WO2025172489A1

WO2025172489A1 - Enhancements of ai/ml reporting, ai/ml management and ai/ml inference

Info

Publication number: WO2025172489A1
Application number: PCT/EP2025/053945
Authority: WO
Inventors: Thomas Fehrenbach; Baris GÖKTEPE; Tatiana Rykova; Thomas Wirth; Thomas Schierl; Cornelius Hellge; Thomas Wiegand; Alvaro LÓPEZ RAVENTÓS; Buket TORUN
Original assignee: Fraunhofer Gesellschaft zur Foerderung der Angewandten Forschung eV
Current assignee: Fraunhofer Gesellschaft zur Foerderung der Angewandten Forschung eV
Priority date: 2024-02-15
Filing date: 2025-02-13
Publication date: 2025-08-21
Anticipated expiration: 2026-08-15

Abstract

A user device, UE, for a wireless communication network uses at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality for performing one or more tasks associated with an operation of the UE. The at least one AI/ML model or functionality operates on the basis of input data obtained from one or more measurements. The one or more measurements for obtaining the input data are associated with a configured or preconfigured configuration. The UE adapts the input data of the AI/ML model or functionality responsive to a certain event.

Description

ENHANCEMENTS OF AI/ML REPORTING, AI/ML MANAGEMENT AND AI/ML INFERENCE

Description

The present invention relates to the field of wireless communication systems or networks, more specifically a use of at least one Artificial Intelligence/Machine Learning model, AI/ML, model or at least one AI/ML functionality for performing one or more tasks. Embodiments of the present invention concern enhancements of AI/ML reporting, AI/ML monitoring, AI/ML pre-processing, AI/ML management and AI/ML inference.

Fig. 1 is a schematic representation of an example of a terrestrial wireless network 100 including, as is shown in Fig. 1 (A), the core network, CN, 102 and one or more radio access networks RANi, RAN2, ... RANN. Fig. 1(B) is a schematic representation of an example of a radio access network RAN_n that may include one or more base stations gNBi to gNBs, each serving a specific area surrounding the base station schematically represented by respective cells IO61 to IO65. The base stations are provided to serve users within a cell. The one or more base stations may serve users in licensed and/or unlicensed bands. The term base station, BS, refers to a gNB in 5G networks, an eNB in UMTS/LTE/LTE-A/ LTE- A Pro, or just a BS in other mobile communication standards, e.g., a base station in a 6G network. The BS may also comprise of integrated access and backhaul, IAB, nodes, e.g., an IAB Donor and/or IAB Node, consisting of a central unit, CU, as well as of a distributed unit, DU, and/or containing lAB-MTs including IAB mobile termination, MT. The term base station may refer to an access point, AP, in any of the WiFi standards, e.g., belonging to the IEEE 802.11-familiy. A user may be a stationary device or a mobile device. The wireless communication system may also be accessed by mobile or stationary loT devices which connect to a base station or to a user. The mobile or stationary devices may include physical devices, ground based vehicles, such as robots or cars, aerial vehicles, such as manned or unmanned aerial vehicles, UAVs, the latter also referred to as drones, buildings and other items or devices having embedded therein electronics, software, sensors, actuators, or the like as well as network connectivity that enables these devices to collect and exchange data across an existing network infrastructure. Fig. 1(B) shows an exemplary view of five cells, however, the RAN_n may include more or less such cells, and RAN_n may also include only one base station. Fig. 1(B) shows two users UE1 and UE2, also referred to as user device or user equipment, that are in cell IO62 and that are served by base station gNB2. Another user UE3 is shown in cell IO64 which is served by base station gNB4. The arrows IO81, IO82 and IO83 schematically represent uplink/downlink connections for transmitting data from a user UE1, UE2 and UE3 to the base stations gNB2, gNB4 or for transmitting data from the base stations gNB2, gNB4 to the users UE1, UE2, UE3. This may be realized on licensed bands or on unlicensed bands. Further, Fig. 1(B) shows two further devices 110i and HO2 in cell IO64, like loT devices, which may be stationary or mobile devices. The device 110i accesses the wireless communication system via the base station gNB₄ to receive and transmit data as schematically represented by arrow 112i. The device 110₂ accesses the wireless communication system via the user UE₃ as is schematically represented by arrow 1122. The respective base station gNBi to gNBs may be connected to the core network 102, e.g., via the S1 interface, via respective backhaul links 114i to 114s, which are schematically represented in Fig. 1(B) by the arrows pointing to “core”. The core network 102 may be connected to one or more external networks. The external network may be the Internet, or a private network, such as an Intranet or any other type of campus networks, e.g., a private WiFi communication system or a 4G or 5G mobile communication system. Further, some or all of the respective base station gNBi to gNBs may be connected, e.g., via the S1 or X2 interface or the XN interface in NR, with each other via respective backhaul links 116i to H65, which are schematically represented in Fig. 1(B) by the arrows pointing to “gNBs”. A sidelink channel allows direct communication between UEs, also referred to as device-to- device, D2D, communication. The sidelink interface in 3GPP is named PC5. Note, that the term user equipment, UE, or user device may also refer to a station, STA, as used in any of the WiFi standards, e.g., belonging to the IEEE 802.11-familiy.

For data transmission a physical resource grid may be used. The physical resource grid may comprise a set of resource elements to which various physical channels and physical signals are mapped. For example, the physical channels may include the physical downlink, uplink and sidelink shared channels, PDSCH, PLISCH, PSSCH, carrying user specific data, also referred to as downlink, uplink and sidelink payload data, the physical broadcast channel, PBCH, and the physical sidelink broadcast channel, PSBCH, carrying for example a master information block, MIB, and one or more system information blocks, SIBs, one or more sidelink information blocks, SLIBs, if supported, the physical downlink, uplink and sidelink control channels, PDCCH, PLICCH, PSSCH, carrying for example the downlink control information, DCI, the uplink control information, UCI, and the sidelink control information, SCI, and physical sidelink feedback channels, PSFCH, carrying PC5 feedback responses. The sidelink interface may support a 2-stage SCI which refers to a first control region containing some parts of the SCI, also referred to as the 1^st-stage SCI, and optionally, a second control region which contains a second part of control information, also referred to as the 2^nd-stage SCI.

For the uplink, the physical channels may further include the physical random-access channel, PRACH or RACH, used by UEs for accessing the network once a UE synchronized and obtained the MIB and SIB. The physical signals may comprise reference signals or symbols, RS, synchronization signals and the like. The resource grid may comprise a frame or radio frame having a certain duration in the time domain and having a given bandwidth in the frequency domain. The frame may have a certain number of subframes of a predefined length, e.g., 1ms. Each subframe may include one or more slots of 12 or 14 OFDM symbols depending on the cyclic prefix, CP, length. A frame may also have a smaller number of OFDM symbols, e.g., when utilizing shortened transmission time intervals, sTTI, or a mini-slot/non-slot-based frame structure comprising just a few OFDM symbols.

The wireless communication system may be any single-tone or multicarrier system using frequency-division multiplexing, like the orthogonal frequency-division multiplexing, OFDM, system, the orthogonal frequency-division multiple access, OFDMA, system, or any other Inverse Fast Fourier Transform, IFFT, based signal with or without Cyclic Prefix, CP, e.g., Discrete Fourier Transform-spread-OFDM, DFT-s-OFDM. Other waveforms, like non- orthogonal waveforms for multiple access, e.g., filter-bank multicarrier, FBMC, generalized frequency division multiplexing, GFDM, or universal filtered multi carrier, LIFMC, may be used. The wireless communication system may operate, e.g., in accordance with 3GPPs LTE, LTE-Advanced, LTE-Advanced Pro, or the 5G or 5G-Advanced or 6G or 3GPPs NR, New Radio, or within LTE-ll, LTE Unlicensed or NR-U, New Radio Unlicensed, which is specified within the LTE and within NR specifications.

The wireless network or communication system depicted in Fig. 1 may be a heterogeneous network having distinct overlaid networks, e.g., a network of macro cells with each macro cell including a macro base station, like base station gNBi to gNBs, and a network of small cell base stations, not shown in Fig. 1 , like femto or pico base stations. In addition to the above-described terrestrial wireless network also non-terrestrial wireless communication networks, NTN, exist including spaceborne transceivers, like satellites, and/or airborne transceivers, like unmanned aircraft systems. The non-terrestrial wireless communication network or system may operate in a similar way as the terrestrial system described above with reference to Fig. 1 , for example in accordance with the LTE-Advanced Pro or 5G or 5G-Advanced or NR, New Radio, or a possible future 6G radio system. In mobile communication networks, for example in a network like that described above with reference to Fig. 1 , like an LTE or 5G/NR network, there may be UEs that communicate directly with each other over one or more sidelink, SL, channels, e.g., using the PC5/PC3 interface or WiFi direct. UEs that communicate directly with each other over the sidelink may include vehicles communicating directly with other vehicles, V2V communication, vehicles communicating with other entities of the wireless communication network, V2X communication, for example roadside units, RSUs, roadside entities, like traffic lights, traffic signs, or pedestrians. An RSU may have a functionality of a BS or of a UE, depending on the specific network configuration. Other UEs may not be vehicular related UEs and may comprise any of the above-mentioned devices. Such devices may also communicate directly with each other, D2D communication, using the SL channels.

When considering two UEs directly communicating with each other over the sidelink, both UEs may be served by the same base station so that the base station may provide sidelink resource allocation configuration or assistance for the UEs. For example, both UEs may be within the coverage area of a base station, like one of the base stations depicted in Fig. 1. This is referred to as an “in-coverage” scenario. Another scenario is referred to as an “out- of-coverage” scenario. It is noted that “out-of-coverage” does not mean that the two UEs are necessarily outside one of the cells depicted in Fig. 1 , rather, it means that these UEs may not be connected to a base station, for example, they are not in an RRC connected state, so that the UEs do not receive from the base station any sidelink resource allocation configuration or assistance, and/or may be connected to the base station, but, for one or more reasons, the base station may not provide sidelink resource allocation configuration or assistance for the UEs, and/or may be connected to the base station that may not support NR V2X services, e.g., GSM, UMTS, LTE base stations or a WiFi AP.

In a wireless network or communication system Artificial Intelligence (Al) and Machine Learning (ML) may be employed for certain tasks. For example, according to 3GPP, AI/ML techniques and data analytics may be incorporated into the 5G system design for supporting certain tasks, e.g., for supporting network automation, data collection for various network functions, network energy savings, load balancing, mobility optimizations, synchronization, modulation and coding scheme (MCS) selection, AI/ML-based services, AI/ML for the new radio (NR) air interface. For example, when considering the NR air interface, AI/ML models may be employed for one or more of the following use cases:

Channel State Information (CSI):

For example, AI/ML may be used for a time-domain prediction of CSI feedback. In another example, AI/ML may be used for compressing CSI feedback.

Beam Management (BM):

For example, Al for beam management in 5G involves the use of Al and ML techniques to improve the efficiency and reliability of a wireless communication using directional beams. Beam management is the process of steering, tracking, and selecting the best beams for each user and link in a 5G network. This is challenging due to factors such as user mobility, a higher number of antennas, and the adoption of higher frequencies, e.g., the adoption of FR2 or FR3. Al and ML may offer valuable solutions to mitigate this complexity and minimize the overhead associated with beam management and selection, while maintaining system performance. For example, AI/ML may be used to interpolate in the spatial domain using measurements from only a few selected beams. This reduces the number of beams that actually have to be transmitted. In another example, AI/ML may also be used to predict the best beams in the temporal domain.

Unlicensed band operation:

For example Al for channel access in unlicensed bands for 5G involves the use of Al and ML techniques to improve the efficiency and reliability of wireless communication using the unlicensed spectrum. The unlicensed spectrum is the part of the radio frequency spectrum that is not allocated to any specific service or operator, and may be used by anyone who follows certain rules and regulations. The unlicensed spectrum may offer more bandwidth, lower cost, and greater flexibility for 5G applications, especially in scenarios where the licensed spectrum is scarce or expensive. There are some challenges and opportunities of Al for channel access in unlicensed bands for 5G, like: o Channel access methods: There are different methods for accessing unlicensed channels, such as listen before talk, LBT, gap-based channel access, contention-based random access, etc. Each method has its own advantages and disadvantages in terms of latency, throughput, fairness, and overhead. Al and ML may help to design, optimize, and adapt these methods according to the network conditions and user requirements. o Spectrum sharing and coexistence: The unlicensed spectrum is shared by multiple users and technologies, such as Wi-Fi, Bluetooth, LTE-U, LAA, MulteFire, CBRS, NR-U, etc. This may cause interference, congestion, and collisions among different transmissions. Al and ML may help to enhance the spectrum sharing and coexistence mechanisms, such as sensing, coordination, scheduling, power control, beamforming, etc., to improve the spectral efficiency and quality of service. o Private networks and industrial loT: The unlicensed spectrum may enable the deployment of 5G private networks and industrial loT applications, such as smart factories, warehouses, mines, etc. These applications have high demands for reliability, security, and low latency. Al and ML may help to customize and optimize the network performance for these applications, such as intelligent load balancing, proactive network slicing, anomaly detection, etc.

Positioning:

For example, a direct AI/ML positioning approach, e.g., fingerprinting, and an AI/ML assisted positioning approach, e.g., the output of the AI/ML model inference is an additional measurement and/or an enhancement of an existing measurement, may be implemented.

The AI/ML model or functionality may be running at one of the two sides or at both sides of the communication link, e.g., at the gNB or AP, or the network-side, e.g., CN, and/or at the UE or STA. Some AI/ML models may not be specified and left up to implementation, while others, e.g., enabling AI/ML for the air interface, need to be specified. The AI/ML models may be implemented using a model-l D-based Life-Cycle-Management, LCM, or a functionality-based LCM. In the first LCM, AI/ML models may be identified at the network side, and the network may control which AI/ML model is currently used at the UE. The term AI/ML model may further refer not only to a physical model but also to a logical model comprising certain properties. Hence, an AI/ML model may be implemented using different physical AI/ML models. In the functionality-based LCM, the network may have less control over the AI/ML model used at the UE. The actual AI/ML model may be completely transparent to the network, where the network only identifies a functionality that may be enabled by a set of configurations. Nevertheless, a model ID may also be used in a functionality-based LCM.

It is noted that the information in the above section is only for enhancing the understanding of the background of the invention and, therefore, it may contain information that does not form prior art that is already known to a person of ordinary skill in the art. Starting from the above, there may be a need for improvements or enhancements of AI/ML reporting, AI/ML management and AI/ML inference.

Embodiments of the present invention are now described in further detail with reference to the accompanying drawings:

Fig. 1 (A)-(B) illustrate a wireless communication network, wherein Fig. 1 (A) is a schematic representation of an example of a terrestrial wireless network, and Fig. 1(B) is a schematic representation of an example of a radio access network, RAN;

Fig. 2 illustrates the mapping of CRI, SSBRI, RSRP as given in the tables in TS 38.212 V17.7.0;

Fig. 3 is a schematic representation of a wireless communication system including a transmitter, like a base station, and one or more receivers, like user devices, UEs, implementing embodiments of the present invention;

Fig. 4 illustrates a user device, UE, in accordance with embodiments of the first, second and third aspects of the present invention;

Fig. 5 illustrates an embodiment for a beam reporting based on a CSI prediction in accordance with the first aspect of the present invention;

Fig. 6 illustrates a further embodiment for a beam reporting based on a CSI prediction in accordance with the first aspect of the present invention;

Fig. 7 illustrates a differential RSRP reporting in accordance with embodiments of the first aspect of the present invention;

Fig. 8 illustrates AI/ML reporting enhancements may be provided according to embodiments of the second aspect of the present invention;

Fig. 9 illustrates a prediction timing in accordance with embodiments; Fig. 10 illustrates a skipping of a beam management reporting for power saving in accordance with embodiments;

Fig. 11 illustrates an RRC measurement configuration in accordance with embodiments of the present invention;

Fig. 12 illustrates an embodiment of partially adapting samples in the input data of an AI/ML model;

Fig. 13 illustrates an embodiment for the handling of input samples by an AI/ML model;

Fig. 14 illustrates examples of beam subsets and their spatial relation to one another;

Fig. 15 illustrates an antenna grid overlaid by larger or wide-coverage beams; and

Fig. 16 illustrates an example of a computer system on which units or modules as well as the steps of the methods described in accordance with the inventive approach may execute.

