control Channel (PDCCH) signals. Alternatively, one or more received first signals may be Channel State Information-Reference Signals (CSI-RS).

[0010] Furthermore, in one example, using the first beam can include activating the first beam. Alternatively, using the first beam can include continuing to use the first beam.

[0011] In another example, the one or more accuracy parameters may include one or more of the line-of-sight (LOS) parameter, channel parameter, or channel quality indicator (CQI) parameter. In an additional example, the WTRU may also activate the AI / ML model to predict one or more second beams. In one example, the one or more second beams may use beam resources from a second set of beam resources. In additional or alternative examples, the WTRU may continue to use the AI / ML model to predict one or more second beams.

[0012] In an additional example, if one or more accuracy parameters of one or more received first signals are unacceptable, the WTRU may send a request to select and report a third beam. In one example, the measured LOS may be below a LOS threshold, and the measured CQI may be below a CQI threshold.

[0013] For example, the WTRU may send a request if one or more accuracy parameters of one or more received first signals are unacceptable. In one example, the measured LOS may be higher than the LOS threshold, and the measured CQI may be lower than the CQI threshold. The sent request may include a request to update the AI / ML model. The sent request may include a request to retrain the AI / ML model. Furthermore, the sent request may include a request to use the AI / ML model to predict and report the fourth beam.

[0014] Furthermore, if one or more accuracy parameters of one or more received first signals are unacceptable, the WTRU can fall back to a non-AI / ML beam management procedure to select and report a fifth beam. In one example, the measured CQI may be below the CQI threshold, and multiple time instances may have elapsed since the first signal was received using the first beam.

[0015] In another example, the WTRU may use one or more sixth beams to receive one or more fourth signals, and may measure one or more accuracy parameters of the received one or more fourth signals even if the measured accuracy parameters of the received one or more first signals are unacceptable. In one example, the measured LOS may be below the LOS threshold, the measured CQI may be below the CQI threshold, and there may not have been multiple time instances since the first signal was received using the first beam. Attached Figure Description

[0016] A more detailed understanding can be obtained from the following description given by way of example in conjunction with the accompanying drawings, wherein similar reference numerals in the drawings indicate similar elements, and wherein: Figure 1AThis is a system diagram illustrating an example communication system in which one or more of the disclosed implementation schemes may be implemented; Figure 1B This is an example of what can be achieved according to the implementation plan. Figure 1A A system diagram of an example wireless transceiver unit (WTRU) used in the illustrated communication system; Figure 1C This is an example of what can be achieved according to the implementation plan. Figure 1A System diagrams of example radio access networks (RAN) and example core networks (CN) used in the illustrated communication system; Figure 1D This is an example of what can be achieved according to the implementation plan. Figure 1A A system diagram of another example RAN and another example CN used within the illustrated communication system; Figure 2 This is a system diagram illustrating an example of performing beam prediction in a second set of beam resources based on beam resource reports in a first set of beam resources; Figure 3 This is a flowchart illustrating an example of the verification process for beam prediction based on hierarchical spatial relationships; and Figure 4 This is a flowchart illustrating an example of the predicted beam management. Detailed Implementation

[0017] Figure 1A This is a diagram illustrating an example communication system 100 that may implement one or more of the disclosed embodiments. Communication system 100 may be a multiple access system providing content such as voice, data, video, messaging, and broadcasting to multiple wireless users. Communication system 100 enables multiple wireless users to access such content through the sharing of system resources, including wireless bandwidth. For example, communication system 100 may employ one or more channel access methods, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal FDMA (OFDMA), Single Carrier FDMA (SC-FDMA), Zero-Tail Unique Word Discrete Fourier Transform Extended OFDM (ZT-UW-DFT-S-OFDM), Unique Word OFDM (UW-OFDM), Resource Block Filtered OFDM, and Filter Bank Multicarrier (FBMC), etc.

[0018] like Figure 1AAs shown, the communication system 100 may include wireless transceiver units (WTRUs) 102a, 102b, 102c, 102d, a radio access network (RAN) 104, a core network (CN) 106, a public switched telephone network (PSTN) 108, the Internet 110, and other networks 112. However, it should be understood that the disclosed embodiments contemplate any number of WTRUs, base stations, networks, and / or network elements. Each of the WTRUs 102a, 102b, 102c, and 102d can be any type of device configured to operate and / or communicate in a wireless environment. For example, WTRUs 102a, 102b, 102c, and 102d (any of which may be referred to as a Station (STA)) may be configured to transmit and / or receive wireless signals and may include User Equipment (UE), mobile stations, fixed or mobile subscriber units, subscription-based units, pagers, cellular phones, personal digital assistants (PDAs), smartphones, laptops, netbooks, personal computers, wireless sensors, hotspots or Mi-Fi devices, Internet of Things (IoT) devices, watches or other wearable devices, head-mounted displays (HMDs), vehicles, drones, medical devices and applications (e.g., remote surgery), industrial devices and applications (e.g., robots and / or other wireless devices operating in industrial and / or automated processing chain environments), consumer electronics devices, and devices operating on commercial and / or industrial wireless networks, etc. Any of WTRUs 102a, 102b, 102c, and 102d may be interchangeably referred to as a UE.

[0019] The communication system 100 may also include base station 114a and / or base station 114b. Each of base stations 114a and 114b can be any type of device configured to wirelessly interface with at least one of WTRUs 102a, 102b, 102c, and 102d to facilitate access to one or more communication networks, such as CN 106, Internet 110, and / or other networks 112. For example, base stations 114a and 114b can be transceiver base stations (BTS), Node Bs, evolved Node Bs (eNBs), home Node Bs, home evolved Node Bs, next-generation Node Bs (such as gNode Bs (gNBs)), new radio (NR) Node Bs, site controllers, access points (APs), and wireless routers, etc. Although base stations 114a and 114b are each depicted as a single element, it should be understood that base stations 114a and 114b may include any number of interconnected base stations and / or network elements.

[0020] Base station 114a may be part of RAN 104, which may also include other base stations and / or network elements (not shown), such as base station controllers (BSCs), radio network controllers (RNCs), and relay nodes. Base station 114a and / or base station 114b may be configured to transmit and / or receive radio signals on one or more carrier frequencies, which may be referred to as cells (not shown). These frequencies may be licensed spectrum, unlicensed spectrum, or a combination of unlicensed and unlicensed spectrum. The cell may provide coverage of radio services to a specific geographic area, which may be relatively fixed or changeable over time. The cell may be further divided into cell sectors. For example, the cell associated with base station 114a may be divided into three sectors. Thus, in one embodiment, base station 114a may include three transceivers, i.e., one transceiver for each sector of the cell. In one embodiment, base station 114a may employ multiple-input multiple-output (MIMO) technology and may utilize multiple transceivers for each sector of the cell. For example, beamforming can be used to transmit and / or receive signals in a desired spatial direction.

[0021] Base stations 114a and 114b can communicate with one or more of WTRUs 102a, 102b, 102c, and 102d via air interface 116, which can be any suitable wireless communication link (e.g., radio frequency (RF), microwave, centimeter wave, micrometer wave, infrared (IR), ultraviolet (UV), visible light, etc.). Any suitable radio access technology (RAT) can be used to establish air interface 116.

[0022] More specifically, as noted above, communication system 100 can be a multiple access system and can employ one or more channel access schemes, such as CDMA, TDMA, FDMA, OFDMA, and SC-FDMA. For example, base stations 114a and WTRUs 102a, 102b, and 102c in RAN 104 can implement radio technologies (such as Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access (UTRA)) that can use wideband CDMA (WCDMA) to establish air interface 116. WCDMA may include communication protocols such as High-Speed ​​Packet Access (HSPA) and / or evolved HSPA (HSPA+). HSPA may include High-Speed ​​Downlink (DL) Packet Access (HSDPA) and / or High-Speed ​​Uplink (UL) Packet Access (HSUPA).

[0023] In the implementation scheme, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies such as evolved UMTS terrestrial radio access (E-UTRA) that can use Long Term Evolution (LTE) and / or Advanced LTE (LTE-A) and / or Advanced LTE Pro (LTE-A Pro) to establish air interface 116.

[0024] In the implementation scheme, base station 114a and WTRUs 102a, 102b, 102c can implement radio technologies (such as NR radio access) that can use NR to establish air interface 116.

[0025] In one implementation, base station 114a and WTRUs 102a, 102b, and 102c can implement multiple radio access technologies. For example, base station 114a and WTRUs 102a, 102b, and 102c can together implement LTE radio access and NR radio access, for example, using the dual connectivity (DC) principle. Therefore, the air interface utilized by WTRUs 102a, 102b, and 102c can be characterized by multiple types of radio access technologies and / or transmissions to / from multiple types of base stations (e.g., eNBs and gNBs).

[0026] In other implementations, base station 114a and WTRUs 102a, 102b, and 102c can implement radio technologies such as IEEE 802.11 (i.e., Wi-Fi), IEEE 802.16 (i.e., WiMAX), CDMA2000, CDMA2000 1X, CDMA2000 EV-DO, Provisional Standard 2000 (IS-2000), Provisional Standard 95 (IS-95), Provisional Standard 856 (IS-856), Global System for Mobile Communications (GSM), Enhanced Data Rate GSM Evolution (EDGE), and GSM EDGE (GERAN).

[0027] Figure 1ABase station 114b can be, for example, a wireless router, a home node B, a home evolution node B, or an access point, and can utilize any suitable RAT to facilitate wireless connectivity in local areas such as commercial locations, homes, vehicles, campuses, industrial facilities, air corridors (e.g., for use by drones), and roads. In one embodiment, base station 114b and WTRUs 102c, 102d can implement radio technologies (such as IEEE 802.11) to establish a wireless local area network (WLAN). In another embodiment, base station 114b and WTRUs 102c, 102d can implement radio technologies (such as IEEE 802.15) to establish a wireless personal area network (WPAN). In yet another embodiment, base station 114b and WTRUs 102c, 102d can utilize cellular-based RATs (e.g., WCDMA, CDMA2000, GSM, LTE, LTE-A, LTE-A Pro, NR, etc.) to establish picocells or femtocells. Figure 1A As shown, base station 114b may have a direct connection to Internet 110. Therefore, base station 114b may not need to access Internet 110 via CN106.

[0028] RAN 104 can communicate with CN 106, which can be any type of network configured to provide voice, data, application, and / or Voice over Internet Protocol (VoIP) services to one or more of WTRUs 102a, 102b, 102c, and 102d. Data can have different Quality of Service (QoS) requirements, such as different throughput requirements, latency requirements, error tolerance requirements, reliability requirements, data throughput requirements, and mobility requirements. CN 106 can provide call control, billing services, location-based services, prepaid calling, internet connectivity, video distribution, and / or perform advanced security functions such as user authentication. Although not explicitly stated... Figure 1A As shown, but it should be understood that RAN 104 and / or CN 106 can communicate directly or indirectly with other RANs that use the same RAT as RAN 104 or a different RAT. For example, in addition to being connected to RAN 104 which can utilize NR radio technology, CN 106 can also communicate with another RAN (not shown) that uses GSM, UMTS, CDMA 2000, WiMAX, E-UTRA or WiFi radio technology.

[0029] CN 106 may also act as a gateway for WTRUs 102a, 102b, 102c, and 102d to access PSTN 108, the Internet 110, and / or other networks 112. PSTN 108 may include a circuit-switched telephone network providing Common Old-Style Telephone Service (POTS). The Internet 110 may include a global system of interconnected computer networks and devices using common communication protocols such as Transmit Control Protocol (TCP), User Datagram Protocol (UDP), and / or Internet Protocol (IP) from the TCP / IP Internet Protocol suite. Network 112 may include wired and / or wireless communication networks owned and / or operated by other service providers. For example, network 112 may include another CN connected to one or more RANs, which may use the same RAT as RAN 104 or a different RAT.

[0030] Some or all of the WTRUs 102a, 102b, 102c, and 102d in communication system 100 may include multi-mode capability (e.g., WTRUs 102a, 102b, 102c, and 102d may include multiple transceivers for communicating with different wireless networks via different wireless links). For example, Figure 1A The WTRU 102c shown can be configured to communicate with a base station 114a that can employ cellular-based radio technology and with a base station 114b that can employ IEEE 802 radio technology.

[0031] Figure 1B This is a system diagram illustrating the example WTRU 102. For example... Figure 1B As shown, WTRU 102 may include a processor 118, a transceiver 120, a transmitting / receiving element 122, a speaker / microphone 124, a keypad 126, a display / touchpad 128, non-removable memory 130, removable memory 132, a power supply 134, a Global Positioning System (GPS) chipset 136, and / or other peripheral devices 138, etc. It should be understood that, while remaining consistent with the implementation, WTRU 102 may include any sub-combination of the foregoing elements.

[0032] Processor 118 may be a general-purpose processor, a special-purpose processor, a conventional processor, a digital signal processor (DSP), multiple microprocessors, one or more microprocessors associated with a DSP core, a controller, a microcontroller, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), any other type of integrated circuit (IC), and a state machine, etc. Processor 118 may perform signal decoding, data processing, power control, input / output processing, and / or any other functionality that enables WTRU 102 to operate in a wireless environment. Processor 118 may be coupled to transceiver 120, which may be coupled to transmitting / receiving element 122. Although Figure 1B The processor 118 and transceiver 120 are depicted as separate components, but it should be understood that the processor 118 and transceiver 120 may be integrated together in an electronic package or chip.

[0033] Transmitting / receiving element 122 may be configured to transmit signals to or receive signals from a base station (e.g., base station 114a) via air interface 116. For example, in one embodiment, transmitting / receiving element 122 may be an antenna configured to transmit and / or receive RF signals. In another embodiment, transmitting / receiving element 122 may be a transmitter / detector configured to transmit and / or receive, for example, IR, UV, or visible light signals. In yet another embodiment, transmitting / receiving element 122 may be configured to transmit and / or receive both RF signals and optical signals. It should be understood that transmitting / receiving element 122 may be configured to transmit and / or receive any combination of wireless signals.

[0034] Although the transmitting / receiving element 122 is in Figure 1B While depicted as a single element, WTRU 102 may include any number of transmitting / receiving elements 122. More specifically, WTRU 102 may employ MIMO technology. Thus, in one embodiment, WTRU 102 may include two or more transmitting / receiving elements 122 (e.g., multiple antennas) for transmitting and receiving wireless signals via air interface 116.

[0035] Transceiver 120 can be configured to modulate signals to be transmitted by transmitting / receiving element 122 and demodulate signals to be received by transmitting / receiving element 122. As noted above, WTRU 102 may have multi-mode capability. For example, transceiver 120 may therefore include multiple transceivers to enable WTRU 102 to communicate via multiple RATs (such as NR and IEEE 802.11).

[0036] The processor 118 of WTRU 102 may be coupled to a speaker / microphone 124, a keypad 126, and / or a display / touchpad 128 (e.g., a liquid crystal display (LCD) unit or an organic light-emitting diode (OLED) display unit) and may receive user input data therefrom. The processor 118 may also output user data to the speaker / microphone 124, keypad 126, and / or display / touchpad 128. Additionally, the processor 118 may access information from any type of suitable memory (such as non-removable memory 130 and / or removable memory 132) and store data in such suitable memory. Non-removable memory 130 may include random access memory (RAM), read-only memory (ROM), a hard disk, or any other type of memory storage device. Removable memory 132 may include a subscriber identity module (SIM) card, a memory stick, and a secure digital storage (SD) card, etc. In other embodiments, the processor 118 may access information from memory that is not physically located on WTRU 102 (such as on a server or home computer (not shown)) and store data in such memory.

[0037] The processor 118 may receive power from the power supply 134 and may be configured to distribute and / or control power to other components in the WTRU 102. The power supply 134 may be any suitable device for powering the WTRU 102. For example, the power supply 134 may include one or more dry cell battery packs (e.g., nickel-cadmium (NiCd), nickel-zinc (NiZn), nickel metal hydride (NiMH), lithium-ion (Li-ion), etc.), solar cells, and fuel cells, etc.

[0038] The processor 118 may also be coupled to a GPS chipset 136, which may be configured to provide location information (e.g., longitude and latitude) about the current location of the WTRU 102. In addition to or instead of the information from the GPS chipset 136, the WTRU 102 may receive location information from base stations (e.g., base stations 114a, 114b) via air interface 116 and / or determine its location based on the timing of signals received from two or more nearby base stations. It should be understood that, while remaining consistent with the implementation, the WTRU 102 may acquire location information using any suitable location determination method.

[0039] The processor 118 may also be coupled to other peripheral devices 138, which may include one or more software and / or hardware modules that provide additional features, functionality, and / or wired or wireless connectivity. For example, peripheral device 138 may include an accelerometer, electronic compass, satellite transceiver, digital camera (for photos and / or video), Universal Serial Bus (USB) port, vibration device, television transceiver, hands-free headset, Bluetooth.® Modules, FM radio units, digital music players, media players, video game player modules, internet browsers, virtual reality and / or augmented reality (VR / AR) devices, and activity trackers, etc. Peripheral devices 138 may include one or more sensors. Sensors may be one or more of the following: gyroscopes, accelerometers, Hall effect sensors, magnetometers, orientation sensors, proximity sensors, temperature sensors, time sensors; geolocation sensors, altimeters, light sensors, touch sensors, magnetometers, barometers, gesture sensors, biometric sensors, and humidity sensors, etc.

[0040] WTRU 102 may include a full-duplex radio for which the transmission and reception of some or all of the signals (e.g., associated with specific subframes for both UL (e.g., for transmission) and DL (e.g., for reception)) may be concurrent and / or simultaneous. The full-duplex radio may include an interference management unit for reducing and / or substantially eliminating self-interference through signal processing via hardware (e.g., a choke) or via a processor (e.g., a separate processor (not shown) or via processor 118). In an embodiment, WTRU 102 may include a half-duplex radio for which the transmission and reception of some or all of the signals (e.g., associated with specific subframes for UL (e.g., for transmission) or DL ​​(e.g., for reception)) may be concurrent and / or simultaneous.

[0041] Figure 1C This is a system diagram illustrating RAN 104 and CN 106 according to the implementation scheme. As noted above, RAN 104 can communicate with WTRUs 102a, 102b, and 102c via air interface 116 using E-UTRA radio technology. RAN 104 can also communicate with CN 106.