Embodiments of the present invention are now described in more detail with reference to the accompanying drawings, in which the same or similar elements have the same reference signs assigned.

In conventional wireless communication networks or systems, like the one described above with reference to Fig. 1 , which may be a current 5G NR systems, a large overhead is caused by characterizing the radio channel and exchanging an estimate of the radio channel between user equipment, UE, and the base station, gNB. The overhead may be caused by a gNB transmitting training symbols in the NR downlink, by triggering measurement reports at the UE, as well as procedures for transmitting training signals from the UE to a gNB in the NR uplink. Furthermore, the overall system performance may depend on various factors, e.g., other UEs within the vicinity, the position and speed of a UE, the interference caused by neighboring cells, etc. 5G-Advanced and 6G systems may overcome some of the drawbacks mentioned here by utilizing AI/ML algorithms and by adapting procedures and protocols accordingly, and thus improve overall system performance of 5G NR telecommunication systems.

Beam Management

Beam management is a set of techniques to establish and maintain optimal directional links between the base station (gNB) and the user equipment (UE) in 5G networks, especially high frequency bands, e.g., in the millimeter wave (mmWave) frequency bands. Beam management involves the following procedures:

Beam sweeping: This is the process of covering a spatial area with a set of beams transmitted and received according to pre-specified intervals and directions.

Beam measurement: This is the evaluation of the quality of the received signal at the gNB or at the UE, using metrics such as RSRP, RSRQ, SINR, or SNR.

Beam determination: This is the selection of the suitable beam or beams either at the gNB or at the UE, based on the beam measurements.

Beam reporting: This is the procedure used by the UE to send beam quality and/or beam decision information to the gNB.

Beam management is performed in both idle mode, when the UE does not have active data transmission, and in connected mode, when the UE is exchanging data with the gNB. In idle mode, the UE uses the synchronization signal block (SSB) to perform initial access and cell search. The SSB includes primary and secondary synchronization signals (PSS, SSS) and the physical broadcast channel (PBCH), which carry essential information for the UE to synchronize and connect to the gNB. The SSB is transmitted using a fixed beam pattern that covers the entire cell. The UE measures the SSB and reports the best beam index to the gNB. The gNB then uses the reported beam index to steer the beam towards the UE for subsequent transmissions.

In connected mode, the UE and the gNB use different reference signals for beam management. The gNB uses the channel state information reference signal (CSI-RS) to transmit beams to the UE, which are used by the UE to measure and report. The UE uses the sounding reference signal (SRS) to transmit beams to the gNB, which can be used at the gNB to measure and determine the uplink radio channel characteristics. The gNB and the UE also exchange beam failure and recovery information using the radio link control (RLC) and medium access control (MAC) protocols. In connected mode, the gNB configures the UE with multiple CSI-RS resources, which describe a CSI-RS (reference symbol) in terms of the REs it is transmitted on and its periodicity, its bandwidth, its time offset, etc. One or more CSI-RS resources are bundled in CSI resource sets. One or more CSI resource sets belong to a CSI resource configuration that is usually associated with a CSI report configuration (CSI-ReportConfig). The CSI- ReportConfig defines how often and when a UE is supposed to report the measurements, e.g. periodically, aperiodic, or triggered etc. Then the UE reports per CSI resource set. For beam management purposes, the UE is configured to report the L1-RSRP.

L1-RSRP reporting in NR is defined in TS38.214 with the following details given in section 5.2.1.4.3 L1-RSRP Reporting.

For L1-RSRP computation the UE may be configured with CSI-RS resources, SS/PBCH Block resources or both CSI-RS and SS/PBCH block resources, when resource-wise quasi co-located with 'type C and 'typeD' when applicable. the UE may be configured with CSI-RS resource setting up to 16 CSI-RS resource sets having up to 64 resources within each set. The total number of different CSI- RS resources over all resource sets is no more than 128.

For L1-RSRP reporting, if the higher layer parameter nrofReportedRS in CSI-ReportConfig is configured to be one, the reported L1-RSRP value is defined by a 7-bit value in the range [-140, -44] dBm with 1dB step size, if the higher layer parameter nrofReportedRS is configured to be larger than one, or if the higher layer parameter groupBasedBeamReporting is configured as 'enabled', or if the higher layer parameter groupBasedBeamReporting-r17 is configured, the UE shall use differential L1-RSRP based reporting, where the largest measured value of L1-RSRP is quantized to a 7-bit value in the range [-140, -44] dBm with 1dB step size, and the differential L1-RSRP is quantized to a 4- bit value. The differential L1-RSRP value is computed with 2 dB step size with a reference to the largest measured L1-RSRP value which is part of the same L1-RSRP reporting instance. The mapping between the reported L1-RSRP value and the measured quantity is described in [11 , TS38.133],

Fig. 2 illustrates the mapping of CRI, SSBRI, RSRP as given in the tables in TS38.212

V17.7.0. In a reporting occasion, the UE determines up to 4 (dependent on the configuration) strongest beams and reports their CSI-RS resource indicator (CRI) and the associated L1- RSRP. The CRI is the index of a CSI-RS resource within a CSI resource set by which the beam is uniquely identified. Each CSI-RS resource is transmitted using a specific refined beam. Hence, the CRI identifies a CSI-RS resource and by that a specific beam. So the terms CRI and beam or beam ID or CSI-RS or CSI-RS resource may be used interchangeably. Furthermore, when the gNB actually transmits data, i.e. PDSCH, to the UE it uses the Transmit Configuration Indicator (TCI) that may be configured or indicated explicitly in the DCI. The TCI state links a data transmission, PDSCH or PUSCH, to up to two reference signals, e.g. a CSI-RS, SSB, SRS etc. Furthermore, it states shared properties of the beams in the form of the quasi-co-location (QCL) parameter. For example, if a SSB and a PDSCH are linked with QCL Type D, it means that they only share Rx properties. In particular, this means that the gNB may use a fine beam for the PDSCH but a coarse beam for the SSB. Both beams although being different share the same direction, hence they are QCLed Type D. In practice, this means that the UE may use the same Rx beam to receive the PDSCH but cannot assume that other parameters are the same. Furthermore, the UE may link a CSI-RS resource to the PDSCH with QCL Type A, which essentially means that the PDSCH and the CSI-RS have been transmitted using the same beam. Hence, the UE can use more reception parameters, such as the Doppler shift, Doppler spread, average delay, delay spread, that it obtained from measuring the said CSI- RS to equalize and decode the PDSCH reception. Thus, the TCI or TCI state essentially also identifies a certain beam or beam ID and hence, can be used interchangeably. Furthermore, as mentioned previously, SSBs or DMRS may also be transmitted using a certain beam and the UE may use an SSB ID or SSB index or DMRS index to identify the certain beam. Hence, the SSB ID, DMRS index and SSB index may identify a certain beam or beam ID and may be used interchangeably.

Al Beam Management

Al for beam management in 5G is a topic that involves the use of artificial intelligence (Al) and machine learning (ML) techniques to improve the efficiency and reliability of wireless communication using directional beams. As mentioned above, beam management is the process of steering, tracking, and selecting the best beams for each user and link in a 5G network. This is challenging due to factors such as user mobility, a higher number of antennas, and the adoption of higher frequencies, e.g., the adoption of FR2 or FR3. Al and ML offer valuable solutions to mitigate this complexity and minimize the overhead associated with beam management and selection, while maintaining system performance. While the use of one or more AI/ML model or one or more AI/ML functionalities provides advantages, not only in the use case of beam management but also when performing one or more of the above mentioned tasks, there is still a need for enhancements of AI/ML reporting, AI/ML management, AI/ML pre-processing and AI/ML inference.

Embodiments of the present invention provide enhancements of AI/ML reporting, AI/ML management, AI/ML pre-processing and AI/ML inference. Embodiments of the present invention may be implemented in a wireless communication system as depicted in Fig. 1 including base stations and users, like mobile terminals or loT devices. Fig. 3 is a schematic representation of a wireless communication system 310 including a transmitter 300, like a base station, and one or more receivers 302, 304, like user devices, UEs. The transmitter 300 and the receivers 302, 304 may communicate via one or more wireless communication links or channels 306a, 306b, 308, like a radio link. The transmitter 300 may include one or more antennas ANTT or an antenna array having a plurality of antenna elements, a signal processor 300a and a transceiver 300b, coupled with each other. The receivers 302, 304 include one or more antennas ANTUE or an antenna array having a plurality of antennas, a signal processor 302a, 304a, and a transceiver 302b, 304b coupled with each other. The base station 300 and the UEs 302, 304 may communicate via respective first wireless communication links 306a and 306b, like a radio link using the Uu interface, while the UEs 302, 304 may communicate with each other via a second wireless communication link 308, like a radio link using the PC5 or sidelink, SL, interface. When the UEs are not served by the base station or are not connected to the base station, for example, they are not in an RRC connected state, or, more generally, when no SL resource allocation configuration or assistance is provided by a base station, the UEs may communicate with each other over the sidelink. The system or network of Fig. 3, the one or more UEs 302, 304 of Fig. 3, and the base station 300 of Fig. 3 may operate in accordance with the inventive teachings described herein.

First Aspect

A first aspect of the present invention concerns the reporting of certain values and/or beams which have been predicted and/or measured by a user device, UE, to the network side, for example to a base station or gNB of the radio access network, RAN, and/or predicted and/or measured by a base station or gNB of the radio access network, RAN, to UE, for enhancing AI/ML reporting. The present invention provides a user device, UE, for a wireless communication network, wherein the UE is to receive from a network entity of the wireless communication network one or more reference signals, wherein the UE is to obtain for each of one or more performance parameters one or more performance parameter values and/or one or more beams, wherein a respective one of the values or a respective one of the beams is obtained using: a measurement of one or more of the reference signals, and/or a prediction using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the UE is transmit to the network entity a report including the one or more performance parameter values and/or the one or more beams, wherein the report includes at least one predicted performance parameter value and/or at least one predicted beam.

In accordance with embodiments, if the UE predicts at least one performance parameter value and/or at least one beam using the at least one AI/ML model or functionality, the UE is to indicate that the report includes at least one predicted performance parameter value and/or at least one predicted beam.

In accordance with embodiments, the report includes a beam identifier for: each of one or more beams, and/or each of one or more performance parameter values.

In accordance with embodiments, the beam identifier includes one or more of the following: a beam identification, beam ID, an identification of the reference signal or of a reference signal configuration of the beam, e.g. indexed by CRI, an identification of a TCI state of the beam. an identification of a beam configuration of the beam, an identification of a SSB of the beam, an identification of a SRS of the beam, an identification of a DMRS of the beam, an identification of a phase or PT-RS of the beam, an identification of a cell, e.g., a physical cell ID (PCI) in NR or SSID or BSSID used in WiFi networks, an identification of a TRP, e.g., in case of multi-TRP, which can be implemented by using a CORSET pool index.

In accordance with embodiments, the UE is preconfigured or configured with a report configuration, the report configuration configuring the report to include zero or more measured performance parameter values and/or zero or more measured beams, and to include at least one predicted performance parameter value and/or at least one predicted beam, and the UE is to modify the report as defined by the report configuration, and/or modify a transmission of the report, e.g., by transmitting the report with a priority, or a frequency, or a period, or a offset, or a delay, which is different from a priority, or a frequency, or a period, or a offset, or a delay, with which the report is transmitted according to the report configuration, e.g., for prioritizing certain content of the report.

In accordance with embodiments, the UE is to receive the report configuration from one or more of the following: a Radio Access Network, RAN, entity, like a base station or another UE, a Core Network, CN, entity, an over the top, OTT, server, a WiFi access point, AP, or a WiFi station, STA.

In accordance with embodiments, the UE is to modify the report by one or more of the following: omitting an entry from the report for which the performance parameter value/beam has been predicted, setting at least a part of an entry in the report, for which the performance parameter value/beam has been predicted, to a predefined value, e.g., to a default value, like zero, using at least a part of an entry in the report, for which the performance parameter value/beam has been predicted, for signaling additional information, using an unused part of an entry for signaling additional information, e.g., a confidence level or indicate whether AI/ML has been used to determine the value, replacing at least parts of one or more or all entries in the report, for which the value has been predicted, by additional information, replacing one or more or all entries in the report by additional information, compress a content of the report.

In accordance with embodiments, compressing the content of the report comprises one or more or any combination of the following:

Huffman codes,

Run-Length encoding (RLE),

Time series compression, e.g., Gorilla or Prometheus approach,

Entropy codes, e.g. Golomb codes or Rice codes,

Importance sampling, e.g. discard values based on their importance,

Lempel-Ziv-Welsh compression,

Lempel-Ziv compression,

- Quantization,

AI/ML compression.

In accordance with embodiments, the UE is to generate a prediction report which includes only the beam identifiers of the one or more beams, or the beam identifiers of the one or more beams and additional information for one or more of the beams.

In accordance with embodiments, the additional information comprises one or more of the following: prediction information, an indication that one or more of the beams identified in the report are unexpected, one or more earlier measurements or predictions of the performance values, e.g., a last measured or predicted CRI, or a last measured or predicted Layer 1 reference signal received power, L1-RSRP, a relative gain or loss to other beams, a relative gain or loss to one or more of: a last measurement, a last prediction, a last report, or a currently used beam, a rank of the beam in the report, a timestamp of a measurement or a prediction, duration of a measurement or a prediction measured or predicted channel state information, CSI, an AI/ML model or functionality ID.

In accordance with embodiments, the CSI comprises one or more of the following: a measured or predicted Channel Quality Indicator, CQI, a measured or predicted Precoding Matrix Indicator PM I, a measured or predicted CSI Reference Signal, CSI-RS, resource indicator, CRI, a measured or predicted Synchronization Signal/Physical Broadcast Channel, SS/PBCH, Block Resource Indicator, SSBRI, a measured or predicted layer indicator, LI, a measured or predicted rank indicator, Rl, a measured or predicted Layer 1 reference signal received power, L1-RSRP, a measured or predicted Layer 1 Signal to Interference plus Noise Ratio, L1-SINR, a measured or predicted Capability Index, a measured or predicted Doppler and/Doppler delay profile, one or more measured or predicted time-domain channel properties, TDCP

In accordance with embodiments, the prediction information comprises one or more of the following: a prediction certainty, a confidence interval, a model ID of the AI/ML model, location information, like a location of the UE, a scenario in which the UE is used, a cell ID, a UE ID, for allowing the network entity to monitor a performance of the prediction, a timestamp of a measurement or a prediction, a confidence level, validity time, a confidence value indicating a confidence with the entries, a relative gain or loss to other beams, a validity of a prediction, e.g., a time for which a prediction is predicted to be valid, an average prediction error, an indication of an unexpected prediction outcome, e.g., spatially not close to a last used beam, a prediction diversity to evaluate whether a prediction algorithm tends to favor certain beams or directions over others across different scenarios, e.g. over varying channel conditions or user mobility. The present invention provides a network entity, for a wireless communication network, wherein the network entity is to determine for each of one or more performance parameters one or more performance values using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the network entity is to report for each of the one or more performance parameters the one or more of measured or predicted performance values for a performance parameter, wherein a performance value, which is smaller than a greatest performance value, is quantized with reference to a next greater performance value, or a performance value, which is greater than a smallest performance value, is quantized with reference to a next smaller performance value, or a greatest performance value and at least one further performance value, which is smaller than the greatest performance value, are quantized as absolute values, a performance value, which is smaller than the greatest performance value and greater than the at least one further performance value, is quantized with reference to the greatest performance value or with reference to a next greater performance value, and a performance value, which is smaller than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next greater performance value, or a smallest performance value and at least one further performance value, which is greater than the smallest performance value, are quantized as absolute values, a performance value, which is greater than the smallest performance value and smaller than the at least one further performance value, is quantized with reference to the smallest performance value or with reference to a next smaller performance value, and a performance value, which is greater than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next smaller performance value.

In accordance with embodiments, the network entity is to quantize the greatest performance value as an absolute value using a first number of bits, and for all remaining performance values, which are smaller than the greatest performance value, determine a difference of the performance value with reference to the next greater performance value, and quantize the difference using a second number of bits, wherein the second number is smaller than the first number,

In accordance with embodiments, the second number of bits decreases with each difference being quantized, or the second number of bits is larger or smaller than the first number of bits, or the first number of bits is zero, i.e. the greatest performance value is omitted.

In accordance with embodiments, the network entity is to quantize the greatest performance value and at least one further performance value, which is smaller than the greatest performance value, as absolutes values using a first number of bits and a second number of bits, respectively, for all performance values, which are smaller than the greatest performance value and greater than the at least one further performance value, determine a difference of the performance value with reference to the greatest performance value or with reference to the next greater performance value, and quantize the difference using a third number of bits, wherein the third number is smaller than the first number, for all performance values, which are smaller than the at least one further performance value, determine a difference of the performance value with reference to the at least one further performance value or with reference to the next greater performance value, and quantize the difference using a fourth number of bits, wherein the fourth number is smaller than the second number.

In accordance with embodiments, the third number of bits and/or the fourth number of bits decreases with each difference being quantized, or the third number of bits and/or fourth number of bits is larger or smaller than the first number of bits and/or the second number of bits, or the first number of bits and/or second number of bits is zero, i.e. the greatest performance value/at least one further performance value is omitted.

In accordance with embodiments, the first and second numbers of bits are the same or different, and/or the third and fourth numbers of bits are the same or different. In accordance with embodiments, the network entity is pre-configured or configured, e.g. by a base station, by a another network entity, or a UE, with the quantization steps.

In accordance with embodiments, the one or more performance parameters comprise one or more of the following: one or more beams, which are transmitted by another network entity of the wireless communication system and received at the network entity, the performance value indicating a measured or predicted strength of a beam at the network entity, a reference signal received power, RSRP, the performance value indicating the measured or predicted RSRP, a reference signal received quality, RSRQ, the performance value indicating the measured or predicted RSRQ, a signal to noise ratio, SNR, the performance value indicating the measured or predicted SNR, a rank,

- a PMI, a signal to noise and interference ratio, SINR, the performance value indicating the measured or predicted SINR, a radio signal strength indicator RSSI, the performance value indicating the measured or predicted RSSI, an interference level, the performance value indicating the measured or predicted interference level, a doppler parameter, the performance value indicating the measured or predicted doppler parameter, a delay, the performance value indicating the measured or predicted delay,

- a packet loss rate, the performance value indicating the measured or predicted packet loss rate, one or more parameters reported from higher layers, the performance value indicating the measured or predicted values for the one or more parameters, a measured or predicted Doppler and/or Doppler delay profile.

In accordance with embodiments, the one or more performance parameters are for one or more beams, which are transmitted by another network entity of the wireless communication system and received at the network entity, the performance value indicating a measured or predicted strength of a beam at the network entity, the network entity is to determine a plurality of beams from a plurality of beams, which are transmitted by another network entity of the wireless communication system and received at the network entity, and send a report about the plurality of beams to the other network entity, and the network entity is to determine the plurality of beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality.

In accordance with embodiments, the determined plurality of beams are the strongest beams out of the plurality of beams.

In accordance with embodiments, the network entity is to report for each of the plurality of beams one or more measured or predicted values, wherein a measured or predicted value for each beam, which is weaker than a strongest beam, is quantized with reference to a measured or predicted value of a next stronger beam, or a measured or predicted value for the strongest beam and a measured or predicted value for at least one further beam, which is weaker than the strongest beam, are quantized as absolute values, a measured or predicted value for each beam, which is weaker than a strongest beam and stronger than the at least one further beam, is quantized with reference to the measured or predicted value of the strongest beam or with reference to the measured or predicted value of the next stronger beam, and a measured or predicted value for each beam, which is weaker than the at least one further beam, is quantized with reference to the measured or predicted value of the at least one further beam or with reference to the measured or predicted value of the next stronger beam.

In accordance with embodiments, the network entity is to quantize the measured or predicted value for the strongest beam as an absolute value using a first number of bits, and for all remaining beams which are weaker than the strongest beam, determine a difference of the measured or predicted value for a beam with reference to a measured or predicted value for the next stronger beam, and quantize the difference using a second number of bits, wherein the second number is smaller than the first number. In accordance with embodiments, the second number of bits decreases with each difference being quantized.

In accordance with embodiments, the network entity is to quantize the measured or predicted value for the strongest beam and the measured or predicted value for at least one further beam, which is weaker than the strongest beam, as absolutes values using a first number of bits and a second number of bits, respectively, for all beams, which are weaker than the strongest beam and stronger than the at least one further beam, determine a difference of the measured or predicted value for a beam with reference to a measured or predicted value for the strongest beam or with reference to a measured or predicted value for the next stronger beam, and quantize the difference using a third number of bits, wherein the third number is smaller than the first number, for all beams, which are weaker than the at least one further beam, determine a difference of the measured or predicted value for a beam with reference to a measured or predicted value for the next stronger beam or with reference to a measured or predicted value for the at least one further beam, and quantize the difference using a fourth number of bits, wherein the fourth number is smaller than the second number.

In accordance with embodiments, the third number of bits and/or the fourth number of bits decreases with each difference being quantized.

In accordance with embodiments, the first and second numbers of bits are the same or different, and/or the third and fourth numbers of bits are the same or different.

In accordance with embodiments, the network entity is a user device, UE, or a base station.

In accordance with embodiments, the UE comprise one or more of a power-limited UE, or a hand-held UE, like a UE used by a pedestrian, and referred to as a Vulnerable Road User, VRU, or a Pedestrian UE, P-UE, or an on-body or hand-held UE used by public safety personnel and first responders, and referred to as Public safety UE, PS-UE, or an loT UE or Ambient loT UE, e.g., a sensor, an actuator or a UE provided in a campus network to carry out repetitive tasks and requiring input from a gateway node at periodic intervals, or a mobile terminal, or a stationary terminal, or a cellular loT-UE, an industrial loT-UE, I loT, or a SL UE, or a vehicular UE, or a vehicular group leader UE, GL-LIE, or a scheduling UE, S- UE, or an loT or narrowband loT, NB-loT, device, a NTN UE, or a WiFi device or WiFi station, STA, or a ground based vehicle, or an aerial vehicle, or a drone, or a moving base station, or road side unit, RSU, or a building, or any other item or device provided with network connectivity enabling the item/device to communicate using the wireless communication network, e.g., a sensor or actuator, or any other item or device provided with network connectivity enabling the item/device to communicate using a sidelink the wireless communication network, e.g., a sensor or actuator, or any sidelink capable network entity.

In accordance with embodiments, the BS comprises one or more of a macro cell base station, or a small cell base station, or a central unit of a base station, or a distributed unit of a base station, or an Integrated Access and Backhaul, I AB, node, or a road side unit, RSU, or a WiFi access point, AP, or a UE, or a SL UE, or a group leader UE, GL-UE, or a relay or a remote radio head, or an AMF, or an SMF, or a core network entity, or mobile edge computing, MEC, entity, or a network slice as in the NR or 5G core context, or any transmission/reception point, TRP, enabling an item or a device to communicate using the wireless communication network, the item or device being provided with network connectivity to communicate using the wireless communication network.

The present invention provides a wireless communication network, like a 3^rd Generation Partnership Project, 3GPP, system, comprising a one or more user devices, UEs, according to embodiments of the first aspect of the present invention and/or one or more network entities according to embodiments of the first aspect of the present invention.

In accordance with embodiments, the wireless communication network comprises one or more base stations, BSs, wherein the base station may comprises one or more of a macro cell base station, or a small cell base station, or a central unit of a base station, or a distributed unit of a base station, or an Integrated Access and Backhaul, IAB, node, or a road side unit, RSU, or a WiFi access point, AP, or a UE, or a SL UE, or a group leader UE, GL-UE, or a relay or a remote radio head, a satellite payload, e.g., a NTN gNB, or an AMF, or an SMF, or a core network entity, or mobile edge computing, MEC, entity, or a network slice as in the NR or 5G core context, or any transmission/reception point, TRP, enabling an item or a device to communicate using the wireless communication network, the item or device being provided with network connectivity to communicate using the wireless communication network.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: receiving, by the UE, from a network entity of the wireless communication network one or more reference signals, obtaining, by the UE, for each of one or more performance parameters one or more performance parameter values and/or one or more beams, wherein a respective one of the values or a respective one of the beams is obtained using: a measurement of one or more of the reference signals, and/or a prediction using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and transmitting, by the UE, to the network entity a report including the one or more performance parameter values and/or the one or more beams, wherein the report includes at least one predicted performance parameter value and/or at least one predicted beam.

The present invention provides a method for operating a network entity, for a wireless communication network, the method comprising: determining, by the network entity, for each of one or more performance parameters one or more performance values using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and reporting, by the network entity, for each of the one or more performance parameters the one or more of measured or predicted performance values for a performance parameter, wherein a performance value, which is smaller than a greatest performance value, is quantized with reference to a next greater performance value, or a performance value, which is greater than a smallest performance value, is quantized with reference to a next smaller performance value, or a greatest performance value and at least one further performance value, which is smaller than the greatest performance value, are quantized as absolute values, a performance value, which is smaller than the greatest performance value and greater than the at least one further performance value, is quantized with reference to the greatest performance value or with reference to a next greater performance value, and a performance value, which is smaller than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next greater performance value, or a smallest performance value and at least one further performance value, which is greater than the smallest performance value, are quantized as absolute values, a performance value, which is greater than the smallest performance value and smaller than the at least one further performance value, is quantized with reference to the smallest performance value or with reference to a next smaller performance value, and a performance value, which is greater than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next smaller performance value.

Second Aspect

A second aspect of the present invention concerns the reporting of beam measurements and/or predictions performed by a user device, UE, to a base station or gNB of the RAN to be used for beam management procedures, for enhancing AI/ML reporting of one or more beams.

The present invention provides a user device, UE, for a wireless communication network, wherein the UE is to determine one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and send a report about the one or more beams to the network entity, wherein the UE is to determine each of the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the UE is to receive a configuration defining a reporting of the one or more beams to the network entity.

In accordance with embodiments, the UE is to use an AI/ML model or functionality to generate the report, the AI/ML model or functionality for generating the report being the same as or being different from the AI/ML model or functionality used for determining the one or more beams. In accordance with embodiments, the UE is to determine from the configuration whether to report the one or more beams determined using the measurement, or the one or more beams determined using the at least one AI/ML model or functionality, e.g., by prediction, or the one or more beams, wherein at least one is determined using the measurement and at least one is determined using the at least one AI/ML model or functionality, e.g., by prediction.

In accordance with embodiments, determining the one or more beams comprises obtaining a set of measured beams, the set including one or more beams, and/or a set of predicted beams, the set including one or more beams, and wherein the sets of predicted and measured beams may include the same beams, different beams, disjoined beams or overlapping beams.

In accordance with embodiments, the configuration implicitly or explicitly indicates a reporting of measured or predicted beams.

In accordance with embodiments, the configuration implicitly or explicitly indicates a reporting of measured and predicted beams.

In accordance with embodiments, if the configuration does not include the one or more reference signal resources, e.g., lacks any non-zero-power Channel State Information Reference Signal, NZP-CSI-RS, resources, the UE is to predict the one or more beams using the at least one AI/ML model or functionality.

In accordance with embodiments, if the configuration includes the one or more reference signal resources, e.g., non-zero-power Channel State Information Reference Signal, NZP- CSI-RS, resources, the UE is to determine the one or more beams using measurements of the one or more reference signal resources.

In accordance with embodiments, the UE is configured with a measurement reporting configuration and a prediction reporting configuration. In accordance with embodiments, the UE is configured with a measurement reporting configuration or a prediction reporting configuration.

In accordance with embodiments, the set of beams, e.g. CSI resources, configured for measuring with the prediction reporting configuration are a subset of the set of beams configured with the measurement reporting config.

In accordance with embodiments, the configuration includes an indicator, like an Al indicator field, if the indicator has a first value, the UE is to report predictions or predictions and measurements.

In accordance with embodiments, the configuration includes an indicator, like an Al indicator field, if the indicator has a second value, the UE is to report only measurements.

In accordance with embodiments, the UE is to report: one or more predictions, e.g., a predicted CSI-RS resource indicator, CRI, a beam ID of a predicted beam, a predicted Layer 1 reference signal received power, L1- RSRP, and/or one or more measurements of the one or more reference signal resources, e.g., a measured CRI, a beam ID of a predicted beam, a measured L1-RSRP, the one or more measurements being obtained responsive to an earlier configuration indicating a reporting of measured beams.

In accordance with embodiments, the report comprises one or more of the following: a CSI-RS resource indicator, CRI, a beam ID, a transmission reception point identifier, TRP ID, e.g., as may be used to identify multi-TRPs, an identification of a TRP, e.g., in case of multi-TRP, which can be implemented by using a CORSET pool index, one or more beams, e.g., identified by a beam ID or by a resource ID, e.g., CRI, Channel State Information, CSI, e.g., a Channel Quality Indicator, CQI, a precoding matrix indicator, PMI, a CSI reference signal, CSI-RS, resource indicator, CRI, a Synchronization Signal/Physical Broadcast Channel, SS/PBCH, Block Resource indicator, SSBRI, a layer indicator, LI, a rank indicator, Rl, a Layer 1 reference signal received power, a Layer 1 Signal to Interference plus Noise Ratio, L1-RSRP, L1- SINR, a Capability Index or time-domain channel properties, TDCP, a received signal strength indicator, RSSI, a higher layer CSI, e.g., Layer 3 reference signal received power, L3-RSRP, or a Layer 1 Signal to Interference plus Noise Ratio, L3-SINR.

In accordance with embodiments, the UE is to report, in addition to the one or more measurements, also the one or more predictions, e.g., a predicted CRI, a beam ID of a predicted beam, a predicted L1-RSRP.

In accordance with embodiments, the UE is to determine at least one of the one or more beams using a measurement of one or more reference signal resources at one or more first occasions, and predict at least one of the one or more beams using the at least one AI/ML model or functionality at one or more second occasions.

In accordance with embodiments, the UE is to report the one or more beams at a configured or preconfigured reporting occasion, and wherein, dependent on a temporal relationship between the reporting occasion and the first and/or second occasions, the report includes the one or more predictions, and/or the one or more measurements.

In accordance with embodiments, a reporting occasion is triggered by a condition.

In accordance with embodiments, the condition is one or more of a finished computing of an AI/ML calculation, a failed computing of an AI/ML calculation, e.g., the AI/ML module could not finish the computation within a certain time or the calculation failed completely, an indication by the AI/ML model or functionality, e.g., the AI/ML module output of AI/ML calculation, one or more results of AI/ML calculation, is o above or below a configured and/or preconfigure threshold or o a change wrt. the last calculated or reported result or o a change wrt. the last calculated or reported result is above or below a configured and/or preconfigure threshold. The present invention provides user device, UE, for a wireless communication network, wherein the UE is to determine, at one or more configured or preconfigured occasions, one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and send a report about the one or more beams to the network entity, wherein the one or more occasions comprise one or more first occasions at which the UE is to determine the at least one of the one or more beams using a measurement of one or more reference signal resources, and one or more second occasions at which the UE is to determine at least one of the one or more beams using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, e.g., by prediction,

In accordance with embodiments, the UE is to report the one or more beams at a configured or preconfigured reporting occasion, and dependent on a temporal relationship between the reporting occasion and the first and second occasions, the report includes the at least one beam determined using the measurement, and/or the at least one beam determined using the at least one AI/ML model, e.g., by prediction.