[0042] RAN 104 may include evolved Nodes B 160a, 160b, and 160c, but it should be understood that RAN 104 may include any number of evolved Nodes B while remaining consistent with the implementation scheme. Each evolved Node B 160a, 160b, and 160c may include one or more transceivers for communicating with WTRUs 102a, 102b, and 102c via air interface 116. In one implementation, evolved Nodes B 160a, 160b, and 160c may implement MIMO technology. Therefore, evolved Node B 160a may, for example, use multiple antennas to transmit radio signals to and / or receive radio signals from WTRU 102a.

[0043] Each of the evolved nodes B 160a, 160b, and 160c can be associated with a specific cell (not shown) and can be configured to handle radio resource management decisions, handover decisions, and user scheduling in the UL and / or DL, etc. Figure 1C As shown, evolution nodes B 160a, 160b, and 160c can communicate with each other via the X2 interface.

[0044] Figure 1C The CN 106 shown may include a Mobility Management Entity (MME) 162, a Serving Gateway (SGW) 164, and a Packet Data Network (PDN) Gateway (PGW) 166. Although the foregoing elements are depicted as part of CN 106, it should be understood that any of these elements may be owned and / or operated by an entity other than a CN operator.

[0045] MME 162 can connect to each of the evolved nodes B 162a, 162b, and 162c in RAN 104 via the S1 interface and can act as a control node. For example, MME 162 can be responsible for authenticating users of WTRUs 102a, 102b, and 102c, activating / deactivating bearers, and selecting a specific serving gateway during the initial attachment of WTRUs 102a, 102b, and 102c. MME 162 can provide control plane functions for handover between RAN 104 and other RANs (not shown) employing other radio technologies such as GSM and / or WCDMA.

[0046] The SGW 164 can connect to each of the evolved Nodes B 160a, 160b, and 160c in RAN 104 via the S1 interface. The SGW 164 typically routes and forwards user data packets to and from WTRUs 102a, 102b, and 102c. The SGW 164 can perform other functions such as anchoring the user plane during handover between evolved Nodes B, triggering paging when DL data is available for WTRUs 102a, 102b, and 102c, and managing and storing the context of WTRUs 102a, 102b, and 102c.

[0047] SGW 164 can be connected to PGW 166, which provides WTRU 102a, 102b, 102c with access to packet-switched networks (such as Internet 110) to facilitate communication between WTRU 102a, 102b, 102c and IP-enabled devices.

[0048] CN 106 can facilitate communication with other networks. For example, CN 106 can provide WTRUs 102a, 102b, and 102c with access to a circuit-switched network (such as PSTN 108) to facilitate communication between WTRUs 102a, 102b, and 102c and traditional landline communication equipment. For example, CN 106 may include an IP gateway (e.g., an IP Multimedia Subsystem (IMS) server) that acts as an interface between CN 106 and PSTN 108, or be able to communicate with such an IP gateway. Additionally, CN 106 can provide WTRUs 102a, 102b, and 102c with access to other networks 112, which may include other wired and / or wireless networks owned and / or operated by other service providers.

[0049] Despite WTRU in Figures 1A to 1D While described as a wireless terminal, it is conceivable that in some representative implementations, such a terminal may (e.g., temporarily or permanently) use a wired communication interface with a communication network.

[0050] In a representative implementation, the other network 112 may be a WLAN.

[0051] A WLAN in Basic Services Set (BSS) mode may have an access point (AP) for the BSS and one or more stations (STAs) associated with that AP. The AP may have access or an interface to a distribution system (DS) or another type of wired / wireless network that carries traffic to and / or carries traffic out of the BSS. Traffic originating outside the BSS and destined for a STA can be delivered to the STA via the AP. Traffic originating from a STA and destined for a destination outside the BSS can be transmitted to the AP for delivery to the appropriate destination. Traffic between STAs within the BSS can be transmitted via the AP, for example, where a source STA can transmit traffic to the AP, and the AP can deliver the traffic to the destination STA. Traffic between STAs within the BSS can be considered and / or referred to as point-to-point traffic. Point-to-point traffic can be transmitted between a source STA and a destination STA (e.g., directly between them) using Direct Link Establishment (DLS). In some representative implementations, the DLS may use 802.11e DLS or 802.11z Tunneled DLS (TDLS). WLANs using the Standalone BSS (IBSS) mode may not have an access point (AP), and STAs within the IBSS or using the IBSS (e.g., all STAs within a STA) can communicate directly with each other. The IBSS communication mode may sometimes be referred to as a “self-organizing” communication mode in this document.

[0052] When operating in 802.11ac infrastructure mode or a similar mode, the AP can transmit beacons on a fixed channel, such as the primary channel. The primary channel can be of fixed width (e.g., a 20 MHz wide bandwidth) or dynamically configured. The primary channel can be the operating channel of the BSS and can be used by the STA to establish a connection with the AP. In some representative implementations, Carrier Sense Multiple Access / Collision Avoidance (CSMA / CA) can be implemented, for example, in an 802.11 system. For CSMA / CA, each STA (including the AP) can listen to the primary channel. If the primary channel is listened to / detected and / or determined to be busy by a particular STA, that particular STA can back off. A single STA (e.g., only one station) can transmit at any given time within a given BSS.

[0053] High-throughput (HT) STAs can communicate using a 40MHz wide channel (e.g., via a combination of a primary 20MHz channel and adjacent or non-adjacent 20MHz channels) to form a 40MHz wide channel.

[0054] Very High Throughput (VHT) STAs support channels with widths of 20MHz, 40MHz, 80MHz, and / or 160MHz. 40MHz and / or 80MHz channels can be formed by combining consecutive 20MHz channels. A 160MHz channel can be formed by combining eight consecutive 20MHz channels, or by combining two non-consecutive 80MHz channels (this can be referred to as an 80+80 configuration). For the 80+80 configuration, after channel coding, data can be processed by a segment parser that can split the data into two streams. Each stream can be processed individually using Inverse Fast Fourier Transform (IFFT) and time-domain processing. These streams can be mapped to two 80MHz channels, and the data can be transmitted via a transmitting STA. At the receiver of the receiving STA, the operations described above for the 80+80 configuration can be reversed, and the combined data can be transmitted to Media Access Control (MAC).

[0055] 802.11af and 802.11ah support operating modes below 1 GHz. Compared to those used in 802.11n and 802.11ac, 802.11af and 802.11ah reduce channel operating bandwidth and carrier. 802.11af supports 5 MHz, 10 MHz, and 20 MHz bandwidths in the TV white space (TVWS) spectrum, while 802.11ah supports 1 MHz, 2 MHz, 4 MHz, 8 MHz, and 16 MHz bandwidths using non-TVWS spectrum. According to representative implementations, 802.11ah may support instrument-type control / machine-type communication (MTC), such as MTC devices in macro coverage areas. MTC devices may have certain capabilities, such as limited capabilities, including support (e.g., only support) certain bandwidths and / or limited bandwidths. MTC devices may include batteries with battery life above a threshold (e.g., to maintain a very long battery life).

[0056] WLAN systems supporting multiple channels and channel bandwidths (such as 802.11n, 802.11ac, 802.11af, and 802.11ah) include channels that can be designated as the primary channel. The primary channel can have a bandwidth equal to the maximum common operating bandwidth supported by all STAs in the BSS. The bandwidth of the primary channel can be set and / or limited by STAs operating in the BSS (each supporting a minimum bandwidth operating mode). In the 802.11ah example, for STAs supporting (e.g., only supporting) a 1MHz mode (e.g., MTC type devices), the primary channel can be 1MHz wide, even if the AP and other STAs in the BSS support 2MHz, 4MHz, 8MHz, 16MHz, and / or other channel bandwidth operating modes. Carrier Sense and / or Network Allocation Vector (NAV) settings can depend on the status of the primary channel. If the primary channel is busy, for example because an STA (which only supports a 1MHz operating mode) is transmitting to the AP, all available frequency bands may be considered busy even if most available bands remain idle.

[0057] In the United States, the available frequency band for 802.11ah is 902MHz to 928MHz. In South Korea, the available frequency band is 917.5MHz to 923.5MHz. In Japan, the available frequency band is 916.5MHz to 927.5MHz. The total available bandwidth for 802.11ah is 6MHz to 26MHz, depending on the country code.

[0058] Figure 1DThis is a system diagram illustrating RAN 104 and CN 106 according to the implementation scheme. As noted above, RAN 104 may employ NR radio technology to communicate with WTRUs 102a, 102b, and 102c via air interface 116. RAN 104 may also communicate with CN 106.

[0059] RAN 104 may include gNBs 180a, 180b, and 180c, but it should be understood that RAN 104 may include any number of gNBs while remaining consistent with the implementation. Each of gNBs 180a, 180b, and 180c may include one or more transceivers for communicating with WTRUs 102a, 102b, and 102c via air interface 116. In one implementation, gNBs 180a, 180b, and 180c may implement MIMO technology. For example, gNBs 180a and 180b may utilize beamforming to transmit and / or receive signals from gNBs 180a, 180b, and 180c. Therefore, gNB 180a may, for example, use multiple antennas to transmit radio signals to and / or receive radio signals from WTRU 102a. In another implementation, gNBs 180a, 180b, and 180c may implement carrier aggregation technology. For example, gNB180a can transmit multiple component carriers to WTRU 102a (not shown). A subset of these component carriers may be on unlicensed spectrum, while the remaining component carriers may be on licensed spectrum. In implementations, gNBs 180a, 180b, and 180c may implement Coordinated Multipoint (CoMP) technology. For example, WTRU 102a may receive coordinated transmissions from gNBs 180a and 180b (and / or gNB 180c).

[0060] WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using transmissions associated with scalable digitization. For example, the OFDM symbol spacing and / or OFDM subcarrier spacing can be varied for different transmissions, different cells, and / or different portions of the radio transmission spectrum. WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using subframes or transmission time intervals (TTIs) of various or scalable lengths (e.g., including a varying number of OFDM symbols and / or a continuously varying length of absolute time).

[0061] gNBs 180a, 180b, and 180c can be configured to communicate with WTRUs 102a, 102b, and 102c in standalone and / or non-standalone configurations. In standalone configuration, WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c without needing to access other RANs (e.g., evolved Node Bs 160a, 160b, and 160c). In standalone configuration, WTRUs 102a, 102b, and 102c can utilize one or more of gNBs 180a, 180b, and 180c as mobile anchor points. In standalone configuration, WTRUs 102a, 102b, and 102c can communicate with gNBs 180a, 180b, and 180c using signals in unlicensed frequency bands. In a non-standalone configuration, WTRUs 102a, 102b, and 102c can communicate / connect with gNBs 180a, 180b, and 180c, and also with another RAN such as evolved Node Bs 160a, 160b, and 160c. For example, WTRUs 102a, 102b, and 102c can implement DC principles to communicate substantially simultaneously with one or more gNBs 180a, 180b, and 180c, and one or more evolved Node Bs 160a, 160b, and 160c. In a non-standalone configuration, evolved Node Bs 160a, 160b, and 160c can act as mobile anchors for WTRUs 102a, 102b, and 102c, and gNBs 180a, 180b, and 180c can provide additional coverage and / or throughput to serve WTRUs 102a, 102b, and 102c.

[0062] Each of gNBs 180a, 180b, and 180c can be associated with a specific cell (not shown) and can be configured to handle radio resource management decisions, handover decisions, user scheduling in UL and / or DL, network slicing support, interoperability between DC, NR, and E-UTRA, routing of user plane data to User Plane Functions (UPF) 184a and 184b, and routing of control plane information to Access and Mobility Management Functions (AMF) 182a and 182b, etc. Figure 1D As shown, gNB 180a, 180b, and 180c can communicate with each other via the Xn interface.

[0063] Figure 1DThe CN 106 shown may include at least one AMF 182a, 182b, at least one UPF 184a, 184b, at least one Session Management Function (SMF) 183a, 183b, and possibly a Data Network (DN) 185a, 185b. Although the foregoing elements are depicted as part of CN 106, it should be understood that any of these elements may be owned and / or operated by an entity other than a CN operator.

[0064] AMF 182a and 182b can connect to one or more of the gNBs 180a, 180b, and 180c in RAN 104 via the N2 interface and can act as control nodes. For example, AMF 182a and 182b can be responsible for authenticating users of WTRU 102a, 102b, and 102c, supporting network slicing (e.g., handling different Protocol Data Unit (PDU) sessions with different requirements), selecting specific SMF 183a and 183b, managing registration areas, terminating Non-Access Stratum (NAS) signaling, and mobility management. AMF 182a and 182b can use network slicing to customize CN support for WTRU 102a, 102b, and 102c based on the service types utilized by WTRU 102a, 102b, and 102c. For example, different network slices can be established for different use cases such as: services relying on Ultra Reliable Low Latency (URLLC) access, services relying on Enhanced Massive Mobile Broadband (eMBB) access, services for MTC access, etc. AMF 182a and 182b can provide control plane functions for handover between RAN 104 and other RANs (not shown) employing other radio technologies such as LTE, LTE-A, LTE-A Pro, and / or non-3GPP access technologies such as WiFi.

[0065] SMFs 183a and 183b can connect to AMFs 182a and 182b in CN 106 via the N11 interface. SMFs 183a and 183b can also connect to UPFs 184a and 184b in CN 106 via the N4 interface. SMFs 183a and 183b can select and control UPFs 184a and 184b, and configure traffic routing through UPFs 184a and 184b. SMFs 183a and 183b can perform other functions, such as managing and allocating UE IP addresses, managing PDU sessions, controlling policy enforcement and QoS, and providing DL data notifications. PDU session types can be IP-based, non-IP-based, and Ethernet-based, etc.

[0066] UPF 184a and 184b can be connected via the N3 interface to one or more of the gNBs 180a, 180b, and 180c in RAN 104. These gNBs can provide WTRU 102a, 102b, and 102c with access to packet-switched networks (such as the Internet 110) to facilitate communication between WTRU 102a, 102b, and 102c and IP-enabled devices. UPF 184 and 184b can perform other functions such as routing and forwarding packets, enforcing user plane policies, supporting multihomed PDU sessions, handling user plane QoS, buffering DL packets, and providing mobility anchoring.

[0067] CN 106 can facilitate communication with other networks. For example, CN 106 may include an IP gateway (e.g., an IP Multimedia Subsystem (IMS) server) that acts as an interface between CN 106 and PSTN 108, or be able to communicate with such an IP gateway. Additionally, CN 106 may provide WTRUs 102a, 102b, and 102c with access to other networks 112, which may include other wired and / or wireless networks owned and / or operated by other service providers. In one embodiment, WTRUs 102a, 102b, and 102c can be connected to DNs 185a and 185b via UPFs 184a and 184b through their N3 interfaces and their N6 interfaces with local DNs 185a and 185b.

[0068] Given Figures 1A to 1D as well as Figures 1A to 1D The corresponding descriptions herein refer to one or more of the functions described in one or more of the following items, which may be performed by one or more emulation devices (not shown): WTRU102a-d, base stations 114a-114b, evolved Node B 160a-160c, MME 162, SGW 164, PGW 166, gNB 180a-180c, AMF 182a-182b, UPF 184a-184b, SMF 183a-183b, DN 185a-185b, and / or any other devices described herein. An emulation device can be one or more devices configured to emulate one or more of the functions described herein. For example, an emulation device can be used to test other devices and / or simulate network and / or WTRU functions.

[0069] Simulation devices can be designed to perform one or more tests on other devices in a laboratory environment and / or an operator network environment. For example, one or more simulation devices can perform one or more functions, or all functions, while being fully or partially implemented and / or deployed as part of a wired and / or wireless communication network to test other devices within the communication network. One or more simulation devices can perform one or more functions, or all functions, while being temporarily implemented / deployed as part of a wired and / or wireless communication network. Simulation devices can be directly coupled to another device for testing and / or perform tests using over-the-air wireless communication.

[0070] One or more emulation devices can perform one or more (including all) functions without being implemented / deployed as part of a wired and / or wireless communication network. For example, emulation devices can be used in test scenarios within test laboratories and / or non-deployed (e.g., testing) wired and / or wireless communication networks to test one or more components. One or more emulation devices can be test rigs. Direct RF coupling and / or wireless communication via RF circuitry (e.g., which may include one or more antennas) can be used by the emulation devices to transmit and / or receive data.

[0071] Within frequency range 2 (FR2), conventional beam management can result in beam sweeping and measurements of numerous antennas at both the gNB and WTRU sides. When selecting the optimal beam, the WTRU can report up to four beams during beam management (e.g., based on the reference signal received power (RSRP)).

[0072] Using artificial intelligence (AI) / machine learning (ML) models, FR2 beam selection / prediction can be performed based on Frequency Range 1 (FR1) Channel State Information (CSI) measurements. However, in scenarios with hierarchical spatial relationships and associations between beam resources across different frequency ranges, the implementation of such a framework is constrained by key challenges in addressing beam measurement and reporting, as well as the training and validation of AI / ML models. Furthermore, using AI / ML-based beam prediction may not always be beneficial. As an example, in the case of non-line-of-sight (NLOS) communication, AI / ML-based beam prediction may be inaccurate, and conventional beam management procedures would be more advantageous.

[0073] This leads to varying WTRU behaviors in determining beam resource associations, measurement, and reporting, as well as in the training, validation, activation, and / or deactivation of AI / ML models. Therefore, further research is needed on hierarchical beam prediction in NR AI / ML beam management.

[0074] The implementation schemes and examples in this paper explain how to effectively / dynamically activate / deactivate AI / ML model-based beam prediction. Therefore, the beam management process can be advantageously modified.

[0075] This paper provides methods for activating or deactivating AI / ML models in beam prediction based on beam measurements of different beam resources within the AI / ML framework, as presented in the implementations and examples. In the examples, different beam resources may include resources with different beamwidths, different frequency ranges, etc. This paper proposes determining the accuracy of the AI / ML model considering different use cases and conditions, providing various options for selecting between activating and deactivating the AI / ML model. Iterative retraining / updating of the AI / ML model based on the AI / ML output and the predicted beam is considered; specifically, this paper provides conditions for AI / ML model retraining due to changes in the activation / deactivation set of the Transmit Configuration Indicator (TCI) state. Finally, this paper presents AI / ML model validation based on reciprocity for beam prediction.

[0076] In the following text, “a,” “an,” and similar terms and phrases may be interpreted as “one or more” and “at least one.” Similarly, any term or phrase ending with the suffix “(s)” may be interpreted as “one or more” and “at least one.” For example, the term “may” may be interpreted as “able to.”