In accordance with embodiments, the one or more first occasions comprise one or more measuring windows having a predefined duration, and/or the one or more second occasions comprise one or more prediction windows having a predefined duration.

In accordance with embodiments, a number of first and/or second occasions depends on one or more criteria.

In accordance with embodiments, the one or more criteria comprise one or more of the following: a distance to a base station, e.g. dependent on a position of the UE in a cell of the wireless communication network the UE is configured with a first number of second occasions when being at a first distance from the base station of the cell and with a second number of second occasions when being at a second distance from the base station of the cell, the first number of second occasions and the first distance being greater than the second number of second occasions and the second distance, a distance from the UE to one or more other UEs or to one or more Radio Access Network, RAN, entities,

- whether the UE is indoors or outdoors, dependent on a coherence, a frequency and/or a time of multiple paths if the UE is served by a multi-TRP entity,

- whether an AI/ML model is used for prediction, a prediction accuracy of the AI/ML model used, a mobility of the UE, e.g. speed with which the UE is moving, a required QoS of a service running on the UE, a required HARQ, like a number or ratio of ACKs or NACKs, a carrier frequency, e.g., FR1 or FR2 or FR3, a battery or power state, e.g. battery level or charging state. a scenario, e.g. UMa, UMi, RMa, RMi, indoor, a channel condition, e.g., LOS, NLOS, dependent on the delay of the strongest path, an interference level, e.g. below or above a certain threshold, a type of UE, e.g. loT device, eMBB device, a vehicular UE, V2X, or a pedestrian UE, P-UE.

In accordance with embodiments, the UE is to report the at least one beam determined using the measurement, if a time between a first occasion and the reporting occasion is less than a configured or preconfigured threshold, and/or the UE is to report the at least one beam determined using the at least one AI/ML model or functionality, e.g., by prediction, if the time between a first occasion and the reporting occasion is more than a configured or pre-configured threshold.

In accordance with embodiments, the UE is to predict the one or more beams at a certain second occasion and/or report the predicted one or more beams at the reporting occasion only if a time between a first occasion and the certain second occasion and/or the reporting occasion is more than a configured or pre-configured threshold. In accordance with embodiments, the UE is to skip a prediction of the one or more beams at a certain second occasion and/or a reporting of the predicted one or more beams at the reporting occasion if a time between a first occasion and the certain second occasion and/or the reporting occasion is less than a configured or pre-configured threshold.

In accordance with embodiments, responsive to skipping the prediction and/or the reporting, the UE is to enter into a sleep mode, like a Discontinuous Reception, DRX, mode, e.g. until the next first occasion or the next second occasion or the next reporting occasion.

In accordance with embodiments, the second occasions are periodic or aperiodic.

In accordance with embodiments, the UE is to receive a signaling of the one or more second occasions from the wireless communication network, or the UE is to initiate the one or more second occasions responsive to one or more conditions.

In accordance with embodiments, the one or more conditions comprise one or more of the following: a change of a link performance, e.g., a degradation of a link performance beyond a predefined limit, a measurement being above or below a configured or pre-configured threshold, a higher layer event, like a change of a Quality of Service, QoS, a change of a data rate, a change of a delay demand, a geographical change, like a change of an Al zone or an Al area, an upcoming transmission by the UE or an upcoming reception at the UE.

In accordance with embodiments, a change of the link performance is detected by a measurement or by another indicator, e.g., a number of Hybrid Acknowledge Request, HARQ, non-acknowledgements exceeding a configured or pre-configured threshold.

In accordance with embodiments, the UE is to initiate the one or more second occasions at a configured or pre-configured time before the upcoming transmission and piggyback the prediction onto the transmission, e.g., as a MAC-CE or UCI, wherein the prediction may only be piggybacked if the prediction triggers a pre-configured condition. In accordance with embodiments, the condition comprises one or more of the following: an upcoming UL transmission, e.g. a scheduled or configured PLISCH in the reporting slot, an SPS or configured grant configuration, e.g. a periodic grant that can be used if the prediction needs to be reported. a QoS of the upcoming transmission, a size or type of the upcoming transmission, remaining bits available for piggybacking, a specific location of the UE, a time since the last measurement or prediction, a time since the last report is above a configured or preconfigured threshold.

In accordance with embodiments, a length or duration of the second occasion, like a length of a prediction window, depends on one or more performance parameters, like a link quality, experienced by the UE, and the length or duration of the second occasion increases or decreases if the performance parameter is above or below a configured or pre-configured threshold.

In accordance with embodiments, the UE is to skip a measurement of the one or more beams at a certain first occasion and/or a reporting of the measured one or more beams at the reporting occasion if one or more first conditions apply, and/or the UE is to perform a measurement of the one or more beams at a certain first occasion and/or a reporting of the measured one or more beams at the reporting occasion if one or more second conditions apply.

In accordance with embodiments, the first condition for skipping a measurement comprises one or more of the following: a prediction at a second occasion preceding the certain first occasion exceeds a configured or pre-configured threshold, e.g., the confidence of the prediction exceeds a certain threshold making a new measurement unnecessary or the current beam is still the best in the prediction or the predicted RSRPs have a delta exceeding a threshold. a currently served beam, e.g., a beam quality, has not changed, a battery status of the UE is below a configured or pre-configured threshold, a change in a QoS requirement or in one or more higher layer criteria, like an upcoming high priority transmission, an indication from the network or higher layers.

In accordance with embodiments, the second condition comprises one or more of the following: a confidence associated with the prediction is below a certain threshold, a time gap between the reporting occasion and the measurement is less than a certain threshold, a delta between one or more predicted values is to low, one or more deltas between the last measurement and the prediction exceed a threshold, a time since the last measurement report exceeds a certain threshold, an indication from the network or higher layers, a performance monitoring threshold associated to the prediction.

In accordance with embodiments, responsive to skipping the measurement and/or the reporting, the UE is to enter into a sleep mode, like a Discontinuous Reception, DRX, mode.

In accordance with embodiments, the UE is in sleep mode until one or more of until the next first occasion, the next second occasion, the next reporting occasion, the UE’s next uplink grant, a maximum timer is reached, e.g., based on a configured or preconfigure threshold, the UE receives a wake-up signal, WUS, e.g., a WUS transmitted by a gNB or by another UE or by another RAN or WiFi entity.

In accordance with embodiments, the UE is to report the one or more beams in accordance with a report configuration for a measurement report which reports the one or more beams determined using the measurement, and the UE is to replace one or more or all entries in the measurement report by prediction entries. In accordance with embodiments, the UE is to indicate in the measurement report that the measurement report includes prediction entries, like a predicted CRI index or a predicted Layer 1 reference signal received power, L1-RSRP.

In accordance with embodiments, if the measurement report includes prediction entries, the UE is to perform one or more of the following: set a flag in the measurement report, activate a prediction indicator, add a confidence value indicating a confidence with the associated entries, wherein a first value indicates that an entry is a measured value, and a second value indicates that an entry is a predicted value, set one more bits of a multi-bit value, like a L1-RSR 7-bit value, so as to signal that prediction is used, and set the remaining bits for indicating one or more of the following: o a confidence in the AI/ML model, o an age of calculation performed by the AI/ML model, o an ID of the AI/ML model, o an ID of the AI/ML functionality, o delta values, e.g., values indicating a delta with regard to a previous value.

In accordance with embodiments, the UE comprise one or more of a power-limited UE, or a hand-held UE, like a UE used by a pedestrian, and referred to as a Vulnerable Road User, VRU, or a Pedestrian UE, P-UE, or an on-body or hand-held UE used by public safety personnel and first responders, and referred to as Public safety UE, PS-UE, or an loT UE or Ambient loT UE, e.g., a sensor, an actuator or a UE provided in a campus network to carry out repetitive tasks and requiring input from a gateway node at periodic intervals, or a mobile terminal, or a stationary terminal, or a cellular loT-UE, an industrial loT-UE, I loT, or a SL UE, or a vehicular UE, or a vehicular group leader UE, GL-UE, or a scheduling UE, S- UE, or an loT or narrowband loT, NB-loT, device, a NTN UE, or a WiFi device or WiFi station, STA, or a ground based vehicle, or an aerial vehicle, or a drone, or a moving base station, or road side unit, RSU, or a building, or any other item or device provided with network connectivity enabling the item/device to communicate using the wireless communication network, e.g., a sensor or actuator, or any other item or device provided with network connectivity enabling the item/device to communicate using a sidelink the wireless communication network, e.g., a sensor or actuator, or any sidelink capable network entity. The present invention provides a base station, BS, for a wireless communication network, wherein the BS is to serves a user device, UE, of the wireless communication network, wherein the BS is to configure the UE to determine one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and send a report about the one or more beams to the network entity, wherein the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the BS is to configure the UE with a configuration defining a reporting of the one or more beams to the network entity, the configuration indicating that the one or more beams determined using the measurement, and/or the one or more beams predicted using the at least one AI/ML model or functionality are to be reported.

The present invention provides a base station, BS, for a wireless communication network, wherein the BS is to serves a user device, UE, of the wireless communication network, wherein the BS is to configure the UE to determine, at one or more configured or preconfigured occasions, one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and send a report about the one or more beams to the network entity, wherein the one or more occasions comprise one or more first occasions at which the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and one or more second occasions at which the UE is to predict the one or more beams using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the BS is to configure the UE to report the one or more beams at a configured or preconfigured reporting occasion, wherein, dependent on a temporal relationship between the reporting occasion and the first and second occasions, the report includes the one or more beams determined using the measurement, or the one or more beams predicted using the at least one AI/ML model or functionality. In accordance with embodiments, the BS comprises one or more of a macro cell base station, or a small cell base station, or a central unit of a base station, or a distributed unit of a base station, or an Integrated Access and Backhaul, I AB, node, or a road side unit, RSU, or a WiFi access point, AP, or a UE, or a SL UE, or a group leader UE, GL-LIE, or a relay or a remote radio head, or an AMF, or an SMF, or a core network entity, or mobile edge computing, MEC, entity, or a network slice as in the NR or 5G core context, or any transmission/reception point, TRP, enabling an item or a device to communicate using the wireless communication network, the item or device being provided with network connectivity to communicate using the wireless communication network.

The present invention provides a wireless communication network, like a 3^rd Generation Partnership Project, 3GPP, system, comprising a one or more user devices, UEs, according to embodiments of the second aspect pf the present invention and/or one or more base stations, BSs, according to embodiments of the second aspect pf the present invention.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: determining, by the UE, one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, sending, by the UE, a report about the one or more beams to the network entity, wherein the UE determines each of the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the UE receives a configuration defining a reporting of the one or more beams to the network entity.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: determining, by the UE, at one or more configured or preconfigured occasions, one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and sending, by the UE, a report about the one or more beams to the network entity, wherein the one or more occasions comprise one or more first occasions at which the UE is to determine the at least one of the one or more beams using a measurement of one or more reference signal resources, and one or more second occasions at which the UE is to determine at least one of the one or more beams using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, e.g., by prediction.

The present invention provides a method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, configuring, by the BS, the UE to determine one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and send a report about the one or more beams to the network entity, wherein the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and configuring, by the BS, the UE with a configuration defining a reporting of the one or more beams to the network entity, the configuration indicating that the one or more beams determined using the measurement, and/or the one or more beams predicted using the at least one AI/ML model or functionality are to be reported

The present invention provides a method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, configuring, by the BS, the UE to determine, at one or more configured or preconfigured occasions, one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity of the wireless communication system and received at the UE, and send a report about the one or more beams to the network entity, wherein the one or more occasions comprise one or more first occasions at which the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and one or more second occasions at which the UE is to predict the one or more beams using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and wherein the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and configuring, by the BS, the UE to report the one or more beams at a configured or preconfigured reporting occasion, wherein, dependent on a temporal relationship between the reporting occasion and the first and second occasions, the report includes the one or more beams determined using the measurement, or the one or more beams predicted using the at least one AI/ML model or functionality.

Third Aspect

A third aspect of the present invention concerns AI/ML management, AI/ML pre-processing and inference enhancements by keeping an AI/ML model or functionality used at a UE in an updated state over an operational period of the UE, by providing improved procedures for AI/ML training, and by providing improved procedures for using a currently used AI/ML model or functionality trained on the basis of a beam configuration of a certain base station also for a new base station.

The present invention provides a user device, UE, for a wireless communication network, wherein the UE is to use at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality for performing one or more tasks associated with an operation of the UE, wherein the at least one AI/ML model or functionality operates on the basis of input data obtained from one or more measurements, and wherein the one or more measurements for obtaining the input data are associated with a configured or preconfigured configuration, and wherein the UE is to adapt the input data of the AI/ML model or functionality responsive to a certain event.

In accordance with embodiments, the input data comprises a plurality of samples obtained from the measurements and adapting the input data of the AI/ML model or functionality comprises fully or partially adapting the samples in the input data.

In accordance with embodiments, fully adapting the samples in the input data comprises removing from the input data all samples, or resetting in the input data all samples to a configured or preconfigured value, like null.

In accordance with embodiments, partially adapting the samples in the input data comprises removing from the input data all samples obtained after the event, or resetting in the input data all samples obtained after the event to a configured or preconfigured value, like null, or removing from the input data a subset of samples obtained after the event and/or obtained before the event, e.g., remove or replace outdated samples.

In accordance with embodiments, partially adapting the samples in the input data comprises removing or not removing from the input data samples based on a function, or replacing a set of samples by the output of a function depending on the set of samples.

In accordance with embodiments, the function is one or more of a maximum value, e.g., based on a configured or preconfigure threshold, a minimum value, e.g., based on a configured or preconfigure threshold, arithmetic mean, geometric mean, weighted mean, moving average, a multi-dimensional function a one-dimensional function, a confidence interval, e.g., samples which are within a certain interval wrt. the average of the input data samples, or wrt. a configured or preconfigured average value.

In accordance with embodiments, the function is configured or preconfigured by a network entity, e.g., by a gNB or another UE.

In accordance with embodiments, responsive to the event, the UE is to perform one or more of the following: apply a default configuration, with which the UE is configured or preconfigured, apply a new configuration with which the UE is configured or preconfigured, receive from the wireless communication network, e.g., from a gNB or from the radio access network, RAN, or from another UE, a new configuration and apply the received new configuration, switch to a different AI/ML model or functionality corresponding to a new or default configuration. In accordance with embodiments, the certain event comprises one or more of the following: a change of a configuration associated with the one or more measurements for obtaining the input data, e.g., a beam configuration, an indication from the wireless communication network, e.g., a signaling from one or more entities in wireless communication network, a performance degradation, a change in a channel, a radio link failure, RLF, a mode switch in a vehicular UE, e.g., a UE switches from mode 1 , under control of a gNB, to mode 2, direct communication with other UEs via PC5, or vice versa, a handover, a fulfillment of a conditional handover, CHO, condition, a change in Quasi co-location, QCL, a change of a used MIMO mode, e.g., a rank drops, e.g., in case an antenna is suddenly shielded, which is more relevant in higher frequency ranges, e.g., FR2 or FR3. an AI/ML model update, e.g., a reception of new training data and/or a reception of a pre-trained AI/ML model, a state change, e.g., a transition of the UE from an inactive state to a connected state, a reception of a wake-up signal, WUS, a change in location of the UE, initiating carrier aggregation, CA, or evacuating a given carrier, e.g., if the UE switches off CA or removes a carrier from a multiband configuration, a radio link recovery, a change in signal quality, e.g., the SNR/SINR/RSSI/RSRP is improving or degrading, a successful HO or CHO, a successful beam switch or change of a TRP, a static position of a UE, e.g., a UE stops moving, a performance or confidence of the Al/M L model or functionality is above or below a certain threshold, an indication of the Al/M L model or functionality.

In accordance with embodiments, the UE is no longer to adapt the input data of the AI/ML model or functionality according to the certain event. In accordance with embodiments, the UE is no longer to adapt the input data of the AI/ML model or functionality according to another certain event.

In accordance with embodiments, the performance degradation or the change in a channel comprise a change of one or more measured parameters, like a Signal to Noise Ratio, SNR, a Signal-to- Interference-and-Noise Ratio, SINR, or a Reference Signal Received Power, RSRP, or a Reference Signal Received Quality, RSRQ, or a Reference Signal Strength Indicator, RSSI, or a Channel Quality Indicator, CQI), or an interference level, or more Chanel State Information, CSI, parameters dropping below a configured or preconfigured threshold or dropping during a configured or preconfigured time period by more than a configured or preconfigured amount.

In accordance with embodiments, the change in location of the UE comprises one or more of: o a movement of the UE from an environment for which the beam configuration applies to a new environment for which a new beam configuration applies, e.g. moving between an urban environment and a rural environment, o a movement of the UE into a certain region or zone, e.g., from outdoor to indoor, or into a certain distance from the base station, e.g., with respect to a minimum required communication range, o a change in longitude and/or latitude and/or height beyond a configured or preconfigured threshold.

In accordance with embodiments, the indication from the wireless communication network comprises one or more of: o an indication from a base station included, e.g., in Downlink Control Information, DCI, a Medium Access Control Control Element, MAC CE, or a Radio Resource Control, RRC, signaling, or through broadcast/multicast messages to multiple UEs, o a signaling of a new beam configuration, e.g., when the UE moves from an environment for which the beam configuration applies to a new environment for which the new beam configuration applies, e.g. moving between an urban environment and a rural environment, o a signaling of a new configuration, e.g., when the UE moves from an environment for which a localization configuration applies to a new environment for which a new localization configuration applies, e.g. moving between an urban environment and a rural environment, o a signaling of new assistance information replacing current assistance information, e.g., when the UE moves from an environment for which current assistance information applies to a new environment for which the new assistance information applies, e.g. moving between an urban environment and a rural environment, o a signaling from another UE, e.g., via sidelink assistance information messages, e.g., AIM.

In accordance with embodiments, the input data is stored in one or more of the following: in a memory that is part of the AI/ML model or functionality, a cloud server, another device, e.g., an associated UE, a base station, e.g., gNB or WiFi AP, a mobile edge cloud, MEC, located closely to a serving base station BS, a dedicated RAN entity, e.g., an AI/ML network function, NF.

In accordance with embodiments, the one or more of tasks comprise one or more of the following:

- AI/ML model based access to a RAN,

- AI/ML model based network energy saving,

- AI/ML model based load balancing, an AI/ML model based mobility optimization,

- AI/ML model based use cases, like o channel state information, CSI, feedback, like a CSI compression and/or a CSI prediction, or o beam management, or o positioning, like a direct AI/ML positioning (e.g., fingerprinting) and/or an AI/ML assisted positioning,

- AI/ML model based mobility management, e.g., a handover, HO, prediction and/or conditional handover, CHO, prediction,

- AI/ML model based modulation and coding scheme, MCS, selection,

- AI/ML model based synchronization,

- AI/ML model based encoding and/or decoding and/or precoding,

- AI/ML model based modulation and/or demodulation, - AI/ML model based positioning or ranging,

- AI/ML model based joint communication and sensing, JSAC,

- AI/ML model based feedback calculation, e.g., CSI/CQI/PMI/RI feedback,

- AI/ML model based interference management,

- AI/ML model based quality of experience, QoE, and/or quality of service, QoS, predictions,

- AI/ML model based network traffic forecasting.

In accordance with embodiments, the at least one AI/ML model or functionality performs beam management, and the UE is to determine one or more beams for a communication with one or more network entities of the wireless communication system using the at least one AI/ML model or functionality.

In accordance with embodiments, the UE is to obtain the one or more measurements from measuring the one or more beams.

In accordance with embodiments, a beam is identified by one or more of the following: a beam ID, a resource ID, a channel state information, CSI, or a signal derived from the CSI, a time index, e.g., the UE is configured or preconfigured with a certain timing and can derive from this, when certain beams can be decoded, a multi-TRP identifier, a physical cell ID, PCI, of a base station or BSSID of a WiFi access point.

In accordance with embodiments, the Chanel State Information, CSI, comprises one of more of the following: a Channel Quality Indicator, CQI, a precoding matrix indicator, PM I, a Chanel State Information Reference Signal, CSI-RS, resource indicator, CRI, a sounding reference signal, SRS,

Synchronization Signal/Physical Broadcast Channel, SS/PBCH, Block Resource Indicator, SSBRI, a SSB index, e.g., mapped to a certain beam, a layer indicator, LI, a spatial signature, e.g., the direction or direction of arrival or direction of departure of a signal, a rank indicator, Rl, a Layer 1 reference signal received power, L1-RSRP, a Layer 1 Signal to Interference plus Noise Ratio, L1-SINR, a Capability Index, one or more time-domain channel properties, TDCP, a received signal strength indicator, RSSI, higher layer CSI, e.g., a Layer 3 reference signal received power, L3-RSRP, or a Layer 3 Signal to Interference plus Noise Ratio, L3-SINR training fields, e.g., as used within the Preamble in WiFi systems.

The present invention provides a user device, UE, for a wireless communication network, wherein the UE is to perform one or more measurements of reference signal resources associated with respective beams, which are transmitted by a network entity of the wireless communication system, like a base station serving the UE, wherein one or more beams, which are to be used by the UE for a communication with the network entity, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the UE is to obtain the one or more beams to be used by the UE for the communication with the network entity or to be used by the network entity for the communication with the UE from at least one AI/ML model or functionality, which is operated at the network entity according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE, and/or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE.

In accordance with embodiments, the at least one AI/ML model or functionality determines the one or more beams in accordance with the antenna configuration for the respective beams and is operated by the network entity, and wherein the UE is to report the one or more measurements to the network entity.

In accordance with embodiments, the at least one AI/ML model or functionality determines the one or more beams in accordance with the antenna configuration for the respective beams and is operated by the network entity, and wherein the UE is to receive from the network entity feedback, which indicates one or more beams to be used by the UE for communication with the network entity.

In accordance with embodiments, the at least one AI/ML model or functionality determines the one or more beams in accordance with the antenna configuration for the respective beams, and wherein the UE is to transmit to the network entity feedback, which indicates one or more beams to be used by the network entity for communication with the UE.

In accordance with embodiments, the at least one AI/ML model or functionality operated by the network entity receives one or more measurements of reference signal resources associated with respective beams from one or more further UEs, e.g., for dynamically adjusting at least one AI/ML model or functionality using beamforming parameters based on one or more beam management reports.

In accordance with embodiments, the network entity is to receive the beam management reports from one or more of the following: one or more further UEs of the wireless communication system, the network, e.g., from a network function, NF, or from a neighboring gNB, e.g., a target gNB in case of a HO or potential CHO gNB candidate.

In accordance with embodiments, at least one AI/ML model or functionality analyzes one or more signal quality indicators and/or locations of the UE and the one or more further UEs to optimize a beam direction and/or a beam strength of the one or more beams to be used by the UE for the communication with the network entity.

In accordance with embodiments, the one or more signal quality indicators comprises one or more of the following: a pathloss or signal quality, e.g., using an uplink received signal strength indicator, RSSI, measured at the base station, a signal quality based on reference signals transmitted in the uplink by the UE, e.g., o based on sounding reference signals, SRS, o demodulation reference signals, DM-RS, o phase-tracking reference signals, PT-RS, a reciprocity-based signal quality, e.g., in case of a Time Division Duplex, TDD, system, based on CSI feedback provided in an uplink, feedback information sent by the UE, e.g., HARQ-ACKs or NACKs, or CBG-based ACKs or NACKs, provided by the UE, higher-layer statistics or feedback, e.g., packet delay or measured TCP slow-starts, or L3-RSRP or L3-latency.

In accordance with embodiments, the UE is to operate the at least one AI/ML model or functionality, receive from the network entity the antenna configuration for the respective beams, and determine the one or more beams to be used by the UE for the communication with the network entity by the at least one Al/M L model or functionality using the received antenna configuration for the respective beams and the one or more measurements.

In accordance with embodiments, the respective beams received by the UE comprise a proper subset of respective beams.

In accordance with embodiments, the UE is to receive different proper subsets of respective beams over time and use them as input data, e.g., to reconstruct a radio channel using different input values.

In accordance with embodiments, the UE is to receive the proper subset of respective beams together with specific configuration parameters.

In accordance with embodiments, the UE is to receive from the network entity at least one AI/ML model and/or parameters of the AI/ML model or functionality, e.g., a model trained at the network entity using the antenna configuration for the respective beams, and determine the one or more beams to be used by the UE for the communication with the network entity by the AI/ML model or functionality using the one or more measurements.

In accordance with embodiments, the UE is to use the AI/ML model or functionality to estimate or predict a beam for a communication with the network entity, e.g., by interpolating between beams by using the reference signal resources of which are sparsely measured to predict a different beam, e.g., a more optimal beam. The present invention provides a user device, UE, for a wireless communication network, wherein the UE supports at least one AI/ML model or functionality, that uses the measurements of one or more first beams from a first set of beams to determine one or more second beams from a second set of beams, wherein the UE is to determine whether the at least one AI/ML model or functionality is applicable based on a configuration received from a network entity, or wherein the UE is to indicate to a network entity information about the at least one AI/ML model or functionality.

In accordance with embodiments, the UE is to measure one or more first beams from a first set of beams received from a network entity, wherein the UE is to determine one or more second beams from a second set of beams received from the network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data.

In accordance with embodiments, the UE is to receive a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for another network entity of the wireless communication system.

In accordance with embodiments, the network entity is a serving base station, or a target base station, or a TRP, e.g., as in multi-TRP.

In accordance with embodiments, the signaling includes network entity specific information allowing the UE to identify the network entity or the another network entity or describing beam specifics of the network entity.

In accordance with embodiments, network entity specific information comprises one or more of the following: an identification of the network entity or the another network entity, e.g., a vendor ID, or a configuration ID, or a gNB ID, or a UE ID, or a target gNB ID in case of a handover, HO, or a potential conditional handover, CHO, a cell ID or a Physical Cell ID, PCI of the cell comprising the network entity or the another network entity, an antenna configuration of the network entity or the another network entity, an antenna array ID at the network entity or the another network entity, e.g., an ID of a transmission reception point, TRP, at the network entity comprising a multi-TRP, a beam configuration of the network entity or the another network entity.

In accordance with embodiments, the information about the at least one AI/ML model or functionality indicated to the network entity or the another network entity comprises one or more of the following:

• beam information about the first and second set of beams,

• an identification of a network entity with which the at least one AI/ML model or functionality is associated, e.g., a vendor ID, or a configuration ID, or a gNB ID, or a UE ID, or a target gNB ID in case of a handover, HO, or a potential conditional handover, CHO,

• a cell ID or a Physical Cell ID, PCI of the cell comprising a network entity with which the at least one AI/ML model or functionality is associated,

• an antenna configuration of a network entity with which the at least one AI/ML model or functionality is associated,

• an antenna array ID at a network entity with which the at least one AI/ML model or functionality is associated, e.g., an ID of a transmission reception point, TRP,

• a beam configuration of a network entity with which the at least one AI/ML model or functionality is associated.

In accordance with embodiments, the beam information comprises one or more of the following: one or more beam IDs of beams that belong to the first set of beams, one or more beam IDs of beams that belong to the second set of beams, a size of the first set of beams, e.g., a number of beams included in the first set of beams, a size of the second set of beams, e.g., a number of beams included in the second set of beams, a number of beams per dimension, like the number of beams per azimuth, or per elevation, angular differences between the beams of the first and/or second set, e.g., a phase offset between the beams of pi/4 resulting in 8 beams per dimension, amplitude differences between the beams of the first and/or second set, e.g., a , e.g., a power difference between the beams of the first and/or the second set, information about the network entity, like a cell ID, a gNB ID, a vendor ID, or a gNB Type, from which the first and second sets of beams have been obtained for training the AI/ML model or functionality.

In accordance with embodiments, the second set of beams includes more beams than the first set of beams.

In accordance with embodiments, the first set of beams is a proper subset of the second set of beams.

In accordance with embodiments, if the UE determines from the signaling that the AI/ML model or functionality is not working for the another network entity, the UE is to perform one or more of the following: modify or update the AI/ML model or functionality so as to meet the requirements for with the another network entity, replace the AI/ML model or functionality by a new AI/ML model or functionality meeting the requirements for with the another network entity, e.g., by selecting the new AI/ML model or functionality from a set of AI/ML models or functionalities stored in the UE, or by obtaining the new AI/ML model or functionality from the wireless communication network or from an external storage to which the UE is connectable via the wireless communication network, switching to a default AI/ML model or functionality, stop using the AI/ML model or functionality.

In accordance with embodiments, for a communication with the another network entity, the UE is configured by the another network entity with the first and/or second set of beams that is compatible with the first and/or second set of beams of the network entity that is associated with the at least one AI/ML model or functionality.

In accordance with embodiments, compatible means the first and/or second set of beams from both network entities are identical, or the first and/or second set of beams from the another network entity is a proper subset of the first and/or second of beams from the network entity. In accordance with embodiments, beams are defined as identical if one or more of the following applies: the beams have the same ID, the beams are spatially aligned, e.g., relative to each other, the beams have the same QCL, the beams have the same orientation in azimuth and/or elevation, relative to each other, the beams are originating from the same antenna configuration, e.g., same number of antenna elements, e.g., as in a Massive MIMO array, the beams are coming from the same PLMN, the beams are coming from a same gNB configuration, e.g., wrt. gNB sectorization or TRP configuration, e.g., multi-TRP, or antenna configuration, e.g., array antenna with the same number of antenna elements, the same are coming from the same gNB or the same gNB type or same vendor.

In accordance with embodiments, the UE is to receive from a network entity the first set of beams, and use the AI/ML model or functionality to estimate or interpolate one or more corresponding characteristics, like the RSRP, the RSSI, the RSRQ, the SINR, the SNR of one or more beams of the second set which are not received at the UE.

In accordance with embodiments, wherein the UE is to receive from a network entity only a plurality of proper subsets of the beams in the first or second set, wherein the plurality of proper subsets comprises a first subset and a second subset, use the first and second subsets for one or more initial measurements of one or more characteristics, like the RSRP, the RSSI, the RSRQ, the SINR, the SNR, and o report the one or more initial measurements to the network entity or o use an AI/ML model or functionality to predict one or more corresponding characteristics of at least a third subset of beams or the first or second set of beams.

In accordance with embodiments, the UE is to receive the plurality of proper subsets periodically or at configured or preconfigured intervals, wherein the one or more corresponding characteristics of the at least third subset or first or second set of beams are predicted for a following period or interval.

In accordance with embodiments, each of the first or second beams of the first or second set is a wide-coverage beam, e.g., SSB beams having a coverage area comprising coverage areas of a plurality narrow-coverage beams, and wherein the UE is to measure one or more of the wide-coverage beams.

In accordance with embodiments, the UE is to predict from the one or more measurements of the wide-coverage beams one or more of the narrow-coverage beams, e.g. CSI-RS beams.

In accordance with embodiments, the UE is to receive for each of the plurality of proper subsets a reference signal resource configuration, like a CSI-RS resource config or a CSI- RS resource set.

In accordance with embodiments, a first reference signal resource configuration of a first subset of beams is a proper subset of a second reference signal resource configuration of a second subset of beams.