[0077] As used in the implementations and examples herein, AI can be broadly defined as behavior exhibited by machines. Such behavior may, for example, mimic cognitive functions such as sensing, inference, adaptation, and action.

[0078] As used in the implementations and examples herein, ML can refer to a type of algorithm based on learning to solve problems through experience (“data”) without explicit programming (“configuration rule set”). ML can be considered a subset of AI. Different machine learning paradigms can be envisioned based on the nature of the data or feedback available for learning the algorithm. For example, supervised learning methods may involve learning a function that maps inputs to outputs based on labeled training examples, where each training example can be a pair consisting of an input and a corresponding output. For example, unsupervised learning methods may involve detecting patterns in data without pre-existing labels. For example, reinforcement learning methods may involve performing a series of actions in an environment to maximize cumulative rewards. In some solutions, it is possible to apply machine learning algorithms using combinations or interpolations of the methods mentioned above. For example, semi-supervised learning methods may use a combination of a small amount of labeled data and a large amount of unlabeled data during training. In this respect, semi-supervised learning falls between unsupervised learning (no labeled training data) and supervised learning (labeled training data only).

[0079] As used in the implementations and examples herein, deep learning (DL) can refer to a class of machine learning algorithms that employ artificial neural networks (such as deep neural networks (DNNs)) loosely inspired by biological systems. DNNs are a special class of machine learning models inspired by the human brain, where the input is linearly transformed and passed multiple times through a non-linear activation function. DNNs typically consist of multiple layers, each composed of a linear transformation and a given non-linear activation function. DNNs can be trained using training data via a backpropagation algorithm. Recently, DNNs have demonstrated state-of-the-art performance in various fields (e.g., speech, vision, natural language, etc.) and for various supervised, unsupervised, and semi-supervised machine learning settings. The term "AI / ML-based (AIML) methods / processes" can refer to learning behaviors and / or conforming to requirements based on data, without an explicit configuration of the sequence of action steps. Such methods enable the learning of complex behaviors that might be difficult to specify, implement, or both when using older methods.

[0080] The WTRU can transmit or receive physical channels or reference signals (RS) based on at least one spatial domain filter. As used in the embodiments and examples herein, the term "beam" can be used to refer to a spatial domain filter.

[0081] The WTRU can use the same spatial domain filter used to receive blocks of RS (such as Channel State Information Reference Signal (CSI-RS)) or Synchronization Signal (SS) to transmit physical channels or signals. The WTRU transmission may be referred to as the "target," and the received RS or SS block may be referred to as the "reference" or "source." In such cases, the WTRU is alleged to transmit the target physical channel or signal based on the spatial relationships of the referenced RS or SS blocks.

[0082] The WTRU can transmit the first physical channel or signal using the same spatial domain filter as the spatial domain filter used to transmit the second physical channel or signal. The first transmission and the second transmission can be referred to as the "target" and the "reference" (or "source"), respectively. In such cases, the WTRU is alleged to transmit the first (target) physical channel or signal based on the spatial relationship of the reference second (reference) physical channel or signal.

[0083] Spatial relationships can be implicit, configured by Radio Resource Control (RRC) signaling, or signaled by the MAC Control Element (CE) or Downlink Control Information (DCI). For example, a WTRU can implicitly transmit the Physical Uplink Shared Channel (PUSCH) transmission and the Demodulation Reference Signal (DM-RS) of the PUSCH based on the same spatial domain filter as the Sounding Reference Signal (SRS), which is indicated by an SRS Resource Indicator (SRI) configured in the DCI or by RRC signaling. As another example, spatial relationships can be configured by RRC signaling for the SRI or signaled by the MAC CE for the Physical Uplink Control Channel (PUCCH). Such spatial relationships can also be referred to as "beam indications."

[0084] The WTRU can receive a first (target) downlink channel or signal based on the same spatial domain filter or spatial reception parameters as the second (reference) downlink channel or signal. For example, such an association can exist between a physical channel (such as the Physical Downlink Control Channel (PDCCH) or Physical Downlink Shared Channel (PDSCH)) and its corresponding DM-RS. This association can exist at least when the first and second signals are reference signals, and when the WTRU is configured with a quasi-co-location (QCL) assumption type D between the corresponding antenna ports. This association can be configured as a TCI state. The WTRU can indicate the association between the CSI-RS or SS block and the DM-RS by indexing a set of TCI states (configured by RRC signaling and / or signaled via MAC CE). This indication can also be referred to as a "beam indication."

[0085] As used herein, Transmitting and Receiving Points (TRPs) may be used interchangeably with one or more of Transmitting Points (TPs), Receiving Points (RPs), Radio Remote Headers (RRHs), Distributed Antennas (DAs), Base Stations (BSs), Sectors (sectors of a BS), and Cells, but remain consistent with the implementations and examples provided herein. In one example, a cell may be a geographic cell area served by a BS. Furthermore, as used herein, multiple TRPs may be used interchangeably with one or more of MTRPs, M-TRPs, and multiple TRPs, but remain consistent with the implementations and examples provided herein.

[0086] The WTRU can report a subset of CSI components, which may correspond at least to the CSI-RS Resource Indicator (CRI), the Synchronization Signal Block (SSB) Resource Indicator (SSBRI), an indication of the panel used for reception at the WTRU (such as panel identifier or group identifier), measurement results such as L1-RSRP, L1-SINR (e.g., cri-RSRP, cri-SINR, ssb-Index-RSRP, ssb-Index-SINR) obtained from the SSB or CSI-RS, and other channel state information such as the Rank Indicator (RI), Channel Quality Indicator (CQI), Predecoding Matrix Indicator (PMI), and / or Layer Indicator (LI).

[0087] This paper's implementation includes activation / deactivation of beam prediction based on AI / ML modeling. Specifically, the implementation includes AI / ML model activation / retraining / deactivation / rollback options. Furthermore, the implementation includes determining the accuracy of the AI / ML model. Additionally, the implementation includes use cases and conditions for using the AI / ML model.

[0088] The implementation scheme presented in this paper includes dynamic retraining / updating of AI / ML models. Specifically, the implementation scheme includes iterative retraining / updating of AI / ML models based on AI / ML. Furthermore, the implementation scheme includes dynamic retraining / updating of AI / ML models based on changes in the activation / deactivation set of TCI states. Additionally, the implementation scheme includes reciprocity-based AI / ML model beam prediction verification.

[0089] The implementations and examples presented herein include activation / deactivation of FR2 beam prediction based on AI / ML modeling. In the examples provided herein, the WTRU is configured with one or more CSI-RS resources in FR1 for channel measurements. The WTRU derives one or more FR1 CSI parameters. One or more of the following may be applied: The WTRU may determine one or more FR2 beam resources, predict one or more FR2 beam resources, or both. In one example, the WTRU may perform determination, prediction, or both based on an AI / ML model. In one example, beam resources may consist of TCI states for downlink, CSI-RS or SSB, SRS resources, or TCI states for uplink. Furthermore, the WTRU may report one or more FR1 CSI parameters (e.g., multiple CRIs), and the base station or gNB may perform FR2 beam prediction accordingly. In one example, beam prediction may be based on an AI / ML model.

[0090] Furthermore, the WTRU can receive one or more FR2 beam resources, which can be predicted FR2 resources. Additionally, the WTRU can receive one or more thresholds for accuracy levels during the verification process. For example, the WTRU can receive thresholds for LOS probability, CQI, block error rate (BLER), Doppler drift, etc. Furthermore, the WTRU can perform a verification process on, for example, the received FR2 beam resources based on one or more accuracy parameters.

[0091] The implementation scheme described in this paper includes AI / ML model activation / retraining / deactivation / rollback options. In one example, based on the measured / determined accuracy parameters, WTRU can determine whether to use one or more of the following options. If the accuracy parameters are acceptable, the first option can be to use / activate the AI / ML model in beam prediction.

[0092] If the accuracy parameter is unacceptable, in the second option, the WTRU can select from other candidate beam resources based on AI / ML predictions. Furthermore, the WTRU can transmit one or more FR2 candidate beam resources, for example, indicated by the base station or gNB, or determined by the WTRU based on an AI / ML model. The WTRU can start the corresponding timer / counter. Additionally, the WTRU can monitor / measure the candidate beams. If the accuracy metric is acceptable for the beam resources, the WTRU can indicate the corresponding beam via the Physical Random Access Channel (PRACH) or PUCCH. If the counter / timer has exceeded the corresponding maximum count / time, the WTRU can switch to the fourth option, which will be explained further below.

[0093] The third option allows for updating / retraining parameters / models. The WTRU can transmit requests to update / retrain, for example, AI / ML parameters / models. The WTRU can start the corresponding timer / counter. The WTRU can then use the updated / retrained parameters / model for beam prediction in FR2. If the accuracy metric is acceptable for the predicted FR2 beam resources, the WTRU can indicate the appropriate beam via PRACH or PUCCH. If the counter / timer has exceeded its corresponding maximum count / time, the WTRU can switch to the fourth option, which is explained below.

[0094] The fourth option can include rollback. The WTRU can transmit requests to deactivate the AI / ML model and / or roll back to the regular beam management mechanism.

[0095] The implementation described in this paper includes determining the accuracy of the AI / ML model. The WTRU can determine the accuracy parameters for the predicted FR2 beam based on a verification process and corresponding thresholds. For example, for the predicted beam resources in FR2, if the measured CSI parameters and / or the assumed (Hyp.) PDCCH BLER are higher and / or lower than the corresponding thresholds, the WTRU can determine, for example, that the accuracy parameters for the AI / ML model are acceptable. In the example, the measured CSI parameters may include one or more of RSRP, Signal-to-Interference-plus-Noise Ratio (SINR), CQI, etc.

[0096] Furthermore, the implementation schemes and examples described herein include the ability of the WTRU to determine accuracy based on the correlation of one or more parameters. In the examples, these one or more parameters may include one or more of CQI, assumptions, PDCCH BLER, RSRP, SINR, LOS probability, Doppler drift, Doppler spread, average delay, delay spread, etc. For example, for a predicted beam resource in FR2, the LOS probability is higher than a first threshold (e.g., LOS_th); however, the derived CQI is lower than a corresponding threshold (e.g., CQI_th). Thus, the WTRU can determine to execute a third option. As another example, for a predicted beam resource in FR2, the LOS probability is lower than a first threshold (e.g., LOS_th) and the derived CQI is lower than a corresponding threshold (e.g., CQI_th). Thus, the WTRU can, for example, determine to execute a second or fourth option based on the determined LOS probability.

[0097] Additionally or alternatively, the WTRU may be configured with one or more use cases, such as a subset of one or more use cases, and / or conditions for which AI / ML models may be used, such as LOS, antenna panel configuration, etc. The WTRU may, for example, determine and transmit a request to the base station or gNB to fall back to FR2 beam management and deactivate AI / ML beam prediction based on the configured use cases.

[0098] Therefore, one or more of the following can be applied: An indication of LOS, a LOS probability, or both can be applied. Specifically, the WTRU can request a backoff if the LOS probability is below a corresponding threshold for any CSI-RS resource in the CSI-RS resource. For example, a base station or gNB can configure multiple FR1 beams to find one FR1 beam with the optimal LOS probability to be used in AI / ML FR2 beam prediction.

[0099] Furthermore, antenna panel configurations can be applied. In one example, if there is no identical QCL type D assumption between the FR1 antenna port and the FR2 antenna port and / or panel on the WTRU side, the WTRU can determine to send a request to the base station to fall back to FR2 beam management and deactivate AI / ML beam prediction.

[0100] In addition, WTRU can use other conditions to activate / deactivate AI / ML. For example, other conditions may include the number of supported FR1 and FR2 beams. Furthermore, other conditions may include the line-of-sight of the antenna arrays at the WTRU side, the base station or gNB side, or both sides, the beam direction of the antennas, or the antenna array configuration for FR1 and FR2. Additionally, other conditions may include WTRU capabilities.

[0101] The implementation schemes and examples in this paper include iterative training / updating of AI / ML models based on AI / ML output-predicted beams. In one example, the WTRU receives one or more sets of FR1 and FR2 beam resources and derives beam resource parameters. Beam resources may consist of one or more of the following: TCI states for downlink, CSI-RS or SSB, SRS resources for uplink, or TCI states for uplink.

[0102] During initial training, the WTRU can use measured CSI parameters and TCI states to train the beam prediction AI / ML model. In one example, the measured CSI parameters may include one or more of CQI, PMI, CRI, etc. The WTRU can, for example, determine one or more optimal FR2 beams based on RSRP. The WTRU can then report the predicted FR2 beams to the base station or gNB. The WTRU can, for example, receive one or more FR2 beams based on the reported FR2 beams. The WTRU can then determine whether the accuracy of the AI / ML and predictions is acceptable.

[0103] Retraining can be used optionally. If the accuracy is not within acceptable limits, the WTRU can use the received FR2 beams to retrain and / or update the corresponding AI / ML model parameters. For example, the WTRU can determine the amount of additional information, such as the number of FR2 beams used for retraining.

[0104] WTRU can use the measured CSI parameters in retrained / updated beam prediction AI / ML models and, accordingly, determine one or more optimal FR2 beams based on, for example, RSRP. In one example, the measured CSI parameters may include one or more of CQI, PMI, CIR, etc. WTRU can determine whether model retraining and updating has resulted in different outputs than the previous model, such as different predicted FR2 beams.

[0105] If model retraining has resulted in new / different outputs, such as a different predicted FR2 beam, the WTRU can report the predicted FR2 beam to the base station or gNB. Otherwise, the WTRU can perform the retraining step once or multiple times, for example, based on a timer or counter, and if unsuccessful, the WTRU determines to follow one or more of the activation / deactivation / rollback options.

[0106] The implementation schemes and examples described herein include AI / ML model retraining due to changes in the set of activated / deactivated TCI states. The WTRU receives the set of activated / deactivated TCI states. In one example, the WTRU may receive this set in the MAC-CE. If the WTRU further receives a second (e.g., updated / changed) set of activated / deactivated TCI states (e.g., in the MAC-CE), one or more of the following may apply: The WTRU may determine whether retraining of the AI / ML model is required. The WTRU may determine whether the second set of activated / deactivated TCI states partially overlaps with the first set. If the overlap is greater than a threshold, the WTRU may determine that retraining is not required. Otherwise, if the overlap is less than a threshold, the WTRU may determine that AI / ML retraining is required.

[0107] The implementation described herein includes reciprocity-based beam prediction AI / ML model validation. In the example, the WTRU determines / predicts one or more FR2 beams based on FR1 beam / CSI measurements. The WTRU performs transmissions to the base station or gNB based on the FR2 beams predicted by the WTRU. In one example, the transmission may include one or more of SRS, Hybrid Automatic Repeat Request (HARQ) Acknowledgments (Ack), or CSI-RS reports. The base station or gNB may measure the channel based on the received FR2 signal, such as RSRP. The base station or gNB may then modify or validate the beam selection on the WTRU side. The WTRU may receive one or more FR2CSI-RS measurement and reporting configurations, which may be based on the selected beam at the base station or gNB. In one example, the measurement and reporting configuration may include one or more of CSI RS resources, QCL information, TCI status, etc. The WTRU may measure FR2CSI and use the measured parameters to update / retrain the AI / ML model. The WTRU may select and report the optimal beam and the corresponding CSI quantities, such as CSI-RSRP, CIR, etc.

[0108] The WTRU can use channel and / or interference measurements. For example, the WTRU can receive synchronization signal / physical broadcast channel (SS / PBCH) blocks. SS / PBCH blocks (SSBs) may include primary synchronization signal (PSS), secondary synchronization signal (SSS), and / or physical broadcast channel (PBCH). The WTRU can monitor, receive, or attempt to decode SSBs during initial access, initial synchronization, radio link monitoring (RLM), cell search, cell handover, etc.

[0109] In addition, the WTRU can measure and report CSI, where CSI for each connectivity mode can include or be configured with one or more of the following: CSI reporting configuration, CSI-RS resource set, or non-zero power (NZP) CSI-RS resource. CSI reporting configuration can include one or more of the following: CSI reporting quantity, such as CQI, RI, PMI, CRI, LI, etc.; CSI reporting type, such as aperiodic, semi-persistent, or periodic; CSI reporting codebook configuration, such as type I, type II, type II port selection, etc.; or CSI reporting frequency. CSI-RS resource set can include one or more of the following CSI resource settings: NZP-CSI-RS resource for channel measurement; NZP CSI-RS resource for interference measurement; or CSI-IM resource for interference measurement. NZP CSI-RS resources may include one or more of the following: NZP CSI-RS resource identifier (ID); periodicity and offset; QCL information and TCI-status; or resource mapping, such as number of ports, density, code division multiplexing (CDM) type, etc.

[0110] A WTRU may indicate, define, or be configured with one or more reference signals. The WTRU can monitor, receive, and measure one or more parameters based on the corresponding reference signals. For example, one or more of the following may be applied. The following parameters are non-limiting examples of parameters that may be included in the measurement of the reference signals. One or more of these parameters may be included. Other parameters may be included.

[0111] SS-RSRP (SS-RSRP) can be measured based on a synchronization signal (e.g., the demodulation reference signal (DMRS) or SSS in the PBCH). SS-RSRP can be defined as a linear average of the power contribution of resource elements (REs) carrying the corresponding synchronization signal. Power scaling of the reference signal may be necessary when measuring RSRP. When SS-RSRP is used for L1-RSRP, the measurement can also be performed based on a CSI reference signal in addition to the synchronization signal.

[0112] CSI-RSRP can be measured based on a linear average of the power contribution of the RE carrying the corresponding CSI-RS. CSI-RSRP measurement can be configured within the measurement resources of the configured CSI-RS timing.

[0113] SS-SINR can be measured based on a synchronization signal (e.g., DMRS or SSS in the PBCH). SS-SINR can be defined as the linear average of the power contribution of the RE carrying the corresponding synchronization signal divided by the linear average of the noise and interference power contributions. When SS-SINR is used for L1-SINR, noise and interference power measurements can be performed based on resources configured by higher layers.

[0114] CSI-SINR can be measured by dividing the linear average of the power contribution of the RE carrying the corresponding CSI-RS by the linear average of the noise and interference power contributions. When CSI-SINR is used for L1-SINR, noise and interference power measurements can be performed based on resources configured at higher levels. Otherwise, noise and interference power can be measured based on resources carrying the corresponding CSI-RS.

[0115] The Received Signal Strength Indicator (RSSI) can be measured based on the average of the total power contribution in the configured OFDM symbols and bandwidth. Power contributions can be received from different resources, such as co-channel serving and non-serving cells, adjacent channel interference, thermal noise, etc.