The present invention provides a user device, UE, for a wireless communication network, wherein the UE is configured or preconfigured with a plurality of Artificial

Intelligence/Machine Learning modes, AI/ML models or functionality, for performing a certain task, wherein the UE is to use one or more of the AI/ML models or functionality for performing the certain task, wherein the UE is to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to-use state allowing the UE to activate one or more of the non-used AI/ML models for performing the certain task, and wherein the UE is to put the one or more of the of the AI/ML models, which are currently not used for performing the certain task, into a ready-to-use state responsive to a first signaling. In accordance with embodiments, the first signaling comprises one or more of an RRC signaling, a MAC CE, or a higher layer control signaling or signaling via sidelink, e.g., PC5 RRC or PSCCH or control signaling embedded into PSSCH.

In accordance with embodiments, the UE is to activate one or more of the non-used AI/ML models or functionalities for performing the certain task responsive to a second signaling.

In accordance with embodiments, the UE is to deactivate one or more of the used AI/ML models or functionalities responsive to the second signaling.

In accordance with embodiments, the second signaling comprises a DCI indicating a switch, an activation or a deactivation, e.g., using an index or a bitmap indicating used and nonused AI/ML models or functionalities to be deactivated/activated.

In accordance with embodiments, the UE comprise one or more of a power-limited UE, or a hand-held UE, like a UE used by a pedestrian, and referred to as a Vulnerable Road User, VRU, or a Pedestrian UE, P-UE, or an on-body or hand-held UE used by public safety personnel and first responders, and referred to as Public safety UE, PS-UE, or an loT UE or Ambient loT UE, e.g., a sensor, an actuator or a UE provided in a campus network to carry out repetitive tasks and requiring input from a gateway node at periodic intervals, or a mobile terminal, or a stationary terminal, or a cellular loT-UE, an industrial loT-UE, I loT, or a SL UE, or a vehicular UE, or a vehicular group leader UE, GL-UE, or a scheduling UE, S- UE, or an loT or narrowband loT, NB-loT, device, a NTN UE, or a WiFi device or WiFi station, STA, or a ground based vehicle, or an aerial vehicle, or a drone, or a moving base station, or road side unit, RSU, or a building, or any other item or device provided with network connectivity enabling the item/device to communicate using the wireless communication network, e.g., a sensor or actuator, or any other item or device provided with network connectivity enabling the item/device to communicate using a sidelink the wireless communication network, e.g., a sensor or actuator, or any sidelink capable network entity or just a network entity.

The present invention provides a base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, wherein the BS is to transmit respective beams to the UE, wherein the BS is to receive from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, wherein the BS is to determine from the respective beams one or more beams for a communication with the UE using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the AI/ML model or functionality operates on the basis of input data, the input data comprising the measurements received from the UE, and wherein the BS is to adapt the input data of the AI/ML model or functionality responsive to a change of the beam configuration.

The present invention provides a base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, wherein the BS is to transmit respective beams to the UE respective beams, wherein the BS is to receive from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, wherein one or more beams, which are to be used by the BS for a communication with the UE, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the one or more beams to be used for the communication are to be obtained from at least one AI/ML model or functionality, which is operated at the BS according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE, or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE.

The present invention provides a base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, the UE measuring one or more first beams from a first set of beams supported by the UE and determining one or more second beams from a second set of beams supported by a first network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data, and wherein the BS is to transmit to the UE a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for a BS.

In accordance with embodiments, the signaling includes BS specific information allowing the UE to identify the BS or describing beam specifics of the BS.

In accordance with embodiments, the BS is to receive from the UE the beam information about the first and second set of beams, and the signaling includes a notification to the UE whether the AI/ML model or functionality at the UE is also working for the BS, the notification generated by the BS using the beam information from the UE.

The present invention provides a base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, wherein the UE is configured or preconfigured with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionalities, for performing a certain task, wherein the BS is to signal to the UE to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to- use state allowing the UE to activate one or more of the non-used AI/ML models or functionalities for performing the certain task.

In accordance with embodiments, the BS comprises one or more of a macro cell base station, or a small cell base station, or a central unit of a base station, or a distributed unit of a base station, or an Integrated Access and Backhaul, I AB, node, or a road side unit, RSU, or a WiFi access point, AP, or a UE, or a SL UE, or a group leader UE, GL-UE, or a relay or a remote radio head, or an AMF, or an SMF, or a core network entity or a network entity, or mobile edge computing, MEC, entity, or a network slice as in the NR or 5G core context, or any transmission/reception point, TRP, enabling an item or a device to communicate using the wireless communication network, the item or device being provided with network connectivity to communicate using the wireless communication network.

The present invention provides a wireless communication network, like a 3^rd Generation Partnership Project, 3GPP, system, comprising a one or more user devices, UEs, according to embodiments of the third aspect of the present and/or one or more base stations, BSs, according to embodiments of the third aspect of the present invention.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: using, by the UE, at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality for performing one or more tasks associated with an operation of the UE, wherein the at least one AI/ML model or functionality operates on the basis of input data obtained from one or more measurements, and wherein the one or more measurements for obtaining the input data are associated with a configured or preconfigured configuration, and adapting, by the UE, the input data of the AI/ML model or functionality responsive to a certain event.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: performing, by the UE, one or more measurements of reference signal resources associated with respective beams, which are transmitted by a network entity of the wireless communication system, like a base station serving the UE, determining, by the UE, one or more beams, which are to be used by the UE for a communication with the network entity, using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, obtaining, by the UE, the one or more beams to be used by the UE for the communication with the network entity or to be used by the network entity for the communication with the UE from at least one AI/ML model or functionality, which is operated at the network entity according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE, and/or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: supporting, by the UE, at least one AI/ML model or functionality, that uses the measurements of one or more first beams from a first set of beams to determine one or more second beams from a second set of beams, determining, by the UE, whether the at least one AI/ML model or functionality is applicable based on a configuration received from a network entity, or indicating, by the UE, to a network entity information about the at least one AI/ML model or functionality.

The present invention provides a method for operating a user device, UE, for a wireless communication network, the method comprising: configuring por preconfiguring the UE with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionality, for performing a certain task, using, by the UE, one or more of the AI/ML models or functionality for performing the certain task, maintaining, by the UE, one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to-use state allowing the UE to activate one or more of the non-used AI/ML models for performing the certain task, and putting, by the UE, the one or more of the of the AI/ML models, which are currently not used for performing the certain task, into a ready-to-use state responsive to a first signaling.

The present invention provides a method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, transmitting, by the BS, respective beams to the UE, receiving, by the BS, from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, determining, by the BS, from the respective beams one or more beams for a communication with the UE using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the AI/ML model or functionality operates on the basis of input data, the input data comprising the measurements received from the UE, and adapting, by the BS, the input data of the AI/ML model or functionality responsive to a change of the beam configuration. The present invention provides a method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, transmitting, by the BS, respective beams to the UE respective beams, receiving, , by the BS, from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, determining, by the BS, one or more beams, which are to be used by the BS for a communication with the UE, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, obtaining the one or more beams to be used for the communication from at least one AI/ML model or functionality, which is operated at the BS according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE, or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE.

The present invention provides a method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, the UE measuring one or more first beams from a first set of beams supported by the UE and determining one or more second beams from a second set of beams supported by a first network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data, and transmitting, by the BS, to the UE a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for a BS.

The present invention provides a method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, wherein the UE is configured or preconfigured with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionalities, for performing a certain task, signaling, by the BS, to the UE to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to- use state allowing the UE to activate one or more of the non-used AI/ML models or functionalities for performing the certain task.

Computer Program Product

The present invention provides a computer program product comprising instructions which, when the program is executed by a computer, causes the computer to carry out one or more methods in accordance with the present invention.

Embodiments of the present invention are now described in more detail with reference to the accompanying drawing.

It is noted that the subsequently outlined and described aspects or embodiments may be combined such that some or all of the aspects/embodiments are implemented within one embodiment. Reference is made herein one or more AI/ML models and/or to one or more AI/ML functionalities. It is noted that when referring only to an AI/ML model, this is to be understood to refer also to an AI/ML functionality, and that when referring only to an AI/ML functionality, this it to be understood to refer also to an AI/ML model. AI/ML functionality may refer to an AI/ML-enabled Feature/Feature Group, FG, enabled by one or more configurations, where the one or more configurations may be supported based on one or more conditions indicated by a UE capability. An AI/ML-enabled Feature refers to a Feature where AI/ML may be used. It is noted that a UE may have one AI/ML model for the functionality, or the UE may have multiple AI/ML models for the functionality.

Further, when referring herein to determining one or more beams, this may mean selecting a set of beams. It may also mean to measure one or more beams, i.e. , determine certain parameters that characterize each of the one or more beams. For example, such parameters may comprise but are not limited to SINR, RSRP, RSRQ, SNR, etc. Furthermore, when referring to predicting a performance parameter or performance parameter value of a beam or predicting a beam using an AI/ML model or functionality, this may refer to the output of the AI/ML model or functionality with respect to said beam, where the output may be a predicted performance parameter value, e. g, a RSRP value, RSRQ value, SINR value, or SNR value, or an indication that the beam belongs to the set of Top- K beams, or a ranking index defining an order over the one or more beams. First Aspect

In the following, embodiments of the first aspect of the present invention is described in more detail.

Modified/New Value/Beam Reporting

Fig. 4 illustrates a user device, UE, 400 in accordance with embodiments of the first aspect of the present invention. The UE 400 includes a signal processor or signal processing module 402 and one or more antennas 404. The UE 400 receives, via the antenna 404, one or more reference signals from a network entity of the wireless communication network, for example the UE 400 receives from a gNB 406 the one or more reference signals over the Uu interface 408. The UE 400 may also receive the reference signals from a further UE 410 over the sidelink or PC5 interface 412. The UE 400 comprises a measurement module 414 to perform measurements of one or more of the received reference signals so as to obtain for each of one or more performance parameters, one or more performance parameter values. The measurement module 414 may also determine one or more beams received at the UE, e.g., from the gNB 406 or the UE 410 which are identified by the one or more reference signals. Further addition, the UE 400 includes an AI/ML module 416 operating or running at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality. The AI/ML module 416 may operate on the basis of input data 415, like the measurements obtained by the measurement module 414. Using the measurements 415, the AI/ML module 416 predicts, for a certain performance parameter, one or more values. The AI/ML module 416 may also predict, using the measurements 415, one or more of the above-mentioned beams.

In accordance with embodiments, the measurement module 414 and the AI/ML module 416 enable the UE 400 to obtain one or more of the performance values or beams by a measurement while one or more other performance values or beams are obtained by prediction. In accordance with yet other embodiments, the UE 400 is capable to obtain all of the values and/or all of the beams either by measurements performed by the measurement module 414 or by prediction performed by the AI/ML module 416.

The U E 400 further includes a reporting module 418 allowing the U E to transmit to a network entity, like the gNB 1406 or the UE 410, a report including the one or more performance parameter values and/or the one or more beams measured and/or predicted using the measurement module 414 and/or the AI/ML module 416. The report created by the reporting module 418 and transmitted by the UE 400 via the one or more antennas 404 to the network entity includes at least one predicted performance parameter value and/or at least one predicted beam. The report may be sent to the network entity which provided the one or more reference signals, like the gNB 406 or the UE 410, or to another network entity.

In accordance with embodiments, when predicting a parameter value or a beam using the AI/ML module 416, the reporting module 418 of the UE 400 generates the report and indicates that the report includes at least one predicted value and/or beam. For example, the reporting module 418 may include into the report a beam identifier for each of the one or more beams (measured or predicted) in the report and/or for each of one or more performance parameter values (measured or predicted) each associated with one of the one or more beams.

In accordance with embodiments, the UE 400 is preconfigured or configured with a report configuration. The report configuration configures the reporting module 418 of the UE 400 to include into the report generated by the reporting module 418 no (zero) or one or more measured performance parameter values and/or beams measured by the measurement module 141 , and at least one predicted performance parameter value and/or at least one predicted beam predicted by the AI/ML module 416.

For indicting that the report holds predicted values/beams, the reporting module 418 of the UE 400 modifies the report as defined by the report configuration, and/or modifies a transmission of the report, e.g., by transmitting the report with a priority, or a frequency, or a period, or a offset, or a delay, which is different from a priority, or a frequency, or a period, or a offset, or a delay, with which the report is transmitted according to the report configuration, e.g., for prioritizing certain content of the report.

In accordance with embodiments, the UE 400 receives the report configuration from one or more of the following: a Radio Access Network, RAN, entity, like the base station 406 or from another UE, like UE 410, a Core Network, CN, entity, an over the top, OTT, server, a WiFi access point, AP, or a WiFi station, STA.

In accordance with embodiments, the reporting module 418 of the UE 400 modifies the report by one or more of the following: omitting an entry from the report for which the performance parameter value/beam has been predicted, setting at least a part of an entry in the report, for which the performance parameter value/beam has been predicted, to a predefined value, e.g., to a default value, like zero, using at least a part of an entry in the report, for which the performance parameter value/beam has been predicted, for signaling additional information, using an unused part of an entry for signaling additional information, e.g., a confidence level or indicate whether AI/ML has been used to determine the value, replacing at least parts of one or more or all entries in the report, for which the value has been predicted, by additional information, replacing one or more or all entries in the report by additional information, compress a content of the report.

In accordance with embodiments, the reporting module 418 of the UE 400 compresses the content of the report using one or more or any combination of the following:

Huffman codes,

Run-Length encoding (RLE),

Time series compression, e.g., Gorilla or Prometheus approach,

Entropy codes, e.g. Golomb codes or Rice codes, Importance sampling, e.g. discard values based on their importance, Lempel-Ziv-Welsh compression, Lempel-Ziv compression,

- Quantization,

- AI/ML compression.

In accordance with other embodiments, rather than modifying a report as defined by a configured or preconfigured report configuration, the reporting module 418 of the UE 400 generates a new report, referred to herein as prediction report. The prediction report includes only the beam identifiers of the one or more beams, or the beam identifiers of the one or more beams and additional information for one or more of the beams. In accordance with other embodiments, the additional information is one or more of the following: prediction information, an indication that one or more of the beams identified in the report are unexpected, one or more earlier measurements or predictions of the performance values, e.g., a last measured or predicted CRI, or a last measured or predicted Layer 1 reference signal received power, L1-RSRP, a relative gain or loss to other beams, a relative gain or loss to one or more of: a last measurement, a last prediction, a last report, or a currently used beam, a rank of the beam in the report, a timestamp of a measurement or a prediction, duration of a measurement or a prediction, measured or predicted channel state information, CSI, an AI/ML model or functionality ID.

In accordance with other embodiments, the CSI may be one or more of the following: a measured or predicted Channel Quality Indicator, CQI, a measured or predicted Precoding Matrix Indicator PM I, a measured or predicted CSI Reference Signal, CSI-RS, resource indicator, CRI, a measured or predicted Synchronization Signal/Physical Broadcast Channel, SS/PBCH, Block Resource Indicator, SSBRI, a measured or predicted layer indicator, LI, a measured or predicted rank indicator, Rl, a measured or predicted Layer 1 reference signal received power, L1-RSRP, a measured or predicted Layer 1 Signal to Interference plus Noise Ratio, L1-SINR, a measured or predicted Capability Index, a measured or predicted Doppler and/Doppler delay profile, one or more measured or predicted time-domain channel properties, TDCP.

In accordance with other embodiments, the prediction information may one or more of the following: a prediction certainty, a confidence interval, a model ID of the AI/ML model, location information, like a location of the UE, a scenario in which the UE is used, a cell ID, a UE ID, for allowing the network entity to monitor a performance of the prediction, a timestamp of a measurement or a prediction, a confidence level, validity time, a confidence value indicating a confidence with the entries, a relative gain or loss to other beams, a validity of a prediction, e.g., a time for which a prediction is predicted to be valid, an average prediction error, an indication of an unexpected prediction outcome, e.g., spatially not close to a last used beam, a prediction diversity to evaluate whether a prediction algorithm tends to favor certain beams or directions over others across different scenarios, e.g. over varying channel conditions or user mobility.

The above-described embodiments of the first aspect of the present invention provide AI/ML reporting enhancements.

A further embodiment implementing a predicted beam reporting in accordance with the first aspect of the present invention is now described for a user device which predicts a downlink transmit beam, DL Tx beam, received at the UE from the network entity, like the gNB 406 or the UE 410 in Fig. 4. For predicting the DL Tx beam, the UE 400 uses the AI/ML model or functionality implemented by the AI/ML module 416 for determining a beam by predicting the best beam in the angular domain (beam management, BM, Case 1) or in the time domain (BM-Case2). The UE 400 predicts the beam IDs or the CSI-RS resource indicators, CRIs, or the TCI states or RSRP values associated with the beams using measurements obtained by the measurement module 414 as an input 415 into the AI/ML module 416.

A conventional reporting framework, as described above with reference to Fig. 2, requires that a UE reports the CRI and the L1-RSRP of a measured beam. Such a report is configured using the CSI-ReportConfig information element, which causes the reporting module 418 of the UE 400 to generate a report that includes the CRIs and the L1-RSRPs of up to the four most powerful or strongest beams, dependent on the actual configuration. The CSI-RS resource indicator, CRI, is a reference to the CSI-RS resource previously received for the channel measurements performed by the measurement module 414 of the UE 400. The bottom table of Fig. 2 illustrates a conventional beam report for up to four beams. The UE provides a CRI for each reported beam and further reports an associated RSRP value for each reported beam.

However, the conventional beam report includes the RSRP values which may be based on outdated measurements by the measurement module 414. Such values may not contribute in helping the gNB 406 or the UE 410 to decide which of the DL Tx beams to use for communication with the UE 400. Hence, in accordance with embodiments of the first aspect of the present invention, the reporting module 418 may simply omit the RSRP fields from the report which is configured in accordance with the CSI-ReportConfig. Fig. 5 illustrates the beam reporting based on the prediction in accordance with embodiments of the first aspect of the present invention. A base station or gNB 406 provides a plurality of DL Tx beams #0 to #4 and the UE 400 predicts, using the AI/ML module 416, the top three CSI- RS at a time instance #n, as is indicated at 420 in Fig. 5, and, at a time instance #n+k, the same or different top-CRIs, as is indicated at 422. The reporting module 418 of the UE 400 creates the CSI prediction report 420 at time #n and the CSI prediction report 422 at time #n+k from which the actual L1-RSRP values or fields are omitted. In other words, the CSI prediction report 420, 422 only includes the CRIs of the three strongest or most powerful beams predicted at the time instance #n and at the time instance #n+k. In accordance with other embodiments, instead of omitting the RSRP parameter or RSRP field from the report, the predicted values may be replaced by a default value indicating to a receiver of the report that the respective beams indicated in the report are predicted beams.

Fig. 6 illustrates another example for a beam reporting based on a prediction in accordance with embodiments of the first aspect of the present invention. Again, a base station or gNB 406 provides a plurality of DL Tx beams #0 to #4 and the UE 400 predicts, using the AI/ML module 416, the top three CSI-RS at a time instance #n, as is indicated at 420 in Fig. 6, and, at a time instance #n+k, the same or different top-CRIs, as is indicated at 422. More specifically, at time instance #n, the CSI prediction report includes the three best beams which have been predicted by the AI/ML module 416 the DL Tx beams #2, #3 and #4, and in the CSI prediction report, the respective CRI is followed by an associated confidence value predicted by the AI/ML module 416. At the time instance #n+k, one may see that the CSI prediction report includes different three most powerful beams, namely DL Tx beams #3, #4 and #2, again indicated together with the confidence values, where a high value may indicate a high confidence that the associated beam is one of the strongest K beams, K being a natural number, and a low value may indicate a low confidence that the associated beam belongs to the top-K beams.

The gNB 406, upon receiving a report 420’ or 422’ determines from the missing RSRP parameter or from the respective beams being associated with a default value that the report includes beams which have been predicted at the UE 400 using the UE’s AI/ML module 416. The gNB 406 may use this knowledge and take one or more of the following actions:

Verify the prediction by triggering a measurement.

Reduce the CSI-RS to beams that are predicted to be of good link quality.

Prioritize scheduling of resources for the predicted beams to optimize network efficiency and user experience.

Adjust the beamforming strategy to enhance the signal quality for the predicted beams.

Configure additional measurements for UE to refine the prediction accuracy.

Dynamically control power for the predicted beams to maintain optimal signal strength and reduce interference.

Trigger a handover for the said UE, e.g., in case the UE cannot be served with beams of a certain quality, or configure the UE with a CHO.

Serve the UE by using a different TRP, e.g., in case of multi-TRP deployments.

Inform neighboring gNBs about the report from the said UE, e.g., in order for neighboring gNBs to preconfigure resources for a potential HO or CHO.

Aggregate a further carrier for the said UE, e.g., in case the UE cannot be served better dependent on the report sent by the UE.

In accordance with other embodiments of the first aspect of the present invention, the reporting module 418 of the UE 400 may be configured or preconfigured to operate according to a new reporting mode if the report to be provided to the gNB 406 or the UE 410 includes at least one or some predicted values or beams. This report may be referred to as a CRI-Predict-Report and, in accordance with embodiments, the reporting module 418 of the UE is configured or preconfigured to use the new reporting mode to only report the CRIs and/or the SSBRIs of the predicted top-k beams, for example the k strongest beams. The value k may be a configurable number larger or equal to one. The CRI-Predict-Report may be used in certain situations as it reduces the size of the CSI report substantially which is transmitted via the PLICCH. For example, the CRI-Predict-Report may be used in situations in which the data rate is limited, so that any traffic on the PLICCH is to be reduced.

In accordance with further embodiments of the present invention, the above-described report generated by the reporting module 418, i.e. , the report generated in accordance with a CSI-ReportConfig information element or the report generated in accordance with the new reporting mode, namely the CRI-Predict-Report, may include one or more CRIs which are actually measured so that in such scenarios, for example, the CSI prediction reports 420’ and 422’ from Fig. 6 may include for one of the indicated CRIs the measured L1-RSRP, and the CRI-Predict-Report may include, in addition to the CRIs and/or SSBRIs of the predicted beams, the CRIs and/or SSBRIs of at least one measured beam. In accordance with further embodiments, the reporting module 418 may create the reports in such a way that in addition to the L1-RSRP or instead of the L1-RSRP also a confidence or reliability of the CRIs is included in the report is indicated. The confidence or reliability of the CRIs may be associated with a value ranging from zero to a fixed or configured number n, where zero describes a very low confidence that the associated beam/CRI is the top-1 beam, and n describes a very high confidence that the associated beam/CRI is the top-1 beam, e.g., the best or strongest beams among those indicated in the report.

In accordance with embodiments, the confidence of each beam/CRI may be signaled as an absolute value or such that only the highest value is signaled explicitly while the remaining confidence values are signaled relative to the highest or a previous value using less bits than for the first value. This is described in more detail below with reference to the embodiments concerning a differential reporting.

In accordance with embodiments, the reporting module 418 may apply a certain ordering of the CRIs in the report provided to the gNB 406 or to the UE 410. In accordance with embodiments, the ordering of the CRIs may be based on a measured L1-RSRP, a predicted L1-RSRP or a confidence value, meaning that within the report, if present, CRIs having the highest measured or predicted L1-RSRP or the highest confidence level, are the first CRI in the report.

In accordance with embodiments, the CRI-Predict-Report may include one or more of the following: an ID to identify the beam/CRI, a TRP ID, e.g., to identify a TRP in case of multi-TRP systems, a last measured L1-RSRP, a predicted RSRP, a confidence level, a relative gain or loss to other CRI/beams, a relative gain or loss to last measurement / prediction I report / currently used beam, an order of the beams, e.g. if a beam it is the best, 2nd best ... , a validity of prediction (time the prediction is predicted to be valid) reported in e.g. ms, s, slots, radio frames, an average prediction error based on past measurements, a prediction diversity to evaluate whether the prediction algorithm tends to favor certain beams or directions over others across different scenarios, e.g. varying channel conditions or user mobility, an AI/ML model or functionality ID.

In accordance with further embodiments, the above described reports generated by the reporting module 418 of the UE 400 may include other information, e.g., information of on the channel state, like a channel quality indicator, CQI, a precoding matrix indicator, PMI, a CSI-RS resource indicator, CRI, a SSI/PPCH block resource indicator, SSI BRI, a layer indicator, LI, a rank indicator, Rl, L1-RSRP, L1-SINR, a capability index or time-domain channel properties, TDCP. The SSBRI may be reported with L1-RSRP measurements of a secondary synchronization signal, SSS.

In accordance with further embodiments, the above-described reports generated in accordance with the conventional report configuration and the new report, CRI-Predict- Report may be combined, thereby, for example, enhancing the CRI-Predict-Report by including one or more measured values of certain performance parameters, like a broadband measured L1-RSRP measured on the SSS. In another embodiment, the basic structure of the conventional approaches, as indicated in the tables of Fig. 2 above, is not changed, rather, the UE 400 reports a default value for the L1-RSRP values, for example zero.

In accordance with yet further embodiments, the report may be based on a conventional report configuration, for example using the structure as described above with reference to Fig. 2. The L1-RSRP value is defined by a 7-bit value in the range of [-140, -44] dBm with a 1dB step size. Using 7 bits allows the coding of 128 values, however, only 96 values, namely the values -140 to -44 are coded 32 values remain unused. In accordance with embodiments, these unused values or states are used for other or additional information. For example, assuming a signaling range of 0 to 255 but only using 0 to 200 to indicate a signaling strength, the values 201 to 255 may be used for indicating further information, for example a surprise or confidence level. If such information is more important than the actual beam strength, the beam strength information may also be overwritten. In another example, for beams being stronger than expected it may be indicated whether this was expected or not. For example, if a beam of strength 190 was expected, one might signal 190 but 245 is signaled if the beam was unexpected.

Differential Reporting of Measured/Predicted Values

Further embodiments of the first aspect of the present invention provide for enhancements in the reporting of measured and/or predicted values in a report provided by a UE.

For example, when considering the use of AI/ML and the need for transmitting information via the air interface, enhancements of the reporting of measurement result or predictions are needed, for example in situations in which an AI/ML model receives input data, like certain measurement results, over the NR air interface as it may be true for the above- mentioned BM-Case1 and BM-Case2. For example, when considering the L1-RSRP reporting, the measurement results of more than four beams for the reporting of other performance parameter values may be required at one reporting instance. Measuring more beams than are currently measured in conventional approaches, namely more than four beams, may be desired for a more efficient and more accurate performance of an AI/ML model receiving the measurements as an input. However, such a larger number of measurements may contribute to a significant amount of signaling that needs to be reported regularly and thus contributing to the signaling overhead. As has been described above, in the conventional approach using CSI-RS reporting, there is one reporting location for which a UE determines up to the four strongest beams and reports the CRI and the associated L1-RSRPs for these beams. In order to reduce the overhead, the UE may use a differential L1-RSRP based reporting in accordance with which the larges measured value of L1-RSRP is quantized to a 7-bit value in the range of [-140, -44] dBm with a 1dB step size. The remaining L1-RSRPs having a measured value smaller than that of the L1-RSRP, are differentially reported. A differential L1-RSRP is quantized to a 4-bit value computed with a 2dB step size with reference to the largest measured L1-RSRP value. While the conventional approach works reliably with a small number of beams having closely located L1-RSRP values with a small relative value to the largest measured value, it no longer works well when the number of beams to be reported increases, especially when the number of beams increases substantially. For example, when considering an increase of the beams to be reported to 64 beams or even more beams as expected to be used in 6G systems, the above-described conventional approach is no longer feasible since there may be very weak beams with a large relative value to the largest measured value.

For addressing this issue, further embodiments of the first aspect of the present invention concern a new differential reporting of one or more values for certain performance parameters, like the above-described L1-RSRP. In accordance with such embodiments, a UE 400, as it is depicted and explained above with reference to Fig. 4 may be employed for reporting only measured values, or only predicted values or for both measured and predicted values. In other words, the new differential value reporting may be employed also for conventional reports including only measured values or for the above-described reports, like the CRI-Predict-Report, including predicted values or for reports which include both predicted and measured values. Further, although the embodiment is described with reference to the differential value reporting in a report provided by a UE, like UE 400, it is noted that the differential value reporting in accordance with embodiments of the first aspect of the present invention may also be implemented in another entity, like the gNB 406 which may provide to the UE 400 measurements to be used at the UE 400, for example for deciding on a receive beam to be formed by the UE 400 using, for example, the AI/ML module 416.

In accordance with embodiments, the reporting module 418 may receive one or more measured or predicted values from the measurement module 414 or from the AI/ML module 416, and the plurality of measured values of a certain performance parameter, like the plurality of L1-RSRP values obtained for the plurality of beams to be included into a report, may be differentially indicated as follows. The largest measured or predicted L1-RSRP value is quantized as an absolute value to k-bits. All remaining weaker beams are differential L1-RSRP values which are computed with a reference to a previous L1-RSRP value and which are quantized to m-bits with m being smaller than k. Fig. 7 illustrates a differential RSRP reporting in accordance with the just described embodiment. The vertical axis labelled delta indicates the respective quantization levels from 1 to 6, and the vertical axis labelled RSRP shows a series of measurement values for beams 5, 7, 9, 3 and 13 as arrows, i.e., in the embodiment of Fig. 7 it is assumed that the report includes the measured/predicted RSRP values for the DL Tx beams 5, 7, 9, 3 and 13 indicated by the UE to be the strongest beams at a certain time instance. As is indicated in Fig. 7, beam 5, which is the beam having the largest RSRP value, is encoded with an absolute quantization. The next weaker RSRP value of beam 7 and is encoded relative to beam 5, the RSRP value of beam 9 is encoded relative to beam 7, the RSRP value of beam 3 is encoded relative to beam 9, and the RSRP value of beam 13 is encoded relative to the RSRP value of beam 3.

The above-described quantization steps may be coarser or finer and may be configured or preconfigured, for example, by the gNB 406.

In accordance with further embodiments, the largest measured or predicted L1-RSRP value is quantized as an absolute value to k-bits, and all remaining weaker beams are differential L1-RSRP values which are computed with reference to a previous measured L1-RSRP, in a similar way as described above with reference to Fig. 7, however, they are quantized to a vector of [rm, m2, ... , m_n]-bits, where rm < rm < rm ... < m_n < k. In accordance with yet other embodiments, there may be more than one absolute quantized value, and the delta quantization is then relative to one of the absolute quantized values. For example, every n- th measured or predicted L1-RSRP value may be quantized as an absolute value and all the remaining weaker beams are differential L1-RSRP values which are computed with a reference to a previous measured L1-RSRP value. For example, when assuming six L1- RSRP values V1 to V6, and when assuming that two of the L1-RSRP values, like V1 and V4, are quantized as an absolute value, the L1-RSRP values V2 and V3 may be encoded relative to V1 and V2, respectively, and the L1-RSRP values V5 and V6 may be encoded relative to the value V4 and V5, respectively, as described above with reference to Fig. 7, i.e., the following values are encoded: V1 , V1-V2, V2-V3, V4, V4-V5, V5-V6. In accordance with other embodiments, the differential L1-RSRP values may also be differentially encoded with reference to the preceding or previous absolute L1-RSRP value. Assuming the above embodiment of the RSRP values V1-V6, this means that the values V2 and V3 are differentially encoded with reference to the absolutely encoded value V1 and that the values V5 and V6 are relatively encoded with reference to the absolutely encoded value V4, i.e., the following values are encoded: V1, V1-V2, V1-V3, V4, V4-V5, V4-V6. In accordance with further embodiments, a performance value, which is greater than a smallest performance value, is quantized with reference to a next smaller performance value. For example, when assuming again the six L1-RSRP values V1 to V6, the smallest of the L1-RSRP values, V6, is quantized as an absolute value, the remining L1-RSRP values V1 to V5 are encoded relative to the next smaller value as follows: V6, V5-V6, V4- V5, V3-V4, V2-V3, V1- V2.