[0116] Cross-layer interference received signal strength indicator (CLI-RSSI) can be measured based on the average of the total power contribution in the configured OFDM symbols across the configured time and frequency resources. Power contributions can be received from different resources (e.g., cross-layer interference, co-channel serving and non-serving cells, adjacent channel interference, thermal noise, etc.).

[0117] The probe reference signal RSRP (SRS-RSRP) can be measured based on a linear average of the power contribution of the RE carrying the corresponding SRS.

[0118] CSI report configurations (e.g., CSI-ReportConfigs, beam / CSI report configurations, etc.) can be associated with, for example, a single bandwidth portion (BWP) indicated by BWP-Id, where one or more of the following parameters are configured: CSI-RS resources and / or sets of CSI-RS resources for channel and interference measurements; CSI-RS report configuration type, including periodic, semi-persistent, and aperiodic; CSI-RS transmission periodicity for periodic and semi-persistent CSI reports; CSI-RS transmission slot offset for periodic, semi-persistent, and aperiodic CSI reports; a list of CSI-RS transmission slot offsets for semi-persistent and aperiodic CSI reports; time constraints for channel and interference measurements; report band configuration (wideband / subband CQI, PMI, etc.); thresholds and calculation modes for report quantities (CQI, RSRP, SINR, LI, RI, etc.); codebook configuration; group-based beam reporting; CQI table; subband size; non-PMI port indication; port index; etc.

[0119] The examples provided herein may include CSI-RS resource configuration. A CSI-RS resource set (e.g., NZP-CSI-RS-ResourceSet) may include one or more CSI-RS resources (e.g., NZP-CSI-RS-Resource and CSI-ResourceConfig), wherein the WTRU may be configured in the CSI-RS resource with one or more of the following: CSI-RS periodicity and slot offset for periodic and semi-persistent CSI-RS resources; CSI-RS resource mapping for defining the number, density, CDM type, OFDM symbols, and subcarrier occupancy of CSI-RS ports; bandwidth portions of the configured CSI-RS; or indexing including QCL source RSs and corresponding QCL type TCI states.

[0120] The examples provided in this document may include RS resource set configurations. One or more of the following configurations may be used for RS resource sets. Specifically, a WTRU may be configured with one or more RS resource sets. In addition, RS resource set configurations may include one or more of the following: RS resource set ID; one or more RS resources for the RS resource set; repetition (i.e., on or off); non-periodic trigger offset (e.g., one of 0 to 6 time slots); or tracking reference signal (TRS) information (e.g., true or false).

[0121] The examples provided in this document may include RS resource configurations. One or more of the following configurations may be used for RS resources. For example, a WTRU may be configured with one or more RS resources. In addition, RS resource configurations may include one or more of the following: RS resource ID; resource mapping, such as RE in a Physical Resource Block (PRB); power control offset (e.g., a value of -8, ..., 15); power control offset with SS (e.g., -3dB, 0dB, 3dB, 6db); scrambling code ID; periodicity and offset; or QCL information (e.g., based on TCI state).

[0122] In the following text, the attributes of an authorization or assignment may consist of at least one of the following: frequency allocation; time allocation aspects, such as duration; priority; modulation and decoding scheme; transport block size; number of spatial layers; number of transport blocks; TCI status, CRI, or SRI; number of repetitions; whether the repetition scheme is type A or type B; whether the authorization is a configured authorization type 1, type 2, or dynamic authorization; whether the assignment is a dynamic assignment or a semi-persistent scheduling (configuration) assignment; configuration authorization index or semi-persistent assignment index; periodicity of configuration authorization or assignment; Channel Access Priority Class (CAPC); or any parameters provided in the DCI by the MAC or by the RRC for scheduling authorization or assignment.

[0123] In the following, the indications made by the DCI may consist of at least one of the following: explicit indications made by the DCI field or by the RNTI used to mask the PDCCH; or implicit indications made by attributes such as the DCI format, DCI size, Coreset or search space, aggregation level, and the first resource element of the received DCI (e.g., the index of the first control channel element), wherein the mapping between attributes and values ​​may be signaled by the RRC or MAC.

[0124] Examples provided herein may include beam quality monitoring, radio link monitoring, or both. For example, a WTRU may use / receive / or be configured with one or more sets of reference signals for each BWP for monitoring and detecting beam faults. For example, the term q0 may be used for a beam fault detection set. In another example, the terms q0,0 or q0,1 may be used as a beam fault detection set. A beam fault detection set (e.g., set q0, q0,0, or q0,1) may include one or more reference signals, where the reference signals may be CSI-RS resource configuration indices and / or SSB indices. The reference signals included in the beam fault detection RS set may be the same as the reference signals configured / used / received for an RLM.

[0125] If the WTRU is not provided / configured with a beam fault detection RS set for the BWP (e.g., set q0, q0,0, or q0,1), the WTRU can determine the appropriate RS set. For example, the WTRU can determine the RS signals to be included in the beam fault detection RS set for the BWP based on a periodic CSI-RS resource configuration index, where the WTRU uses the periodic CSI-RS resource configuration index to monitor the PDCCH in the corresponding CORESET as indicated by the TCI status.

[0126] The WTRU can measure reference signals included in the beam fault detection RS set and thus estimate radio link quality. The WTRU can use one or more thresholds / ranges to monitor and estimate radio link quality. For example, an asynchronous threshold (e.g., Q_out), a synchronous threshold (e.g., Q_in), or both can be used, where the thresholds Q_out, Q_in, or both can be used to estimate the quality of the radio link and / or the corresponding beam. The terms Q_out and Q_in can be used to represent one or more attributes or parameters, and corresponding values ​​for those attributes or parameters.

[0127] The threshold Q_out can be used to determine the quality of radio links and / or beams that may not reliably receive signal transmissions, corresponding to the synchronization block error rate (BLER_out). Additionally or alternatively, the threshold Q_in can be used to determine the quality of radio links and / or beams that can reliably receive signal transmissions, corresponding to the synchronization block error rate (BLER_in). BLER_out, BLER_in, or both can be explicitly determined by the base station or gNB.

[0128] If the base station or gNB does not explicitly determine BLER_out and / or BLER_in, they can be estimated based on one or more parameters. For example, the WTRU can use, receive, or be configured with PDCCH transmission parameters to perform asynchronous, synchronous, or both assessments. In one example, the number of OFDM symbols, aggregation level, the ratio of assumed PDCCH RE energy to average SSS RE energy, the ratio of assumed PDCCH DMRS energy to average SSS RE energy, BWP in the PRB quantity, subcarrier spacing, etc., can be used to determine the BLER_out threshold, BLER_in threshold, or both thresholds.

[0129] Tables 1 and 2 show examples of PDCCH transmission parameters that can be included when evaluating the Q_out and Q_in thresholds, respectively. These tables are non-limiting examples of parameters that can be included in evaluating the asynchronous and synchronous thresholds. One or more of these parameters may be included. Values, the number of PRBs, and the selection of each parameter are examples. Other values, the number of PRBs, or selections may be included.

[0130] property The value of BLER configuration #0 DCI format 1-0 Control the number of OFDM symbols 2 Convergence Grade (CCE) 8 The assumed ratio of PDCCH RE energy to the average SSS RE energy 4dB The assumed ratio of PDCCH DMRS energy to the average SSS RE energy 4dB Bandwidth (PRB) 24 Subcarrier spacing (kHz) SCS of the DL BWP event DMRS pre-decoder granularity REG bundle size REG bundle size 6 CP length normal Mapping from REG to CCE distributed Table 1: PDCCH transmission parameters used for asynchronous evaluation property The value of BLER configuration #0 DCI payload size 1-0 Control the number of OFDM symbols 2 Convergence Grade (CCE) 4 The assumed ratio of PDCCH RE energy to the average SSS RE energy 0dB The assumed ratio of PDCCH DMRS energy to the average SSS RE energy 0dB Bandwidth (PRB) 24 Subcarrier spacing (kHz) SCS of the DL BWP event DMRS pre-decoder granularity REG bundle size REG bundle size 6 CP length normal Mapping from REG to CCE distributed Table 2: PDCCH transmission parameters used for synchronous evaluation In the following text, the term "RS" may be used interchangeably with one or more of RS resource, RS resource set, RS port, and RS port group, but remains consistent with the embodiments and examples provided herein. Furthermore, the term "RS" may be used interchangeably with one or more of SSB, CSI-RS, SRS, DM-RS, TRS, PRS, and Phase Tracking Reference Signal (PTRS), but remains consistent with the embodiments and examples provided herein.

[0131] In the following text, the phrase “reference signal” may be used interchangeably with one or more of the following, but remains consistent with the implementation schemes and examples provided herein: SRS, CSI-RS, DM-RS, PT-RS and / or SSB.

[0132] In the following text, the term "channel" may be used interchangeably with one or more of the following, but remains consistent with the implementations and examples provided herein: PDCCH, PDSCH, PUCCH, PUSCH, PRACH, etc. In the following text, the phrase "RS resource set" may be used interchangeably with one or more of RS resources and beamgroups, but remains consistent with the implementations and examples provided herein.

[0133] In the following text, the phrase “beam report” may be used interchangeably with one or more of CSI measurement, CSI reporting, and beam measurement, but remains consistent with the implementation schemes and examples provided herein. The proposed solution for beam resource prediction can be used for beam resources belonging to a single or multiple cells and a single or multiple TRPs, and remains consistent with the implementation schemes and examples provided herein.

[0134] In the following text, the phrase “CSI report” may be used interchangeably with one or more of CSI measurement, beam reporting, and beam measurement, but remains consistent with the implementations and examples provided herein. Similarly, the term “quality” or the phrase “measurement quality” may be used interchangeably with one or more of RSRP, Reference Signal Received Quality (RSRQ), SINR, CQI, Modulation and Decoding Scheme (MCS), Assumed PDCCH BLER, PDSCH BLER, LOS probability, etc., but remains consistent with the implementations and examples provided herein.

[0135] This article provides implementation schemes and examples for activating / deactivating beam prediction based on AI / ML modeling. In one example, the WTRU may receive one or more CSI report configurations. For example, the WTRU may receive CSI-ReportConfig. In another example, the WTRU may receive one or more CSI report configurations from the base station. The CSI report configuration may include a CSI report quantity, which may indicate CSI parameters that may need to be measured / estimated / derived and reported. In one example, the CSI report quantity may be one or more of CQI, RI, PMI, CRI, LI, SINR, RSRP, etc.

[0136] CSI reporting configurations can be associated with one or more CSI resource settings (e.g., CSI-ResourceConfig) used for channel / interference measurements. Resource settings can include a list of CSI resource sets, which may include references to one or more CSI-RS resource sets, SSB sets, or both.

[0137] Figure 2 This is a system diagram illustrating an example of beam prediction in a second set of beam resources based on beam resources reported in a first set of beam resources. Figure 2 The example shown presents a WTRU configured with a first set of beam resources. For example, the first set of beam resources could be CSI-RS resources, TCI states, etc. Furthermore, the first set of beam resources could be in a first frequency range (e.g., FR1) and / or a first beamwidth (e.g., a wide beamwidth for wide beams), which are shown, for example, as C... 1,1 C1,2 and C 1,3 .

[0138] Furthermore, the WTRU is configured with a second set of beam resources. For example, the second set of beam resources could be CSI-RS resources, TCI states, etc. Additionally, the second set of beam resources could be in a second frequency range (e.g., FR2) and / or a second beamwidth (e.g., a narrow beamwidth for narrow beams), these narrow beams in Figure 2 For example, it is shown as B. 2,1 B 2,2 ... B 2,9 .

[0139] The WTRU can perform measurements on one or more CSI-RS resources and derive one or more CSI parameters. In one example, the WTRU can determine which beam resource in a first set of beam resources is the optimal beam resource. Figure 2 C in 1,2 For example, WTRU can determine beam C. 1,2 It is the optimal beam because it has the highest RSRP or LOS. For example, the WTRU can determine the optimal PMI in the first set of CSI-RS resources, which is indicated by a directional arrow pointing from base station 214 to WTRU 202 and obstruction 203. In the example shown in system diagram 200, obstruction 203 reflects the signal received from base station 214 to WTRU 202.

[0140] WTRU can target the corresponding selected beam resources, such as C 1,2 The determined parameters, such as CSI parameters like RSRP, RI, LI, SINR, PMI, CQI, etc., are reported in the first set of CSI-RS resources. In one example, this could be at base station 214 of a gNB. The reported CSI parameters can be used, for example, based on an AI / ML model to predict one or more optimal beam resources in a second frequency range (e.g., FR2).

[0141] Additionally or alternatively, the WTRU may, for example, determine one or more optimal beam resources in a second set based on an AI / ML model. For instance, selection / determination / prediction may be based on, for example, the optimal PMI. Figure 2 Beam B in 2,4 and B 2,6In this way, the WTRU can report the determined / selected beam resources (e.g., beam index or CRI) and the corresponding predicted RSRP / SINR for up to a maximum number of beams (e.g., up to four beams). In one example, the corresponding predicted RSRP / SINR may include L1-RSRP, L1-SINR, etc. In other words, the optimal beam resources in the second frequency range can be determined based on an AI / ML model without excessive beam sweeping at base station 214 (which may be a gNB) or WTRU 202.

[0142] In a framework based on beam prediction for a second frequency range and measurements in a first frequency range (e.g., based on an AI / ML model), beam prediction may be less accurate and may require further verification and validation. Those skilled in the art will recognize that the embodiments and examples provided herein address one or more of these problems. For example, how is beam prediction (e.g., based on AI / ML) validated? How can one distinguish whether low-quality predictions (e.g., predicted beams with low RSRP) are due to poor behavior of the AI / ML model or other causes? How can cases with low prediction quality, such as predicted beams with low RSRP / CQI, be addressed? In one example, the WTRU can activate the AI / ML model in beam prediction based on one or more accuracy parameters for candidate beam resources. The WTRU can select additional candidate beam resources based on the AI / ML beam prediction. Furthermore, the WTRU can determine the accuracy parameters for the beam predicted by the AI / ML model. Additionally, the WTRU can deactivate AI / ML beam prediction based on one or more of the following: line-of-sight (LOS) indication, LOS probability, antenna panel configuration, number of supported beams, antenna array, or WTRU capabilities. Furthermore, the WTRU can use measured CSI parameters and TCI states in beam prediction used for the initial training of the AI / ML model.

[0143] In the example solution, the WTRU can be determined or configured to perform verification on the AI / ML output. For example, the AI / ML model can be used to predict beam resources in a second frequency range based on measurements (such as CSI measurements) in a first frequency range. For example, the AI / ML model can be executed at the WTRU and / or gNB. Beam resources can consist of TCI states for downlink, CSI-RS or SSB, SRS resources, or TCI states for uplink.

[0144] A WTRU can identify, indicate, or be configured to receive and measure one or more CSI / beam resources to verify the validity and / or accuracy of the AI / ML output. In one example, the WTRU can suggest, report, or request a base station or gNB to transmit one or more channels / signals corresponding to a predicted beam resource, such as the same QCL, the same spatial relationship, or both. In one example, these one or more channels / signals could be a PDCCH, CSI-RS signal, etc. Alternatively, the WTRU can be configured to receive and measure one or more channels / signals (e.g., PDCCH, CSI-RS signal, etc.) corresponding to a predicted beam resource (e.g., the same QCL, the same spatial relationship, or both). As an example, the requested / configured beam resource could be in a second frequency range, such as FR2.

[0145] The WTRU can determine or be configured to derive measurements of one or more parameters relating to the configured / determined beam resources. For example, the WTRU can be configured to derive CSI parameters, such as LOS probability, PMI, CQI, RSRP, SINR, Doppler drift, etc., based on the configured / received CSI-RS signal. As another example, the WTRU can be configured to derive parameters corresponding to the received channel, such as the PDCCH assumption BLER. Furthermore, the WTRU can determine or be configured with one or more thresholds associated with the parameters determined / configured for measurement. Additionally, the WTRU can determine or be configured with one or more limits / maximum / minimum values ​​associated with timers / counters used during the calibration process.

[0146] WTRU can determine / report the validity of an AI / ML model, classifying it as either "valid" (meaning the measured accuracy is acceptable) or "invalid" (meaning the measured accuracy is unacceptable). Thus, if the measured parameters or their associations / combinations are within acceptable limits, WTRU can determine that the AI / ML model is "valid" and has "acceptable accuracy." Conversely, if the measured parameters or their associations / combinations are not within acceptable limits, WTRU can determine that the AI / ML model is "invalid" and does not have "acceptable accuracy."

[0147] In the example solution, the WTRU can be configured with one or more options to select based on different conditions of measurement, threshold, and verification scenarios. Therefore, one or more examples in the following examples are applicable.

[0148] Under Option 1, WTRU can use an AI / ML model in beam prediction, activate an AI / ML model in beam prediction, or perform both. For example, WTRU can determine that the corresponding AI / ML model is valid and its accuracy is acceptable. Thus, WTRU can determine to use / activate the corresponding AI / ML model, for example, for beam prediction.

[0149] Under Option 2, the WTRU can select from other candidate beam resources based on AI / ML predictions. For example, the WTRU can determine or be configured with one or more candidate (predicted) beams, for instance, based on an AI / ML model. Thus, if the WTRU detects / determines that the optimal (predicted) beam is not showing acceptable accuracy, the WTRU can determine one or more candidate (predicted) beams to monitor. In the example, acceptable accuracy is not shown and can be represented by a CQI below a threshold, an RSRP below a threshold, or both, or by the PDCCH assumption BLER above a threshold.

[0150] Under option 3, parameters, the model, or both can be updated, retrained, or both. For example, the WTRU can determine that the reason why a predicted beam based on AI / ML does not have acceptable accuracy is due to the AI / ML model. In this way, the WTRU can determine, recommend, or transmit a request to update the AI / ML model, retrain the AI / ML model, or perform both.

[0151] Under option 4, the WTRU can fall back to a legacy procedure, deactivate the AI / ML model, or both. For example, the WTRU might determine that the quality, accuracy, or both of the (predicted) beam are below one or more thresholds, where the WTRU might determine that the AI / ML model is invalid. Therefore, the WTRU could determine to deactivate the AI / ML model, fall back to a legacy procedure, or perform both. In one example, falling back to a legacy procedure could include using a procedure based on a non-AI / ML model.

[0152] In the example solution, the WTRU can determine to change the selected option based on the measured accuracy parameter, one or more timers / counters, etc. In one example, the WTRU might determine to operate in option 2, where the WTRU can start a timer or counter, for example, for the number of candidate beams being monitored. If the timer (counter) times out (reaches its maximum value) before the WTRU is able to find another candidate beam with acceptable accuracy, the WTRU might determine to select another option. For example, the WTRU might determine to select option 3 or option 4. Additionally or alternatively, if the WTRU determines that one or more candidate beams among the candidate beams have acceptable accuracy, such as before the timer (counter) times out (reaches its maximum value), the WTRU might report the identified beam and the WTRU might determine to select the option to use / activate the AI / ML model, for example, as in option 1.

[0153] For example, the WTRU can determine to operate in option 3, where the WTRU can start a timer, a counter, or both. If the model update / retraining causes the WTRU to determine one or more beams with acceptable accuracy before the timer (counter) times out (reaches its maximum), the WTRU can determine to select whether to use, activate, or both (retrained / updated) the AI / ML model, for example, as in option 1. Conversely, if the timer times out (and / or the counter reaches its maximum) and the WTRU determines that the AI / ML model update / retraining has not yet resulted in acceptable accuracy, the WTRU can determine the option to roll back or deactivate the AI / ML model, for example, as in option 4.

[0154] In the example solution, the WTRU can determine and / or establish accuracy parameters and verification procedures based on the association / combination of one or more CSI, beam, channel, environment, and / or mobility parameters. For example, the WTRU can determine the association based on one or more of the following parameters.

[0155] WTRU can determine the establishment of associations based on beam resource parameters. For example, WTRU can determine parameters corresponding to beam resources and CSI quantities, such as RSRP, SINR, CQI, PMI, RI, LI, assumed PDCCH BLER, and corresponding thresholds.

[0156] WTRU can determine the establishment of associations based on channel, mobility, and environmental parameters. For example, WTRU can determine parameters corresponding to channel, environment, and mobility, such as LOS probability, Doppler drift, Doppler spread, average delay, delay spread, and corresponding thresholds.

[0157] For example, the WTRU can determine accuracy parameters based on the correlation / combination of CSI and / or beam parameters, as well as environmental, mobility, and channel parameters for (predicted) beam resources relative to corresponding thresholds. For example, the WTRU can determine one or more accuracy levels based on the correlation of CQI, combinations of CQI, or both, and LOS probability. For example, if the measured LOS probability and the measured received power and / or channel quality (e.g., CQI, RSRP, SINR, etc.) are above the corresponding thresholds, the WTRU can determine that the AI / ML model performance is acceptable, and therefore the WTRU can determine to validate the AI / ML model and use / activate the corresponding AI / ML model, for example, as in Option 1.

[0158] For example, the measured LOS probability may be higher than a corresponding threshold, while the measured received power and / or channel quality (e.g., CQI, RSRP, SINR, etc.) may be lower than a corresponding threshold. In this case, the WTRU can determine that the accuracy of the AI / ML is unacceptable. Therefore, the WTRU can determine to update the AI / ML model, for example, as in option 3. For example, if the measured LOS probability and the measured received power and / or channel quality (e.g., CQI, RSRP, SINR, etc.) are lower than a corresponding threshold, the WTRU can determine to monitor / measure one or more candidate (predicted) beams, for example, as in option 2.

[0159] In the example solution, the WTRU can suggest, request, or report the results of verification and the determined options, such as activating / deactivating the AI / ML model, to the base station or gNB. In one example, the WTRU can suggest, request, or report the results and options via uplink control information (UCI) in the PUSCH as part of the CSI report, as a marker in the HARQ-ACK, as a parameter in the PUCCH, as a parameter in the PRACH, etc.

[0160] In another example solution, the WTRU may identify or be configured with one or more use cases (or subsets of use cases), for which the WTRU determines whether to use, activate, or deactivate the AI / ML model. For example, the WTRU may determine to activate, deactivate, or both of the AI / ML model based on one or more of the following: an indication of LOS, the LOS probability, or both; antenna panel configuration; other conditions; or all of these.

[0161] For example, using an indication of LOS, LOS probability, or both, the WTRU can deactivate the AI / ML model when the LOS probability is below a corresponding threshold for any of the configured / determined beam resources, such as in a first frequency range, e.g., FR1. For example, the WTRU can be configured to identify, report, or identify and report the beam with the best LOS probability, highest LOS probability, or both among multiple beams in the first frequency range (e.g., FR1). Thus, the identified / reported beam can be used at the base station or gNB and / or WTRU for AI / ML beam prediction, e.g., in a second frequency range (e.g., FR2).

[0162] In the example using the antenna panel configuration, the WTRU can deactivate the AI / ML model if there is no identical QCL type D assumption between the antenna port, panel, or both (at the WTRU side) for the first and second frequency ranges (e.g., for FR1 and FR2).

[0163] WTRU can use other conditions to activate / deactivate AI / ML, such as: the number of beams supported, beam attributes, WTRU capabilities, or all of these. For example, using the number of beams supported, WTRU can determine or be configured to use / activate the AI / ML model only for a set of beams in a first frequency range and a second frequency range (e.g., in FR1 and FR2).

[0164] For example, using beam properties, the WTRU can determine or be configured to activate the AI / ML model if one or more determined or configured parameters are within acceptable ranges, such as the line of sight of the antenna array for a first frequency range and a second frequency range, the beam direction of the antenna, or the antenna array configuration, such as at the WTRU, at the gNB or base station, or at both the WTRU and the gNB or base station. Otherwise, the WTRU can determine to deactivate the AI / ML model.

[0165] For example, using WTRU capabilities, when one or more WTRU capabilities are available (e.g., processing time, antenna switching time, BWP switching time, etc.), the WTRU can determine or be configured to activate the AI / ML model. Otherwise, the WTRU can determine to deactivate the AI / ML model.

[0166] This document provides AI / ML model activation / retraining / deactivation / rollback options. One or more of the following configurations can be used for CSI / beam reporting configurations. WTRU can be configured with one or more CSI reporting configurations. Additionally, WTRU can be configured with one or more beam reporting configurations. CSI reporting configuration may include one or more of the following: reporting configuration type (e.g., periodic, semi-persistent on PUCCH, semi-persistent on PUSCH, or aperiodic); reporting quantity (e.g., CRI-RI-PMI-CQI, CRI-RI-i1, CRI-RI-i1-CQI, CRI-RSRP, SSB-Index-RSRP, CRI-RI-LI-PMI-CQI, CRI-SINR, SSB-Index-SINR); reporting frequency configuration; CQI format indicator, such as wideband CQI or subband CQI; PMI format indicator, such as wideband PMI or subband PMI; CSI reporting band; time limit for channel measurements; time limit for interference measurements; codebook configuration; group-based beam reporting; CQI table; subband size; non-PMI port indicator; reporting slot configuration / offset list; CSI reporting period and offset; one or more PUCCH resources for CSI reporting; port index; or any combination of all of these.

[0167] One or more of the following configurations can be used for CSI / beam reporting measurement configurations. The WTRU can be configured with one or more CSI measurement configurations. Additionally, the WTRU can be configured with one or more beam measurement configurations. CSI measurement configurations may include one or more of the following: RS for channel measurements; RS for interference measurements (zero power or non-zero power); report trigger size; a non-periodic trigger state list; a semi-persistent trigger state list on the PUSCH; associated CSI resource configuration; associated CSI reporting configuration; or any combination of all of these. Similar parameters can be included in the beam measurement configuration.

[0168] One or more of the following configurations can be used for CSI resource configuration. A WTRU can be configured with one or more CSI resource configurations. A CSI resource configuration may include one or more of the following: a CSI resource configuration ID; one or more RS resource sets for channel measurements; one or more RS resource sets for interference measurements; a bandwidth portion ID; or a resource type, such as aperiodic, semi-persistent, or periodic.

[0169] In the example solution, the WTRU can activate / apply the AI / ML model, deactivate the AI / ML model, update / retrain the AI / ML model, or trigger a process for selecting one or more new beams. Activation of the AI / ML model can include one or more of the following: activation of one or more RS resources / resource sets associated with the AI / ML model; activation of one or more CSI reporting configurations associated with the AI / ML model; activation of one or more measurement configurations associated with the AI / ML model; activation of one or more CSI resource configurations associated with the AI / ML model; resetting / starting one or more counters associated with the AI / ML model; resetting / starting one or more timers associated with the AI / ML model.

[0170] Deactivation of an AI / ML model may include one or more of the following: deactivation of one or more RS resources / resource sets associated with the AI / ML model; deactivation of one or more CSI reporting configurations associated with the AI / ML model; deactivation of one or more measurement configurations associated with the AI / ML model; deactivation of one or more CSI resource configurations associated with the AI / ML model; resetting one or more counters associated with the AI / ML model; resetting one or more timers associated with the AI / ML model.

[0171] The process for updating / retraining an AI / ML model or its associated parameters / weights may include one or more of the following: transmitting a request / instruction to update the AI / ML model or its associated parameters / weights; resetting / starting one or more counters associated with the process; resetting / starting one or more timers associated with the process; updating the AI / ML model and its associated parameters / weights; applying / using the updated AI / ML model and its associated parameters / weights; or selecting one or more RS / beams based on the updated AI / ML model and its associated parameters / weights. The WTRU may select one or more RS / beams based on quality. For example, the WTRU, base station, or gNB may select one or more RS / beams with the best quality.

[0172] The process for updating / retraining the AI / ML model or associated parameters / weights may also include one or more of the following: measuring one or more selected RS / beams based on the updated AI / ML model and associated parameters / weights; indicating one or more selected RS / beams; or, if the process is unsuccessful, indicating deactivation of the AI / ML model and / or reverting to the regular beam management mechanism. In an example solution regarding indicating one or more selected RS / beams, the WTRU or base station or gNB may indicate one or more selected RS / beams. For example, if the measured quality of one or more selected RS / beams is greater than or equal to (≥) a threshold, the WTRU and / or base station or gNB may indicate one or more selected RS / beams.

[0173] Example solutions include instructing the AI / ML model to be deactivated and / or falling back to the regular beam management mechanism if the process fails. For example, the WTRU may determine that the process failed if one or more of the following conditions are met: a timer (the WTRU may determine that the process failed if the timer associated with the process times out); a counter (the WTRU may increment the counter when it measures a candidate RS and / or the measured quality of the candidate RS is less than one or more first thresholds; if the counter is greater than a second threshold, the WTRU may determine that the process failed); or the measured quality (the WTRU may determine that the process failed if the measured quality of one or more candidate beams is less than a threshold).

[0174] The process for selecting one or more new beams may include one or more of the following: triggering / requesting one or more candidate beam resources; resetting / starting one or more counters associated with the process; resetting / starting one or more timers associated with the process; monitoring / measuring candidate beams / RS; selecting one or more RS / beams; indicating one or more selected beams; or, if the process is unsuccessful, indicating deactivation of the AI / ML model and / or reverting to the regular beam management mechanism. In the example regarding the selection of one or more RS / beams, the WTRU may select one or more beams based on quality. For example, the WTRU, base station, or gNB may select one or more RS / beams with the best quality.

[0175] In an example concerning indicating one or more selected beams, the WTRU may indicate one or more selected beams to the base station or gNB. Furthermore, the WTRU may indicate one or more beams by transmitting one or more UL resources. These one or more UL resources may be one or more of the following: PRACH (e.g., the WTRU may transmit one or more PRACHs (e.g., in PRACH resources associated with one or more selected beams)); PUCCH (e.g., the WTRU may indicate one or more RS indices and / or beam indices (e.g., as part of the CSI) by using one or more PUCCHs (e.g., in PUCCH resources associated with the procedure or one or more selected beams)); or PUSCH (e.g., the WTRU may indicate one or more RS indices and / or beam indices (e.g., as part of the CSI) by using one or more PUSCHs). Additionally, the WTRU may receive acknowledgments from the base station or gNB. For example, the WTRU may receive one or more PDCCHs in one or more CORESET / search spaces associated with the procedure.

[0176] In another example regarding the indication of one or more selected beams, the base station or gNB may indicate one or more selected beams to the WTRU, for example, via one or more of RRC, MAC CE, and DCI. Furthermore, the WTRU may receive the indication based on one or more of the following: TCI status; or beam index. In the example of receiving TCI status, the WTRU may receive an indication of one or more TCI states associated with the selected beam. In the example of receiving beam index, the WTRU may receive an indication of one or more beam indices associated with the selected beam.

[0177] In the example regarding instructing the AI / ML model to be deactivated and / or revert to the regular beam management mechanism if the process fails, the WTRU may determine that the process failed if one or more of the following conditions are met: a timer, a counter, or the measured quality. For example, if a timer associated with the process times out, the WTRU may determine that the process failed. In the counter example, the WTRU may increment the counter when it measures a candidate RS and / or the measured quality of the candidate RS is less than one or more first thresholds. If the counter is greater than a second threshold, the WTRU may determine that the process failed. In the measured quality example, the WTRU may determine that the process failed if the measured quality of one or more candidate beams is less than (<) a threshold.

[0178] The activation, deactivation, updating / retraining of AI / ML models and associated parameters / weights, and triggering of new beam selection processes can be based on one or more of the following: base station or gNB indications, or WTRU indications. In the example solution, the WTRU can receive indications from the base station or gNB for activating / deactivating one or more AI / ML models, updating one or more AI / ML models and associated parameters, or triggering a new beam selection process, such as one or more of RRC signaling, MAC CE, or DCI. Indications can be based on one or more of the following: explicit indications; or indications based on one or more configurations associated with one or more AI / ML models.

[0179] In an example of explicit instruction, the WTRU can receive instructions to activate / deactivate or trigger an update process for one or more AI / ML models or to trigger a beam selection process. Explicit instructions may include one or more of the following.

[0180] For example, the WTRU can receive an indication of the process type. For example, the WTRU can receive one or more of the following: activation, deactivation, update, or new beam selection.

[0181] This instruction can include a signal to trigger the AI / ML model update process. For example, a single bit can indicate whether the AI / ML model update process is triggered. For instance, if the bit is "1", the update process can be triggered. If the bit is "0", the update process cannot be triggered.

[0182] This indication can include a signal to trigger a new beam selection process. For example, a single bit can indicate whether a new beam selection process should be triggered. For instance, if the bit is "1", a new beam selection process can be triggered. If the bit is "0", a new beam selection process cannot be triggered.

[0183] This indication may include AI / ML model IDs. For example, the WTRU may receive one or more AI / ML model IDs to be activated / deactivated / updated. If a new beam selection is triggered, the explicit indication may not include AI / ML model IDs.

[0184] This instruction may include a bitmap of the AI / ML model. For example, each bit of the bitmap may be associated with a specific AI / ML model. For instance, if a bit is "1", the AI / ML model associated with that bit is activated. If the bit is "0", the AI / ML model associated with that bit is deactivated. If a new beam selection is triggered, the explicit instruction may not include the bitmap of the AI / ML model.

[0185] In the example, the instruction may be based on one or more configurations associated with one or more AI / ML models. For example, the WTRU may receive instructions for activation / deactivation / update for one or more configurations. For instance, if the WTRU receives an instruction for activation of a first set of configurations, the WTRU may activate the first set of AI / ML models associated with the first set of configurations. If the WTRU receives an instruction for deactivation of a second set of configurations, the WTRU may deactivate the second set of AI / ML models associated with the second set of configurations. If the WTRU receives an instruction for update of a third set of configurations, the WTRU may update the third set of AI / ML models associated with the third set of configurations. If the WTRU receives an instruction for new beam selection of a fourth set of configurations, the WTRU may select one or more new beams for the third set of AI / ML models associated with the third set of configurations. The one or more configurations may be one or more of the following: CSI reporting configuration; measurement configuration; CSI resource configuration; RS resource configuration; and / or RS resource set configuration.

[0186] The following examples illustrate AI / ML model activation, deactivation, updating / retraining of AI / ML models and associated parameters / weights, and triggering of new beam selection processes based on WTRU indications. In the example solutions, the WTRU can indicate a preferred mode to the base station or gNB, such as one or more of activation, deactivation, updating, and new beam selection. This indication can be based on one or more of the following: explicit indication for all AI / ML models, indication for each AI / ML model, indication for each configuration, or quality measurement.

[0187] WTRU can explicitly indicate the preferred mode for all AI / ML models. For example, an information bit can be used to indicate activation / deactivation. For example, 1 can indicate activation of all AI / ML models or activation of the AI / ML mode, and 0 can indicate deactivation of all AI / ML models or deactivation of the AI / ML mode. For example, an information bit can be used to indicate an update. For example, 1 can indicate an update of all AI / ML models, and 0 can indicate that none of the AI / ML models have been updated. For example, an information bit can be used to trigger a new beam selection process. For example, 1 can indicate the selection of one or more new beams, new RS, or both for all AI / ML models, and 0 can indicate that one or more new beams / RS are not selected.

[0188] WTRU can indicate a preferred mode for each AI / ML model or for each AI / ML model. For example, 1 can indicate activation of the AI / ML model associated with the instruction, and 0 can indicate deactivation of the AI / ML model associated with the instruction. For example, 1 can indicate an update of the AI / ML model associated with the instruction, and 0 can indicate that the AI / ML model associated with the instruction is not updated. For example, 1 can indicate the selection of one or more new beams, new RS, or both for the AI / ML model associated with the instruction, and 0 can indicate that one or more new beams, new RS, or both are not selected for the AI / ML model associated with the instruction.

[0189] WTRU can indicate a specific configuration or a preferred mode for each configuration. For example, 1 can indicate the activation of the AI / ML model associated with that configuration, and 0 can indicate the deactivation of the AI / ML model associated with that configuration. For example, 1 can indicate an update of the AI / ML model associated with that configuration, and 0 can indicate that the AI / ML model associated with that configuration is not updated. For example, 1 can indicate the selection of one or more new beams, new RS, or both for the AI / ML model associated with that indication, and 0 can indicate that one or more new beams, new RS, or both are not selected for the AI / ML model associated with that configuration. The configuration can be one or more of the following: CSI reporting configuration; measurement configuration; CSI resource configuration; RS resource configuration; or RS resource set configuration.

[0190] In the example solution, the WTRU can activate / deactivate / update one or more AI / ML models or trigger a new beam selection procedure based on one or more measured quality parameters. In the example solution, this procedure can be based on thresholds and quality measurements for each procedure, such as the quality reported to the base station or gNB. For example, if the measured quality is greater than or equal to (≥) a threshold, the WTRU can indicate, determine, or both indicate and determine the activation of the AI / ML model. If the measured quality is less than (<) a threshold, the WTRU can indicate, determine, or both indicate and determine the deactivation of the AI / ML model.

[0191] For example, if the measured quality is greater than or equal to (≥) a threshold, the WTRU can indicate, determine, or both indicate and determine that the AI / ML model has not been updated. If the measured quality is less than (<) a threshold, the WTRU can indicate, determine, or both indicate and determine that the AI / ML model has been updated. As another example, if the measured quality is greater than or equal to (≥) a threshold, the WTRU can indicate / determine that there is no new beam selection for the AI / ML model. If the measured quality is less than (<) a threshold, the WTRU can indicate / determine a new beam selection for the AI / ML model.

[0192] In the example solution, the process can be based on two or more thresholds and quality measurements. For example, if the measured quality is greater than or equal to (≥) a first threshold, the WTRU can indicate, determine, or both indicate and determine the activation of the AI / ML model. If the second threshold is less than (<) the measured quality, which may be less than (<) the first threshold, the WTRU can trigger / indicate / determine a new beam selection process. If the measured quality is less than (<) the second threshold, the WTRU can indicate, determine, or both indicate and determine the deactivation of the AI / ML model.

[0193] For example, if the measured quality is greater than or equal to (≥) a first threshold, the WTRU can indicate, determine, or both indicate and determine the activation of the AI / ML model. If the second threshold is less than (<) the measured quality, which may be less than (<) the first threshold, the WTRU can trigger / indicate / determine the AI / ML update process. If the measured quality is less than (<) the second threshold, the WTRU can indicate, determine, or both indicate and determine the deactivation of the AI / ML model.

[0194] In the example solution, the process can be based on two or more quality measurements and two or more thresholds. For example, if the quality of the first measurement (e.g., RSRP, RSRQ, SINR, MCS, or CQI) is greater than or equal to (≥) a first threshold and the quality of the second measurement (e.g., LOS probability) is greater than or equal to (≥) a second threshold, then the WTRU can indicate, determine, or both indicate and determine the activation of the AI / ML model. If the quality of the first measurement (e.g., RSRP, RSRQ, SINR, MCS, or CQI) is less than (<) a first threshold and the quality of the second measurement (e.g., LOS probability) is greater than or equal to (≥) a second threshold, then the WTRU can indicate, determine, or both indicate and determine the update of the AI / ML model. If the quality of the first measurement (e.g., RSRP, RSRQ, SINR, MCS, or CQI) is less than (<) a first threshold and the quality of the second measurement (e.g., LOS probability) is less than (<) a second threshold, then the WTRU may indicate, determine, or both indicate and determine the deactivation of the AI / ML model, trigger a new beam selection process, or both.

[0195] If the WTRU reports two or more measured qualities, the WTRU may instruct activation / deactivation / triggering of a new beamselection process based on one or more of the following: an indication for each measured quality, an indication for all measured qualities, or both. The WTRU may instruct activation / deactivation / triggering of a new beamselection process based on the measured qualities. The WTRU may instruct activation / deactivation / triggering of a new beamselection process based on all measured qualities. The WTRU may determine activation / deactivation / triggering of a new beamselection process based on one or more of the following. For example, the WTRU may determine activation / deactivation / triggering of a new beamselection process based on an average value. Furthermore, in the example where the WTRU may average all measured qualities, the WTRU may determine activation / deactivation / triggering of a new beamselection process, and if the average quality is greater than (≥) a threshold, the WTRU may instruct activation. Additionally, the WTRU may determine activation / deactivation / triggering of a new beamselection process based on multiple measured qualities greater than (≥) a threshold. In addition, the WTRU can determine the activation / deactivation / triggering of a new beam selection process, wherein the WTRU can indicate activation if the quantity of the measured quality is greater than or equal to (≥) a threshold.

[0196] In the example solution, the WTRU can receive acknowledgments of WTRU indications / determinations. For example, the WTRU can receive PDCCHs in one or more CORESET / search spaces associated with the WTRU indication. Alternatively, the WTRU can receive acknowledgment messages via one or more of RRC signaling, MAC CE, or DCI.

[0197] This article provides examples of methods for determining or obtaining the accuracy of AI / ML models. The phrases “accuracy of AI / ML models” and “effectiveness of AI / ML models” are used interchangeably and remain consistent with the examples provided in this article. The phrases “frequency region” or “frequency range” are used interchangeably and remain consistent with the examples provided in this article. The terms “ML,” “AI / ML,” and “AIML” are used interchangeably and remain consistent with the examples provided in this article.

[0198] WTRU can determine the validity or accuracy of an AI / ML model. A WTRU can be configured with resources to perform measurements on it to determine the validity of the AI / ML model. These resources can be located in one or more frequency regions. For example, an AI / ML model can take input from a first frequency region to determine its behavior in a second frequency region. To determine the validity of an AI / ML model, the WTRU can be configured with measurement resources in either the first or second frequency region.

[0199] The WTRU can determine whether the behavior of a second frequency region determined based on an AI / ML model matches the behavior of a second frequency region determined based on measurement resources in that second frequency region. In one example, the WTRU can be configured with a periodic or sparse reference signal in the second frequency region to perform a legacy method (e.g., a non-AI / ML-based method) and compare it with the output of an AI / ML model, which may be based on a reference signal in a first frequency region.

[0200] The validity or accuracy of an AI / ML model can be determined by at least one of the following: the execution of the associated function, the statistical performance of the associated function, the transmission performance of the function associated with the AI / ML model in the frequency region, a comparison of legacy results with AI / ML results, measurements, and / or fault counters. In the example, the validity or accuracy of the AI / ML model can be determined by the execution of the associated function.

[0201] For example, AI / ML models can be used to support, provide feedback on, or enable functions. These functions may include one or more of the following: beam management, CSI reporting, RLM, beam failure detection, persistent listen-before-talk (LBT) failure, mobility, cell (re)selection, random access, or measurement reporting. The WTRU can determine the effectiveness of the AI / ML model based on the performance of the associated functions. The WTRU can be configured with metrics to determine the performance of the associated functions. For example, the WTRU may be configured with an AI / ML model that supports beam management. The WTRU may be configured with metrics such as optimal beam determination. If the AI / ML model determines the optimal beam, the AI / ML model can be considered effective.

[0202] Effectiveness metrics associated with beam management may include at least one of the following: These metrics may include optimal beam prediction. For example, an AI / ML model predicts the optimal beam. These metrics may include predicted beam measurements within a threshold offset from the optimal beam. In one example, the threshold may be configurable. Alternatively, the threshold offset can be used to compare RSRP measurements. These metrics may include N predicted optimal beams that match at least M actual optimal beams. These metrics may include the rate of beam failure detection.

[0203] In the example, the effectiveness or accuracy of an AI / ML model can be determined by the statistical performance of the associated functions. For instance, if a WTRU satisfies the metrics of the associated functions at a certain proportion of times over a period of time, then the WTRU can be considered an AI / ML model.

[0204] In the example, the validity or accuracy of the AI / ML model can be determined by the execution of transmissions in the frequency region of the function associated with the AI / ML model. For example, the WTRU can determine the validity of the AI / ML model with associated functions in a second frequency region based on the execution of transmissions in a second frequency region. The execution of transmissions can be determined based on at least one of the following: BLER, assumed PDCCH BLER, HARQ-ACK / NACK (NACK) performance or ratio, latency, throughput, spectral efficiency, probability of interruption, etc.

[0205] In the example, the validity or accuracy of the AI / ML model can be determined by comparing legacy results (e.g., results based on non-AI / ML) with AI / ML results. For instance, the WTRU can perform measurements in the frequency region of the associated function to compare with the output of the AI / ML model. The WTRU can determine the validity of the AI / ML model based on the difference between the output of the AI / ML model and the output of the associated function based on measurements in the applicable frequency region (e.g., using a legacy method).

[0206] In the example, the effectiveness or accuracy of the AI / ML model can be determined by measurements. For instance, the WTRU can determine effectiveness based on measurements performed on RS. These measurements may include at least one of the following: RSRP, RSSI, RSRQ, CSI, CQI RI, PMI, LI, CRI, channel occupancy (CO), LOS probability, Doppler drift, Doppler spread, average delay, or delay spread. The WTRU can compare at least one measurement to one or more thresholds to determine the model's accuracy or effectiveness.

[0207] In the example, the effectiveness or accuracy of the AI / ML model can be determined by a failure counter. The WTRU can count the number of times the AI / ML model fails. For example, the WTRU can count the number of times an associated function of the AI / ML model fails. Alternatively, the WTRU can count the number of times prediction deviations exceed a (potentially configurable) threshold. The counter can be valid for a period of time. At the end of this period, the counter can be reset. This period can be fixed or configurable. When a failure occurs, the WTRU can start or restart the period.

[0208] When N outputs (where N is configurable) of an AI / ML model are considered accurate (e.g., predictions are within a configurable threshold of actual values), the WTRU can stop for a period of time or reset the counters. When the period of time has elapsed, the WTRU can determine the accuracy or validity of the AI / ML model based on the counter values. Alternatively, the WTRU can determine the accuracy or validity of the AI / ML model based on a fault counter that has reached a specific value. For example, if the fault counter reaches X, the WTRU can consider the model invalid.

[0209] The WTRU can report the validity of AI / ML models to the base station or gNB. The WTRU can report one of two states: valid or invalid. Similarly, the WTRU can report accuracy metrics for the AI / ML models. Accuracy metrics indicate the validity value of the AI / ML model. Validity values ​​provide accuracy parameters for the AI / ML model.

[0210] The validity of an AI / ML model can be reported in PUCCH resources, PUSCH resources, RACH, RRC messages, or MAC CE. New messages can be used to report the validity of an AI / ML model. Alternatively, the validity of an AI / ML model can be implicitly reported, for example, by indicating a failure of an associated function (e.g., beam fault detection) via a WTRU. Such fault reports can include new elements indicating that the failure is due to the AI / ML model no longer being valid.

[0211] The WTRU can request resources, such as DL reference signals, to determine the validity of the AI / ML model. The WTRU can indicate to the base station or gNB the type of resources required, the AI / ML model (e.g., AI / ML model index), and associated functions.

[0212] WTRU can be configured to determine the accuracy of an AI / ML model. When WTRU can determine the accuracy of an AI / ML model, this configuration can include a collection of periodic time instances. The configuration can also include, for example, reporting resources associated with one or more periodic time instances, enabling WTRU to report the accuracy of the AI / ML model.

[0213] The WTRU can also be dynamically triggered to determine and potentially report the accuracy of the AI / ML model. The WTRU can receive triggers in DL signals such as DCI, MAC CE, or RRC commands. The WTRU can be configured with one or more triggers to determine the accuracy of the AI / ML model.

[0214] The WTRU can be triggered to determine the accuracy of the AI / ML model by at least one of the following: time, timer, reception of RS signals, indication from the base station or gNB, execution of functions associated with the AI / ML model, execution of transmission in the frequency region of functions associated with the AI / ML model, beam fault detection or radio link fault determination, cell activation or deactivation, BW change, cell change, measurement and / or fault counter.

[0215] In the example, WTRU can be triggered to determine the accuracy of an AI / ML model over time. For instance, WTRU can be triggered to determine the accuracy of an AI / ML model at a specific time instance (e.g., a time slot, subframe, or symbol).

[0216] In the example, the WTRU can be triggered to determine the accuracy of the AI / ML model via a timer. For example, the WTRU can be triggered to determine the accuracy of the AI / ML model when the timer expires, or after a set number of time instances, slots, subframes, or symbols. The WTRU can start or restart the timer after determining the accuracy of the AI / ML model. The WTRU can start or restart the timer based on signaling from the base station or gNB. The WTRU can also start or restart the timer based on the execution of functions associated with the AI / ML model. For example, if the AI / ML model is used for beam prediction, the WTRU can start or restart the timer if the prediction is determined to be within the required range.

[0217] In the example, the WTRU can be triggered to determine the accuracy of the AI / ML model by receiving the RS signal. For example, the WTRU can be triggered to determine the accuracy of the AI / ML model based on the reception of the RS intended for AI / ML model accuracy determination.

[0218] In the example, the WTRU can be triggered to determine the accuracy of the AI / ML model by executing functions associated with the AI / ML model. For example, if the AI / ML model is used for beam prediction, the WTRU can be triggered to determine the accuracy of the AI / ML model if the prediction is determined to be outside the acceptable range. Other examples of the execution of functions associated with the AI / ML model from the part that determines the validity or accuracy of the AI / ML model may be applicable here.

[0219] In the example, the WTRU can be triggered to determine the accuracy of the AI / ML model by the execution of transmissions in the frequency region associated with the AI / ML model's functionality. For example, the WTRU can be triggered to determine the validity or accuracy of an AI / ML model with associated functionality in a second frequency region based on the performance of transmissions in that second frequency region. Transmission performance can be determined based on at least one of the following: BLER, assumed PDCCH BLER, HARQ-ACK / NACK performance or ratio, latency, throughput, spectral efficiency, outage probability, etc.

[0220] In the examples, WTRU can be triggered to determine the accuracy of an AI / ML model based on cell changes. In one example, the cell change could originate from cell handover (HO). Alternatively, it could originate from cell selection. In another example, it could originate from cell reselection.

[0221] In the example, the WTRU can be triggered to determine the accuracy of an AI / ML model through measurements. For example, the WTRU can be triggered to perform the determination of the accuracy of an AI / ML model based on measurements of RS. These measurements may include at least one of the following: RSRP, RSSI, RSRQ, CSI, CQI RI, PMI, LI, CRI, CO, LOS probability, Doppler drift, Doppler spread, average delay, or delay spread.

[0222] In the example, the WTRU can be triggered to determine the accuracy of an AI / ML model via a fault counter. The WTRU can count the number of times the AI / ML model fails. For example, the WTRU can count the number of times an associated function of the AI / ML model fails. Alternatively, the WTRU can count the number of times prediction deviations exceed a (potentially configurable) threshold. The counter can be valid for a period of time. At the end of this period, the counter can be reset. This period can be fixed or configurable. When a fault occurs, the WTRU can start or restart the period.

[0223] When N outputs of the AI / ML model (where N is configurable) are considered accurate (e.g., predictions are within a configurable threshold of actual values), the WTRU can stop for a period of time or reset the counter. When the period of time expires, the WTRU can be triggered to determine the accuracy of the AI / ML model based on the counter value. Alternatively, the WTRU can be triggered to determine the accuracy or validity of the AI / ML model based on a fault counter that has reached a specific value. For example, if the fault counter reaches X, the WTRU can be triggered to determine the accuracy of the AI / ML model.

[0224] In the example, WTRU may engage in the following behaviors when determining the accuracy of an AI / ML model: WTRU may determine appropriate behavior based on the determined accuracy or validity of the AI / ML model. WTRU behavior may depend on the method used to determine the accuracy or validity of the AI / ML model. WTRU behavior may be determined based on one or more of the measurements or measurements compared to a threshold, or a combination thereof. Measurements may be triggered based on the determination of the accuracy or validity of the AI / ML model. These measurements may include at least one of the following: BLER, hypothetical PDCCH BLER, RSRP, RSSI, RSRQ, CSI, CQI RI, PMI, LI, CRI, CO, LOS probability, Doppler drift, Doppler spread, average delay, or delay spread. WTRU behavior may include at least one of the following: continuing to use the AI / ML model, selecting a secondary output of the AI / ML model, updating or training the AI / ML model, and / or stopping the use of the AI / ML model and using or reverting to a legacy method for the associated functionality.

[0225] Examples in this article include situations where WTRU behavior includes continuing to use the AI / ML model. For instance, if the AI / ML model is considered accurate or effective, WTRU may continue to use it. If the accuracy of the AI / ML model is greater than a threshold, WTRU may consider the AI / ML model accurate.

[0226] Examples in this article include cases where WTRU behavior includes selecting a secondary output of an AI / ML model. For instance, if an AI / ML model is considered averagely accurate but incorrect for a particular result, WTRU can select a secondary output (if available).

[0227] For example, if for the predicted beam resources in FR2, the LOS probability is higher than a first threshold (e.g., LOS_th), and the derived CQI is lower than a corresponding threshold (e.g., CQI_th), then the WTRU can update or retrain the AI / ML model. As another example implementation, if for the predicted beam resources in FR2, the LOS probability is lower than a first threshold (e.g., LOS_th), and the derived CQI is lower than a corresponding threshold (e.g., CQI_th), then the WTRU can, for example, determine the secondary output of the AI / ML model to select or fall back to older operations based on the determined LOS probability.

[0228] The ML model used for beam selection and / or prediction can be located at the WTRU and / or the network. In a first exemplary solution where the ML model is located at the WTRU, the WTRU may be configured with one or more use cases, which may include one or more subsets of use cases for which the ML model can be used. The WTRU may also be configured to perform and potentially report measurements to determine whether the ML model is suitable for use. Use cases / parameters for determining whether the WTRU can use the ML model may include any or more of the following: LOS / NLOS indication / probability, changes to LOS / NLOS indication / probability, signal-to-noise ratio (SNR) / SINR measurement / calculation, additional channel measurements or changes to channel measurements, the number of supported FR1 and FR2 beams, changes to bandwidth portion (BWP), WTRU capabilities, network assistance, antenna panel configuration at the WTRU, other antenna parameters, and / or model validity / accuracy.

[0229] The examples provided in this article include how the WTRU uses LOS / NLOS indications / probabilities to determine whether to use an ML model. In one example, a base station or gNB can configure multiple beams in a first frequency range (e.g., FR1) to find the beam with the optimal LOS probability, making it more likely to exceed a pre-configured LOS probability. In such scenarios, the WTRU can use only the FR1 beam with the optimal LOS probability (such as above a pre-configured threshold) as input to the ML model.

[0230] In one example, the WTRU can determine that the LOS indication is negative or that the LOS probability is below a pre-configured threshold for any CSI-RS resource in the CSI-RS resource. The WTRU can determine to deactivate the AI / ML model and resort to legacy beam management procedures. In another example, the WTRU can make this determination based on historically poor performance of the ML model in NLOS scenarios previously observed by the WTRU.

[0231] The examples provided in this article include how the WTRU uses changes in LOS / NLOS indications / probabilities to determine whether to use the ML model. In one example, the WTRU can determine to activate / deactivate the ML model based on changes in LOS / NLOS conditions. For instance, if the LOS indication changes from "1" to "0" to indicate a loss of LOS, the WTRU can determine to deactivate the ML model and switch back to the legacy beam management procedure.

[0232] For example, the WTRU can measure a sudden drop in LOS / NLOS probability. A drop below a threshold pre-configured by the WTRU, base station, or gNB can trigger the WTRU to deactivate the ML model and switch to the legacy beam management process.

[0233] The examples provided in this document include the WTRU using SNR / SINR measurements / calculations to determine whether to use the ML model. The WTRU can be configured to perform / calculate SNR / SINR measurements, such as SS-SINR or CSI-SINR. The WTRU can determine to deactivate the ML model for FR2 beam selection based on a drop in the SNR / SINR measurement / calculation below a configured or pre-configured threshold. The threshold can also be based on a drop / difference in the SNR / SINR value rather than its absolute value, such that a drop / difference exceeding the threshold can trigger the WTRU to switch to a legacy procedure, which in one example could be a legacy beam management procedure.

[0234] For example, the WTRU can be configured with SNR / SINR ranges, where the use of a second frequency range (e.g., FR2, ML selection / prediction model) is based on measurement inputs in a first frequency range (e.g., FR1) to provide the optimal output for beam prediction. Measurements / calculations of SNR / SINR outside the pre-configured range can trigger the WTRU to deactivate the ML model and revert to the traditional framework, which in one example could be a traditional beam management framework.

[0235] The examples provided in this article include the WTRU using additional channel measurements, such as channel coherence time, channel coherence bandwidth, Doppler spread, BLER, etc., or changes to channel measurements, to determine whether to use the ML model. The WTRU can perform additional channel measurements, such as channel coherence time, channel coherence bandwidth, Doppler spread, BLER, etc. Measurements of channel conditions and / or changes to those measurements can constitute a trigger for using or not using the ML model.

[0236] In one example, the WTRU measures and records large channel coherence times (such as exceeding a configured or pre-configured threshold) that may indicate a slow-fading channel, causing the WTRU to activate an ML model predicting the optimal FR2 beam based on FR1 beam information / measurements. Conversely, channel coherence times below a pre-configured threshold may lead to poor performance of the FR2 prediction model due to fast fading conditions. In this case, the WTRU can deactivate the model and resort to conventional methods for FR2 beam selection.

[0237] In one example, the WTRU can measure a sudden change in any of the channel parameters (e.g., channel coherence time, channel coherence bandwidth, Doppler spread, BLER, etc.). In one example, a change in channel coherence time from a large measurement to a smaller value could indicate a sudden deterioration in channel conditions, allowing the WTRU to determine that the channel conditions are no longer sufficiently effective / stable to use the FR2 beam ML predictor, and thus revert to / back to the older method of FR2 beam selection.

[0238] In one example, the WTRU can be configured with a corresponding range for any of the channel parameters (e.g., channel coherence time, channel coherence bandwidth, Doppler spread, BLER, etc.) such that when the channel measurement is within the configured / pre-configured / determined range, the WTRU can determine to activate only the FR2 beam selection / prediction ML model.

[0239] The examples provided in this document involve the WTRU using the number of supported FR1 and FR2 beams to determine whether to use an ML model. The WTRU can activate / deactivate the ML model based on the number of supported beams in a first and second frequency range. In one example, a lower number of supported beams in the second frequency range (e.g., FR2) can trigger the WTRU to deactivate the ML model used for predictions in the second frequency range (e.g., FR2) because the WTRU can determine that, with fewer supported beams, the legacy measurement method selects the optimal FR2 beam. In such scenarios, the minimum number of supported beams in the second frequency range (e.g., FR2) that would trigger the use of the ML model can be determined by the WTRU using historical data (e.g., past validation of the ML model accuracy relative to the number of supported FR2 beams).

[0240] The examples provided in this article illustrate how WTRU uses changes to the BWP to determine whether to use the ML model. WTRU can determine to deactivate the ML model after a BWP change / switch. In one example, when WTRU changes / switches the BWP due to a timer timeout, WTRU can determine that the model is no longer suitable for the new BWP.

[0241] The examples provided in this paper include WTRUs using their capabilities to determine whether to use an ML model. A WTRU can determine whether to use an ML model based on its capabilities. In one example, a WTRU with reduced capabilities and / or no ML capabilities may not be configured with any ML model and may have to use older methods, such as older methods for beam selection. Similarly, a lower-capacity WTRU may be able to use an ML model for current beam selection / determination in a second frequency range (e.g., FR2) based on beam measurements in a first frequency range (e.g., FR1), but may not be able to predict future beams based on current measurements because the WTRU performs fewer measurements (e.g., a larger number of measurements compared to a WTRU with four (4) receive antennas). In one example, a lower-capacity WTRU may be a WTRU with two (2) receive antennas compared to a WTRU with four (4) receive antennas.

[0242] The examples provided in this document include the WTRU using network assistance to determine whether to use an ML model. The network can transmit assistance to the WTRU based on an indication of when the WTRU can activate its ML model. During registration, the network can transmit a "WTRU Capability Enquiry" to the WTRU specifying which capability the WTRU wants to report. One such capability could be whether the WTRU has ML capabilities. This indication could be a single bit / tag type indicator, where the WTRU reports "1" if it is configured with an ML model and "0" otherwise, or the WTRU could have additional parameters to be reported, such as the SNR range for which the ML model is activated at the WTRU. Assistance regarding when to activate the appropriate ML model can be provided to the WTRU based on measurements from the base station or gNB (such as based on channel conditions).

[0243] The examples provided in this document involve the WTRU using the antenna panel configuration at the WTRU to determine whether to use an ML model. In one example, the antenna panel configuration between the antenna ports in a first frequency range and a second frequency range may be incompatible, such that measurements taken in the first frequency range (e.g., FR1) may not result in the selection of the optimal beam in the second frequency range (e.g., FR2). For example, the same QCL type D assumptions may not exist between the antenna ports and / or panels for different frequency ranges, such as on the WTRU side. The WTRU may determine to deactivate the ML prediction model and revert to / back to a legacy procedure, such as an older beam selection procedure.

[0244] The examples provided in this article include how the WTRU uses other antenna parameters to determine whether to use the ML model. The WTRU may determine to activate / deactivate / retrain its ML model based on other antenna parameters, such as the line of sight of the antenna array at the WTRU and / or the base station or gNB, the beam direction of the antenna, or one or more of the antenna array configuration for a first frequency range and / or a second frequency range. In one example, a change in the line of sight of the antenna array or the beam direction of the antenna may trigger the WTRU to deactivate its ML model, for example, because the model may need to be retrained for the new beam direction.

[0245] The examples provided in this document include WTRU using model validity / accuracy to determine whether to use an ML model. WTRU can activate / deactivate an ML model based on model validity / accuracy, and WTRU can determine model validity / accuracy using any methods as explained elsewhere in the implementations and examples herein. Activation / deactivation of the ML model can be triggered by any triggers as explained elsewhere in the implementations and examples herein.

[0246] In any use case involving determining that a beam selection / prediction method is no longer effective, leading to the adoption of another beam selection / prediction method (e.g., switching from an ML model to a legacy (beam management) procedure), the WTRU can be configured with procedures for a smooth transition. In one example, a temporary procedure might involve triggering a time window after the WTRU determines the transition to the legacy procedure, prior to the ML model being deactivated, to allow for the timing of measurements to the base station or gNB, such as SS and / or CSI-RS measurements and / or SRS.

[0247] In an exemplary solution where the ML model is located at the base station or gNB, the WTRU can provide feedback / reports to the base station or gNB, for example, in the UCI, regarding the accuracy / quality of the beam selected by the base station or gNB. Following the feedback from the WTRU, the base station or gNB can determine whether to continue using the ML model for beam selection / prediction, deactivate the ML model, retrain the ML, or revert to / back to older / non-ML methods, such as those used for beam selection.

[0248] In one example, the WTRU can be configured to perform RSRP measurements on a beam selected by the ML model in a second frequency range (e.g., FR2) at the base station or gNB. This can be based on input / reported data / information from the WTRU in a first frequency range (e.g., FR1). If the WTRU measures an RSRP value below a threshold, the WTRU can report the measurement to the base station or gNB, which can revert to a legacy (beam management) procedure. In one example, the threshold can be one or more of those configured, pre-configured, determined, or predetermined by the WTRU or the base station or gNB.

[0249] In one example, when using an ML model at a base station or gNB, the WTRU can be configured to perform additional measurements, particularly at the beginning of the calibration period, to ensure that the ML model at the base station or gNB is properly calibrated. The WTRU can be configured to report all channel measurement results, or to report channel measurement results only through the network when they are above / below a (pre-)configured / determined threshold, or to report only changes in channel measurement results that are below / above a (pre-)configured / determined threshold.

[0250] This paper provides implementation schemes and examples for dynamic retraining / updating of AI / ML models. Furthermore, it provides examples of iterative retraining / updating of AI / ML models based on the beam output predicted by the AI / ML.

[0251] This document provides examples of AI / ML model configuration. A WTRU can be configured with an AI / ML model to perform predictions of beam resources and / or beam resource attributes associated with a second frequency band (e.g., FR2) based on beam resources and / or beam resource attributes in a first frequency band (e.g., FR1). In this document, beam resources may consist of TCI states for downlink, CSI-RS or SSB, SRS resources, or TCI states for uplink. Beam resource attributes may be any CSI associated with the beam resource, including but not limited to CQI, PMI, RI, RSRP, SNR, SINR, LosS or NLoS information, CIR, or any associated statistics. In the example solution, the WTRU can apply the measured FR1 beam resource attributes as input to the AI / ML model and obtain the optimal beam resource and / or beam resource attributes associated with FR2 as output. In one example, the beam resource attributes of the first frequency band (e.g., FR1) may include one or more of RSRP, CSI, PMI, CIR, Loss probability, etc.

[0252] In the example solution, the WTRU can be configured to report the output of the AI / ML model to the base station or gNB. For example, the WTRU can measure the FR1 beam resource, apply it as input to the AI / ML model, obtain the predicted RSRP of the FR2 beam resource as the output of the AI / ML model, and report the predicted RSRP of the FR2 beam resource to the base station or gNB.

[0253] This document provides examples of monitoring / verifying the accuracy of AI / ML models. A WTRU can be configured to determine the accuracy of an AI / ML model. The mechanism for determining the accuracy of an AI / ML model may depend on the specific functions that the AI / ML model can support. For example, this function could be beam management, CSI feedback generation, beam failure and / or radio link failure determination, mobility, measurement reporting, etc. For instance, the WTRU can compare, for example, a predicted value of FR2 beam resources from the output of an AI / ML model with, for example, an actual value of FR2 beam resources based on measurements of FR2 beam resources. Some example methods for determining the accuracy of AI / ML models are described in implementations and examples elsewhere in this document.

[0254] In the example solution, the WTRU can be configured with an accuracy threshold for operations performed by the AI / ML model. For example, such an accuracy threshold can be configured semi-statically via RRC configuration. In possible examples, this accuracy threshold can be signaled as part of the AI / ML model configuration. Alternatively, the accuracy threshold can be signaled within a MAC control element. It is possible that this accuracy threshold can be signaled along with an AI / ML model activation command. When the model is active, the WTRU can be configured to monitor the accuracy of the AI / ML model. This monitoring may be performed over a pre-configured time period. In one example solution, if the AI / ML model's accuracy does not meet a pre-configured accuracy threshold, the WTRU can autonomously deactivate the AI / ML model and send a report to the network. The WTRU may be configured to fall back to a legacy approach for functions performed by the AI / ML model. The WTRU may also initiate retraining of the AI / ML model.

[0255] In the example solution, the WTRU can be configured to periodically retrain the AI / ML model. For example, the WTRU can be configured to start a timer with pre-configured values. When the timer expires, the WTRU can trigger retraining of the AI / ML model. The WTRU can restart the timer when the retraining process is complete. In another example solution, the WTRU can restart the timer when AI / ML model retraining is successfully executed due to event-based triggering. The periodic AI / ML model retraining timer can be configured as part of the AI / ML model configuration or as part of an AI / ML model trigger.

[0256] In another example solution, the WTRU can be configured to perform retraining of the AI / ML model based on a pre-configured trigger. In this example solution, the trigger can be based on AI / ML model accuracy. For example, the WTRU can be configured to determine the accuracy of the AI / ML model based on one or more triggers as outlined in the embodiments and examples provided elsewhere in this document. If the determined accuracy is below a pre-configured accuracy threshold, the WTRU can trigger retraining of the AI / ML model. In another example solution, the trigger can be based on AI / ML performance. For example, the WTRU can be configured to monitor AI / ML model performance, for instance, based on metrics associated with the functionality enabled by the AI / ML model. Performance metrics could include one or more of the following: reported BLER performance on FR2 beams above or below a threshold, HARQ-ACK or HARQ-NACK ratio, L3 (e.g., RSRP, RSSI, RSRQ, CO) or L1 measurements (e.g., RI, PMI, CQI, LI, CRI, RSRP).

[0257] In another example solution, the trigger could be based on configuration changes. For example, configuration changes could include RS configuration, bandwidth portion configuration, SCell configuration, etc. In yet another example solution, the trigger could be based on mobility events. For example, mobility events could include changes in serving cell / TRP due to HO / Conditional Handover (CHO) / Dual Active Stack (DAPS) handover, radio link failure (RLF), etc. In the example solutions, the trigger condition could be based on network commands. For example, the WTRU could receive implicit or explicit instructions in a deactivation command associated with the AI / ML model, where the deactivation command could instruct the WTRU to retrain the AI / ML model. In possible examples, the deactivation command could additionally indicate the configuration of resources suitable for retraining.

[0258] This document provides examples of iterative processes for training, retraining, or both of AI / ML models. When one or more triggers for training, retraining, or both of an AI / ML model are met, the WTRU can indicate to the network that a retraining process has been triggered, and additionally provide a reason for the retraining. For example, the reason can be expressed as a reason value, and different code points in the reason value can be associated with different triggering conditions. In possible examples, the indication may include AI / ML performance and / or accuracy values. The WTRU may include assistive information to enable the network to configure resources for retraining. For example, assistive information may include one or more of the following: the number of beam resources for retraining, periodicity, the density of RS in the frequency domain, etc.

[0259] WTRU can train, retrain, or both train and retrain AI / ML models based on configuration parameters received from a base station or gNB. For example, such configuration parameters may include one or more of the following: configuration of FR1 beam resources, configuration of FR1 beam resource attributes, configuration of FR2 beam resources, configuration of FR2 beam resource attributes, configuration of parameterized loss functions, thresholds, etc. For example, the loss function may be associated with a metric that indicates the difference between the predicted and actual FR2 beam resources as cross-entropy loss, hinge loss, squared hinge loss, mean square error, etc.

[0260] In the example solution, the WTRU can determine whether retraining is successful based on pre-configured criteria and indicate to the base station or gNB that retraining is complete. For example, the WTRU can determine that retraining is complete when the accuracy of the retrained AI / ML model is higher than a pre-configured threshold.

[0261] In the example solution, the WTRU can be configured to iteratively retrain the model until the AI / ML model output is updated. For example, the WTRU can be configured to use measured CSI parameters (e.g., CQI, PMI, CIR, etc.) in the retrained / updated beam prediction AI / ML model and can determine one or more optimal FR2 beams, for example, based on RSRP. The WTRU can determine that retraining is complete when the model retraining has resulted in different outputs (e.g., predicted FR2 beams different from those of the AI / ML model before retraining). The WTRU can be configured to indicate retraining success to the network via a MAC control element, a pre-configured PUCCH resource, or a preamble resource. Optionally, the WTRU can indicate the accuracy value of the AI / ML model in the retraining success indication.

[0262] In the example solution, the WTRU can determine retraining failure based on pre-configured criteria. For example, if retraining is unsuccessful within a pre-configured time period, the WTRU can declare retraining failure. For example, if the AI / ML model accuracy does not improve beyond a pre-configured accuracy threshold, the WTRU can declare retraining failure. For example, if the model output before and after retraining results in the same output, such as the same predicted FR2 beam, the WTRU can declare retraining failure. For example, if the beam resources configured for training are no longer available, such as if the beam resources can be released / deactivated by the network or due to obstruction or WTRU mobility, and the AI / ML model accuracy remains below a pre-configured accuracy threshold, the WTRU can declare retraining failure. Upon retraining failure, the WTRU can deactivate the AI / ML model (if active) and fall back to the regular beam management mechanism. The WTRU can be configured to send a retraining failure indication to the network via a MAC control element. In another example solution, the WTRU can send the retraining failure indication via a pre-configured PUCCH or a pre-configured preamble resource.

[0263] Figure 3 This is a flowchart illustrating an example of a verification process for beam prediction based on hierarchical spatial relationships. In the example shown in flowchart 300, the WTRU can perform measurements on parameters of a first set of beam resources, and the WTRU can then predict beam resources 310 from a second set of beam resources based on the measured parameters of the first set of beam resources. In one example, the base station can use the first set of beam resources to transmit to the WTRU. The WTRU can then report the predicted beam resources. In another example, the WTRU can report these predicted beam resources to the base station.

[0264] In another example, the WTRU may receive one or more thresholds 315 for accuracy verification. These thresholds can be used when measuring the first set of beam resources. In this example, the thresholds for accuracy verification may include one or more of the following: CQI threshold, RSRP threshold, SINR threshold, LOS probability threshold, assumed BLER threshold, etc. In one example, the WTRU may receive one or more thresholds from the base station.

[0265] Furthermore, the WTRU can receive one or more signals 320 on one or more channels based on beam resources in a second set of beam resources. Additionally, the WTRU can measure one or more accuracy parameters of one or more signals. In one example, the WTRU can receive one or more signals from a base station.

[0266] In Example 325, the one or more signals may include one or more PDCCH signals. Additionally, the one or more signals may include one or more PDSCH signals. As another example, the one or more signals may include one or more CSI-RS signals. Similar signals may also be received in other examples. Furthermore, one or more channels may include one or more PDCCH signals. Additionally, one or more channels may include one or more PDSCH signals. Similar channels may also be used for reception in other examples.

[0267] The WTRU can further perform a verification process and determine whether one or more measured accuracy parameters are within acceptable limits. 330 If one or more measured accuracy parameters are within acceptable limits, the WTRU can determine that the AI / ML model is valid. 340 Furthermore, upon determining that the AI / ML model is valid, the WTRU can activate the use of the AI / ML model to predict the optimal beam. In an additional or alternative example, upon determining that the AI / ML model is valid, the WTRU can continue using the AI / ML model to predict the optimal beam. Furthermore, the WTRU can perform transmission, reception, or both based on the predictions for the determined optimal one or more beams. 390

[0268] If one or more measured accuracy parameters are outside the acceptable range, the WTRU can determine that the predicted beam is invalid (350). Furthermore, the WTRU can select one or more other predicted beams from one or more other candidates. Additionally, the WTRU can restart the verification process. In one example, the beam-specific accuracy parameters may be outside the acceptable range because the measured probability of one or more LOS parameters is below the LOS threshold and one or more channel parameters are below the channel parameter threshold (355). In one example, one or more channel parameters may include one or more CQI parameters.

[0269] If the WTRU restarts the calibration process 350 and the timer has not yet timed out 335, the WTRU can re-determine whether one or more measured accuracy parameters are within acceptable limits 330. The WTRU can then continue using the calibration process. If the timer has timed out 375, the WTRU can fall back to legacy beam management 370.

[0270] During the validation process, if one or more measured accuracy parameters are outside the acceptable range, the WTRU can determine that the AI / ML model is invalid. Furthermore, the WTRU can then update the AI / ML model, retrain it, or perform both. Additionally, the WTRU can predict new beams. Furthermore, the WTRU can restart the validation process. In one example, beam-specific accuracy parameters may be outside the acceptable range because the measured probability of one or more LOS parameters is above the LOS threshold, while one or more channel parameters are below the channel parameter threshold. In one example, one or more channel parameters may include one or more CQI parameters.

[0271] If the WTRU restarts the calibration process 360 and the timer has not yet timed out 335, the WTRU can re-determine whether one or more measured accuracy parameters are within acceptable limits 330. The WTRU can then continue using the calibration process. If the timer has timed out 375, the WTRU can fall back to legacy beam management 370.

[0272] If the WTRU reverts to legacy beam management, it can deactivate the AI / ML model and then use legacy beam management to determine the optimal beam 370. Furthermore, the WTRU can perform transmit, receive, or both based on predictions for one or more of the determined optimal beams 390.

[0273] Figure 4 This is a flowchart illustrating an example of predicted beam management. In the example shown in flowchart 400, the WTRU can perform a measurement 410 on a first set of beam resources. In one example, the base station can use the first set of beam resources to transmit to the WTRU. In one example, the first set of beam resources can be FR1 beam resources. The WTRU can then predict beam resources in a second set of beam resources based on the measurement of the first set of beam resources 420. In one example, the second set of beam resources can be FR2 beam resources. Furthermore, the WTRU can report the predicted beam resources 430. In one example, the WTRU can report the predicted beam resources to the base station.

[0274] Furthermore, the WTRU can use the first beam to receive one or more first signals 440. In one example, the WTRU can receive one or more first signals from the base station. In one example, the first beam can use beam resources from a second set of beam resources.

[0275] Furthermore, the WTRU can perform measurements 450 on one or more accuracy parameters of the received one or more first signals. Additionally, provided that the measured one or more accuracy parameters of the received one or more first signals are acceptable, the WTRU can use the first beam to transmit one or more second signals 460. In one example, these accuracy parameters may be acceptable when the measured LOS is higher than a LOS threshold and the CQI is higher than a CQI threshold. In one example, the WTRU can transmit one or more second signals to a base station.

[0276] In another example, provided that one or more accuracy parameters of the received one or more first signals are acceptable, the WTRU can use the first beam to receive one or more third signals. In one example, the WTRU can receive one or more third signals from a base station.

[0277] In one example, one or more received first signals may be PDCCH signals. Alternatively, one or more received first signals may be CSI-RS.

[0278] Furthermore, in one example, using the first beam can include activating the first beam. Alternatively, using the first beam can include continuing to use the first beam.

[0279] In another example, the one or more accuracy parameters may include one or more line-of-sight (LOS) parameters. Alternatively, the one or more accuracy parameters may include one or more channel parameters. Furthermore, the one or more accuracy parameters may include one or more CQI parameters.

[0280] In the additional examples, WTRU can also activate the AI / ML model to predict one or more second beams. In one example, one or more second beams can use beam resources from a second set of beam resources. In the additional or alternative examples, WTRU can continue to use the AI / ML model to predict one or more second beams.

[0281] In an additional example, if one or more measured accuracy parameters of one or more received first signals are unacceptable, the WTRU may send a request to select and report a third beam. In one example, the measured LOS may be below a LOS threshold, and the measured CQI may be below a CQI threshold. In one example, the WTRU may send the request to the base station. In another example, the base station may respond to the request. Therefore, the WTRU can select a third beam. Furthermore, the WTRU can report the third beam to the base station.

[0282] For example, the WTRU may send a request if one or more accuracy parameters of one or more received first signals are unacceptable. In one example, the measured LOS may be higher than the LOS threshold, and the measured CQI may be lower than the CQI threshold. The sent request may include a request to update the AI / ML model. The sent request may include a request to retrain the AI / ML model. Furthermore, the sent request may include a request to use the AI / ML model to predict and report the fourth beam. In one example, the WTRU may send the request to the base station. Furthermore, in one example, the base station may respond to the request. Thus, the WTRU may update the AI / ML model. In additional or alternative examples, the WTRU may retrain the AI / ML model. Furthermore, the WTRU may use the AI / ML model to predict the fourth beam. Furthermore, the WTRU may report the fourth beam. In one example, the WTRU may report the fourth beam to the base station.

[0283] Furthermore, if one or more accuracy parameters of the received first signal are unacceptable, the WTRU can fall back to a non-AI / ML beam management procedure to select and report a fifth beam. In one example, the measured CQI may be below the CQI threshold, and multiple time instances may have elapsed since the first signal was received using the first beam. In one example, the WTRU can then select a sixth beam. Furthermore, the WTRU can report the sixth beam. In one example, the WTRU can report the sixth beam to the base station.

[0284] In another example, the WTRU can use one or more sixth beams to receive one or more fourth signals, and can measure one or more accuracy parameters of the received one or more fourth signals even if the measured accuracy parameters of the received one or more first signals are unacceptable. In one example, the measured LOS may be below the LOS threshold, the measured CQI may be below the CQI threshold, and there may not have been multiple time instances since the first signal was received using the first beam. In one example, the WTRU can receive one or more fourth signals from the base station.

[0285] This paper provides an example of dynamic retraining / updating of an AI / ML model based on changes in the set of activated / deactivated TCI states. TCI states provide the QCL information necessary for the WTRU to receive various reference signals and / or channels. The WTRU can be configured with several TCI states (e.g., via RRC signaling), and a subset of the configured TCI states can be activated via signaling (e.g., MAC-CE signaling). To receive a reference signal, channel, or both, the WTRU can select at least one TCI state from the set of activated TCI states, for example, based on DCI, indication, or conformance to a (predefined) configuration, such as in the case of DM-RS for receiving PDSCH using the default QCL assumption when the scheduling offset is less than (<) timeDurationForQCL. Once a new signaling / indication is received, for example, via MAC-CE, the WTRU can activate a new set of TCI states.

[0286] For example, a change in the set of active TCI states can be seen as an indication that the radio wave propagation environment has changed due to various factors, such as the rotation, movement, or both of the WTRU, or changes in other objects in the surrounding environment. Therefore, when the set of active TCI states of the WTRU changes, the WTRU can determine the need to evaluate the retraining of the AI / ML model used for beam selection and / or prediction.

[0287] When the AI / ML model used for beam prediction and / or selection is trained, the WTRU can be configured and activated using a first set of TCI states via a first (e.g., MAC-CE) instruction. After AI / ML model training is performed, the WTRU can be activated using a second (e.g., MAC-CE) instruction using a second set of TCI states. Upon receiving a second (e.g., MAC-CE) signaling instructing the activation of the second set of TCI states, the WTRU can determine and / or evaluate the need for retraining the AI / ML model and indicate the need for retraining the AI / ML model to the base station or gNB. The need for retraining the AI / ML model can be reported to the base station or gNB, for example, in a PUCCH, PUSCH, RACH, RRC message, or MAC CE.

[0288] In the example solution, the WTRU can compare a first set of TCI states with a second set of TCI states and determine the overlap level (L_overlap) between the two sets. If the L_overlap of the two sets is lower than a configured / pre-configured / determined threshold level, the WTRU can determine that the AI / ML model needs to be retrained. If the L_overlap is higher than the configured / pre-configured / determined threshold level, the WTRU can determine that the AI / ML model does not need to be retrained. The WTRU can determine and / or receive the threshold for TCI state overlap, for example, from the base station or gNB via RRC / MAC-CE signaling.

[0289] In another example solution, the WTRU can determine the number (N_add) of new TCI states activated in the second set of TCI states that are not part of the first set and / or the number (N_del) of TCI states in the first set of TCI states that are not included in the second set. If N_add and / or N_del are higher than the corresponding configured / pre-configured / determined thresholds, the WTRU can determine that AI / ML model retraining is required. If N_add and / or N_del are lower than the corresponding thresholds, the WTRU can determine that AI / ML model retraining is not required. The WTRU can receive the corresponding thresholds for N_add and N_del from the base station or gNB, for example, via RRC / MAC-CE signaling.

[0290] In additional or alternative example solutions, the WTRU can report one or more calculated parameters L_overlap, N_add, and N_del to the base station or gNB. In one example, the WTRU can report one or more calculated parameters as soft information. For example, the WTRU can report information about accuracy / effectiveness / confidence levels.

[0291] This document provides an implementation scheme and examples for reciprocity-based AI / ML model beam prediction verification. The WTRU can receive and measure one or more parameters (e.g., CSI or beam parameters such as RSRP, CQI, PMI, SINR, etc.) regarding one or more beam resources in a first frequency range (e.g., FR1). The WTRU can determine / predict (e.g., based on an AI / ML model) one or more beam resources in a second frequency range (e.g., FR2) based on the corresponding measurements. Beam resources may consist of TCI states for downlink, CSI-RS or SSB, SRS resources, or TCI states for uplink. The WTRU can define / determine one or more spatial filters for the determined / predicted beam resources. The WTRU can identify the determined / predicted beam resources via a reference ID.

[0292] Those skilled in the art will recognize that the embodiments and examples provided herein solve one or more problems. For example, one problem solved is how to verify the determined / predicted beam resources and the corresponding AI / ML models in a second frequency range.

[0293] In the example solution, the WTRU can perform one or more uplink transmissions (e.g., SRS, PUCCH, PUSCH), where the WTRU can determine associations by considering the spatial relationship between each uplink transmission and one of the determined / predicted (downlink) beam resources. Thus, the WTRU can determine which spatial domain filter to use for the uplink transmission that the WTRU may have already determined for the associated determined / predicted beam resource. The WTRU can indicate a reference ID corresponding to the determined / predicted beam resource, which is associated with the corresponding uplink transmission in the context of the spatial relationship.

[0294] In one example, the WTRU can be scheduled / configured for one or more UL transmissions with signals and / or channels. Thus, the WTRU can determine to use the same spatial filter to transmit the configured UL signals or channels, which can be defined for the determined / predicted beam resources. For example, the WTRU can determine to use the same spatial domain filter, which is determined to transmit (uplink) resource reference signals or channels on the determined beam resources, the predicted beam resources, or either. In other words, the WTRU can determine to consider the same QCL relationship between the determined / predicted (downlink) beam resources and the (uplink) transmitted signals or channels.

[0295] In the example solution, the base station or gNB can measure parameters corresponding to the received uplink signal and channel, such as beam resources, RSRP, CIR, angle of arrival (AoA), and PDCCH assumption BLER. The base station or gNB can change, update, or confirm the determined beam resources, the predicted beam resources, or both.

[0296] The WTRU can receive, for example via DCI, MAC CE, or other means, one or more signaling messages from the base station or gNB indicating whether the base station or gNB has changed, updated, or acknowledged the determined / predicted beam resources. In one example, the WTRU can receive, for instance in the DCI, a flag indicating whether the predicted / determined beam resources are valid or invalid. For example, a flag value of zero can indicate invalidity, and a flag value of one can indicate validity. Furthermore, the WTRU can receive one or more CSI-RS measurement and reporting configurations, such as CSI-RS resources, QCL information, TCI status, etc., which can be based on selected beam resources, such as at the base station or gNB.

[0297] The WTRU can receive one or more signals and channels in one or more beam resources (e.g., in a second frequency range) based on signals / channels transmitted / reported via the uplink. The WTRU can use the received signals to measure CSI and / or beam parameters. The WTRU can also use these measurements to update / retrain AI / ML models. The WTRU can select and report the optimal beam and the corresponding CSI quantities. For example, the corresponding CSI quantities may include CSI-RSRP, CIR, etc.

[0298] Although features and elements have been described above in specific combinations, those skilled in the art will understand that each feature or element may be used alone or in any combination with other features and elements. Furthermore, those skilled in the art will recognize that the aforementioned features and elements include means for performing the methods described herein. Additionally, the methods described herein may be implemented in a computer program, software, or firmware incorporated in a computer-readable medium for execution by a computer or processor. Examples of computer-readable media include electronic signals (transmitted via a wired or wireless connection) and computer-readable storage media. Examples of computer-readable storage media include, but are not limited to, read-only memory (ROM), random access memory (RAM), registers, cache memory, semiconductor memory devices, magnetic media (such as internal hard disks and removable disks), magneto-optical media, and optical media (such as CD-ROM disks and digital versatile optical discs (DVDs)). A processor associated with the software may be used to implement a radio frequency transceiver used in a WTRU, UE, terminal, base station, RNC, or any host computer.

Claims

1. A method for use in a wireless transmit / receive unit (WTRU), the method comprising: Receive first configuration information, the first configuration information including configuration for one or more reference signal (RS) resources measured and configuration for one or more RS resources predicted; Receive second configuration information, the second configuration information including configuration for one or more valid RS resources; Perform measurements on one or more RS resources being measured; A prediction is made for one or more of the best RS resources among the predicted one or more RS resources based on measurements of one or more RS resources. Perform measurements on one or more RS resources for validity; One or more optimal RS resources among one or more RS resources for effectiveness are determined based on measurements of one or more RS resources for effectiveness. Given that at least one RS resource from the prediction of one or more of the best RS resources in the one or more RS resources for prediction corresponds to an RS resource in the one or more of the best RS resources in the one or more RS resources for validity, the prediction of one or more of the best RS resources in the one or more RS resources for prediction is determined to be a valid prediction. as well as The report identifies the one or more optimal RS resources among the predicted one or more RS resources as valid predictions.

2. The method of claim 1, wherein the determination of whether the prediction of the one or more optimal RS resources among the predicted one or more RS resources is a valid prediction is performed within a pre-configured time period.

3. The method of claim 1, wherein the one or more RS resources for measurement are one or more RS resources for channel measurement.

4. The method of claim 1, wherein the one or more RS resources for validity are one or more RS resources for prediction accuracy.

5. The method of claim 1, wherein the one or more RS resources to be measured are one or more beam RS resources to be measured, the one or more RS resources to be predicted are one or more beam RS resources to be predicted, and the one or more RS resources to be valid are one or more beam RS resources to be valid.

6. The method of claim 1, further comprising: The report is derived from the prediction of up to a configured number of RS resources for one or more of the best RS resources among the predicted one or more RS resources.

7. The method of claim 1, wherein the measured one or more RS resources are in a first frequency range, and the predicted one or more RS resources are in a second frequency range.

8. The method of claim 1, further comprising: Receive information that provides a correspondence between one or more RS resources for prediction and one or more RS resources for validity.

9. The method of claim 1, wherein the one or more optimal RS resources among the predicted one or more RS resources have the highest reference signal received power (RSRP) among the predicted one or more RS resources.

10. The method of claim 1, wherein the one or more best RS resources among the one or more valid RS resources have the highest RSRP among the one or more valid RS resources.

11. A wireless transceiver unit (WTRU), comprising: transceiver; as well as The processor is operatively coupled to the transceiver; wherein: The transceiver and the processor are configured to receive first configuration information, the first configuration information including configuration for one or more measured reference signal (RS) resources and configuration for one or more predicted RS resources; The transceiver and the processor are configured to receive second configuration information, the second configuration information including configuration for the validity of one or more RS resources; The transceiver and the processor are configured to perform measurements on one or more RS resources. The processor is configured to determine a prediction of one or more optimal RS resources among the predicted one or more RS resources based on measurements of one or more RS resources. The transceiver and the processor are configured to perform measurements on the validity of one or more RS resources; The processor is configured to determine one or more optimal RS resources among the one or more RS resources for validity based on measurements of validity of one or more RS resources; The processor is configured to determine that the prediction for the one or more best RS resources among the predicted one or more RS resources is a valid prediction, provided that at least one RS resource from the prediction for the one or more best RS resources among the predicted one or more RS resources corresponds to an RS resource among the one or more best RS resources among the valid one or more RS resources; and The transceiver and the processor are configured to report a determination that the prediction for the one or more best RS resources among the predicted one or more RS resources is a valid prediction.

12. The WTRU of claim 11, wherein the determination of the prediction of the one or more optimal RS resources among the predicted one or more RS resources is valid is performed within a pre-configured time period.

13. The WTRU of claim 11, wherein one or more RS resources for measurement are one or more RS resources for channel measurement.

14. The WTRU of claim 11, wherein one or more RS resources for validity are one or more RS resources for prediction accuracy.

15. The WTRU of claim 11, wherein the one or more RS resources for measurement are one or more beam RS resources for measurement, the one or more RS resources for prediction are one or more beam RS resources for prediction, and the one or more RS resources for validity are one or more beam RS resources for validity.

16. The WTRU of claim 11, wherein the transceiver and the processor are further configured to: report up to a configured number of RS resources from the predicted one or more optimal RS resources among the predicted one or more RS resources.

17. The WTRU of claim 11, wherein the measured one or more RS resources are in a first frequency range, and the predicted one or more RS resources are in a second frequency range.

18. The WTRU of claim 11, wherein the transceiver and the processor are further configured to: receive information providing a correspondence between one or more RS resources for prediction and one or more RS resources for validity.

19. The WTRU of claim 11, wherein the one or more optimal RS resources among the predicted one or more RS resources have the highest reference signal received power (RSRP) among the predicted one or more RS resources.

20. The WTRU of claim 11, wherein the one or more best RS resources among the one or more valid RS resources have the highest RSRP among the one or more valid RS resources.

Images