In accordance with yet further embodiments, a smallest performance value and at least one further performance value, which is greater than the smallest performance value, are quantized as absolute values, a performance value, which is greater than the smallest performance value and smaller than the at least one further performance value, is quantized with reference to the smallest performance value or with reference to a next smaller performance value, and a performance value, which is greater than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next smaller performance value. For example, when assuming again six L1-RSRP values V1 to V6, and when assuming that two of the L1-RSRP values, like V3 and V6, are quantized as an absolute value, the L1-RSRP values V5 and V4 may be encoded relative to V6 and V5, respectively, and the L1-RSRP values V2 and V1 may be encoded relative to the value V3 and V2, i.e. , the following values are encoded: V6, V5-V6, V4-V5, V3, V2-V3, V1-V2. In accordance with other examples, the differential L1-RSRP values may also be differentially encoded with reference to the preceding or previous absolute L1-RSRP value. Assuming the above embodiment of the RSRP values V1-V6, this means that the values V1 and V2 are differentially encoded with reference to the absolutely encoded value V3 and that the values V4 and V5 are relatively encoded with reference to the absolutely encoded value V6, i.e., the following values are encoded: V6, V5-V6, V4-V6, V3, V2-V3, V1-V3.

It is noted that the present invention is not limited to the above-described reporting of beams or RSRP values, rather, the above-described approach using the novel differential reporting may be employed for reporting a plurality of values for any performance parameter, e.g., one or more of the following: one or more beams, which are transmitted by a network entity of the wireless communication system and received at the UE, the performance value indicating a measured or predicted strength of a beam at the UE, a reference signal received power, RSRP, the performance value indicating the measured or predicted RSRP, a reference signal received quality, RSRQ, the performance value indicating the measured or predicted RSRQ, a signal to noise ratio, SNR, the performance value indicating the measured or predicted SNR, a rank,

- a packet loss rate, the performance value indicating the measured or predicted packet loss rate, one or more parameters reported from higher layers, the performance value indicating the measured or predicted values for the one or more parameters, a measured or predicted Doppler profile and/or Doppler delay profile.

In accordance with the above-described embodiments, the new differential reporting is not limited to UE, but can also be implemented at the network side, e.g. at a gNB. For example, the gNB can measure Sounding Reference Signals (SRS) and report the performance parameter values back to optimize beamforming at a UE. The gNB may provide the UE with the information on multiple beams, e.g., SINR, SNR, RSRP, RSRQ. Next, the UE can select appropriate one or more beams for uplink transmission. Another case scenario that can happen at the gNB is coordinated beamforming, which involves multiple gNBs. In this case, the gNBs can exchange information and provide guidance to the UE, e.g., a serving gNB can inform the UE about beams transmitted by neighboring gNBs that may cause interference. In a further embodiment, the serving gNB can prepare a handover or CHO for a UE by informing the UE about potential beams or by providing a quality of neighboring beams to the said UE. Thus, in accordance with the above-described embodiments, the reporting module 418 of the UE 400, or a corresponding module of another network entity, like the gNB 406 of the further UE 410 may include into a report for a certain performance parameter the one or more of measured or predicted values such that a performance value, which is smaller than a greatest performance value, is quantized with reference to a next greater performance value, or a performance value, which is greater than a smallest performance value, is quantized with reference to a next smaller performance value, or a greatest performance value and at least one further performance value, which is smaller than the greatest performance value, are quantized as absolute values, a performance value, which is smaller than the greatest performance value and greater than the at least one further performance value, is quantized with reference to the greatest performance value or with reference to a next greater performance value, and a performance value, which is smaller than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next greater performance value, or a smallest performance value and at least one further performance value, which is greater than the smallest performance value, are quantized as absolute values, a performance value, which is greater than the smallest performance value and smaller than the at least one further performance value, is quantized with reference to the smallest performance value or with reference to a next smaller performance value, and a performance value, which is greater than the at least one further performance value, is quantized with reference to the at least one further performance value or with reference to a next smaller performance value.

In accordance with embodiments, the greatest performance value is quantized as an absolute value using a first number of bits, and for all remaining performance values, which are smaller than the greatest performance value, a difference of the performance value with reference to the next greater performance value determined, and the difference is quantized using a second number of bits, wherein the second number is smaller than the first number.

In accordance with other embodiments, the greatest performance value and at least one further performance value, which is smaller than the greatest performance value, are quantized as absolutes values using a first number of bits and a second number of bits, respectively, for all performance values, which are smaller than the greatest performance value and greater than the at least one further performance value, a difference of the performance value with reference to the greatest performance value or with reference to the next greater performance value is determined, and the difference is quantized using a third number of bits, wherein the third number is smaller than the first number, and for all performance values, which are smaller than the at least one further performance value, a difference of the performance value with reference to the at least one further performance value or with reference to the next greater performance value is determined, and the difference is quantized using a fourth number of bits, wherein the fourth number is smaller than the second number.

Second Aspect

Embodiments of the second aspect of the present invention are now described in more detail. The embodiments of the second aspect may be implemented using a user device, like the one described above with reference to Fig. 4.

AI/ML reporting enhancements

In accordance with embodiments, the UE 400 determines one or more beams, for example, one or more strongest beams or determining the strength, e.g., RSRP, RSRQ, SNR, SINR, of one or more beams, from a plurality of beams which are transmitted by a network entity and received at the UE 400. For example, the network entity may be the gNB 406 in Fig. 4 providing, in a way that it is described above with reference to Fig. 5 or Fig. 6, a plurality of DL Tx beams at least some of which are received at the UE 400. In accordance with other embodiments, the beams may also be provided by another network entity, like the further UE 410. Using the measurement module 416, the UE 400 determines from the plurality of DL Tx beams one or more beams by measuring one or more reference signal resources associated with the respective beams. In addition or alternatively, the UE 400 may use the AI/ML module 460 for determining, i.e., for predicting, the one or more beams. The UE 400 uses the reporting module 418 for sending a report about the one or more beams determined by the UE 400 back to the network entity, like the gNB 406 or the UE 410, for example to the network entity which also transmitted the plurality of DL Tx beams towards the UE 400.

In accordance with embodiments, the UE 400 receives a configuration defining how the reporting module 418 is to create the report for reporting the one or more beams to the network entity. In accordance with the embodiments, the UE 400 may select the one or more beams from the plurality of beams based on a configured or preconfigured rule. For example, the UE 400 may select the strongest K beams, e. g., having a high measured or predicted RSRP, from the plurality of beams, where K is a configured or preconfigured number.

In accordance with embodiments, the UE 400 may include, for example in the AI/ML module 416, a further AI/ML model or functionality which may be used by the reporting module 418 to generate the report using the AI/ML model or functionality. In other words, a further AI/ML model or functionality may be provided which is different from the prediction AI/ML model or functionality for predicting the one or more beams. The further AI/ML model or functionality may receive the predictions from the prediction AI/ML model or functionality as well as measurements from the measurement module 414 or only the measurements from the measurement module 414, and generate on the basis of this input the report. In accordance with other embodiments, instead of using a further, separate AI/ML model in addition to the prediction AI/ML model, the reporting module 418 may use the prediction AI/ML model or functionality also for generating the report.

In accordance with embodiments, the reporting module 418 may determine from the configuration how the report is to be generated, and report the measurements provided by the measurement module 414, or the one or more beams predicted by the AI/ML module 416, or to both.

In accordance with further embodiments, the UE 400, more specifically the measurement module 414 and the AI/ML module 416 thereof, may determine a set of measured beams and a set of predicted beams, with each set including one or more beams. The sets of predicted and measured beams include the same beams, different beams, disjoined beams or overlapping beams. The number of beams that a UE can measure is limited, e.g., due to a limited number of reference symbols which can be embedded into the signal. Thus, using AI/ML in addition to measurements can improve picking the best beam for the said UE. Furthermore, the AI/ML module can take other side information into account when predicting beams, which would otherwise not be used when solely relying on measurements. Thus, when comparing and/or combining measurement results and prediction results, it may be possible to choose a better beam or a better proper subset of beams for the said UE, so that the UEs connectivity can be improved. This would result in a better overall performance, e.g., with reference or with respect to (wrt.) data rates or lower delays or less radio link failures, RLF, for the said UE. For example, if the sets include the same beams, the UE 400 may report for each beam a measured value and a predicted value. This allows the network to have a more detailed picture of the channel at the UE 400. In another example, the sets may be disjoint, then the UE would report measured values for some beams and predicted values for some other beams. This would reduce the measurement overhead, as less beams have to be measured. At the same time, the network would receive the actual measurements from the UE, which may be more reliable compared to predicted values. In a further example, where the beams are overlapping, the UE may report measured and predicted values for some beams and, e. g. , only predicted values for some other beams. This approach reduces again the number of required measurements. However, compared to the case of disjoint sets, the network receives more information allowing better decisions at the network.

Embodiments of the second aspect of the present invention provide AI/ML reporting enhancements by reusing an existing measurement reporting frame for reporting the measured and/or predicted beams. For example, when considering a conventional CSI reporting framework, this offers a flexible way to report AI/ML predicted beams to the network. A conventional CSI-ReportConfig may be used to configure the UE 400 with respective CSI reporting opportunities, which may be periodic or aperiodic occasions. In the CSI report the UE reports the CRI and the L1-RSRP or SINR for certain beams, for example for n strongest beams, with n being four 4 in the conventional approaches. In other words, up to the four strongest beams may be reported. In accordance with embodiments of the second aspect of the present invention, such an existing measurement reporting frame is reused for reporting the predicted beams so that instead of the measured beams, like the one or more strongest measured beams, the UE 400 reports the n strongest predicted beams.

Fig. 8 illustrates AI/ML reporting enhancements may be provided according to embodiments of the second aspect of the present invention. In accordance with embodiments, the UE, like UE 400 of Fig. 4, is configured or preconfigured with a CSI-ReportConfig determining the periodic CSI occasions labelled “CSI” at times ti, t2 and t₃. As is indicated at 430, measurements 415 obtained by the UE 400 during a certain period before an actual time to, referred to as the duration or time window to-thistory, are input into the AI/ML module 416 of the UE 400. The UE 400, for example responsive to a certain event, triggers a CSI prediction for predicting the one or more beams, like the n strongest beams at the time to. At the time to, in accordance with the CSI reporting framework, a CSI report is provided by the reporting module 418 which includes the n strongest beams predicted by the AI/ML module 416. In accordance with embodiments of the second aspect of the present invention, the configuration with which the UE may be configured or preconfigured may indicate, implicitly or explicitly, that a report is to be generated which includes measured or predicted beams, or which includes measured and predicted beams.

In accordance with embodiments, for indicating that a report is to include predicted beams, an existing CSI measurement reporting framework may be used, for example, an existing CSI-ReportConig. Conventionally, the CSI-ReportConfig requires a link to a CSI- ResourceConfigld which indicates the presence of at least one CSI-RS resource, i.e. , the presence of at least one beam, which is associated with the reporting configuration. In other words, the UE 400 may determine from the received configuration the respective CSI-RS resources to be measured or monitored, with each CSI-RS resource being associated with a certain DL Tx beam (see Fig. 5 and Fig. 6) provided by the gNB. In accordance with embodiments, for implicitly indicating that a report is to include predicted beams, the CSI- ReportConfig configuring the CSI reporting opportunity at to, referred to as “predicted CSI”, does not include any actually transmitted CSI-RS resources and/or contains at least one Zero-power or non-transmitted CSI-RS resource in the channel measurement configuration or in the channel prediction configuration. In other words, the CSI report configuration or the CSI-ResourceConfig or the CSI-RS resource set does not include any non-zero power-CSI- RS, NZP-CSI-RS, resources and/or at least one zero-power or non-transmitted CSI-RS resource. When determining from the CSI report configuration or from the CSI resource configuration that no resources are indicated and/or only or partially Zero-power or nontransmitted CSI-RS resources are indicated, the UE 400, like its reporting module 418, judges that the next report to be generated at the CSI reporting opportunity defined by the configuration is to include one or more predicted beams. In this way, the UE 400 identifies that the configuration is associated with AI/ML beam management and, hence, may be used to report the prediction results instead of the measurement results.

In accordance with other embodiments, the configuration may explicitly indicate to the UE 400 that predicted beams are to be included into the report. In accordance with such an embodiment, an Al indicator field may be provided in the CSI-ReportConfig and may indicate that the current reporting configuration received by the UE 400 allows for reporting only predicted results or predicted and measured results. This indicates to the UE 400 that it may report in an associated reporting occasion only predictions, i.e., only predicted beams, like a CRI, beam ID, L1-RSRP and the like associated with respective predicted beams, or predictions and measurements, i.e., a CRI, beam ID, L1-RSRP or the like for the beams in the report determined by the measurement module 414 and the prediction module 416.

In accordance with other embodiments, an Al indicator field may indicate that an associated non-Zero-power CSI-RS resource is not available for measurement. In other words, the reporting configuration may include some or only non-zero-power CSI-RS for channel measurement. However if some or all of the non-Zero-power CSI-RS are associated with an Al indicator or an Al reporting occasion, the said non-zero-power CSI-RS would not be transmitted and the UE would not measure them but instead report predictions in the reporting occasion. For example, the Al indication may cause the UE to report only predictions for some or all CSI-RS of a certain reporting configuration. In another example, the UE may be configured or preconfigured with an Al periodicity or measurement periodicity n, which indicates that every n-th reporting occasion includes predictions or every n-th reporting occasion includes measurements.

In accordance with embodiments, the report comprises one or more of the following: a CSI-RS resource indicator, CRI, a beam ID, a transmission reception point identifier, TRP ID, e.g., as may be used to identify multi-TRPs, an identification of a TRP, e.g., in case of multi-TRP, which can be implemented by using a CORSET pool index, one or more beams, e.g., identified by a beam ID or by a resource ID, e.g., CRI, Channel State Information, CSI, e.g., a Channel Quality Indicator, CQI, a precoding matrix indicator, PMI, a CSI reference signal, CSI-RS, resource indicator, CRI, a Synchronization Signal/Physical Broadcast Channel, SS/PBCH, Block Resource indicator, SSBRI, a layer indicator, LI, a rank indicator, Rl, a Layer 1 reference signal received power, a Layer 1 Signal to Interference plus Noise Ratio, L1-RSRP, L1- SINR, a Capability Index or time-domain channel properties, TDCP, a received signal strength indicator, RSSI, a higher layer CSI, e.g., Layer 3 reference signal received power, L3-RSRP, or a Layer 1 Signal to Interference plus Noise Ratio, L3-SINR. In accordance with embodiments, the UE 400 reports, in addition to the one or more measurements, also the one or more predictions, e.g., a predicted CRI, a beam ID of a predicted beam, a predicted L1-RSRP.

Timing of predictions

Further embodiments of the second aspect of the present invention concern the timing of predictions. The UE 400 may report predictions dependent on a time between the occasion for reporting the predictions and a latest measurement or prediction of beams. The reporting module 418 of the UE 400 may determine one or more beams among the beams received at the UE, like the DL Tx beams provided by the gNB (see Fig. 5 and Fig. 6) at one or more configured or preconfigured occasions, for example at one or more first occasions at which the one or more beams are determined by the measurement module 414 or at one or more second occasions at which the one or more beams are predicted by the AI/ML module 416.

Fig. 9 illustrates a prediction timing in accordance with embodiments and the plurality of first occasions 440, which may be measurement windows 440 including one or more reporting occasions CSI1 , CSI2, and the one or more second occasions 442, which may be the prediction windows including reporting occasions 444 at which CSI predictions may be reported. During a measurement window 440 the measurement module 414 may take respective measurements for determining one or more beams, like the n strongest beams, from a plurality of DL Tx beams received at the UE 400. During the prediction windows 442, which are at time periods between the respective measurement windows 440, beams may be predicted using the earlier measurements. In accordance with other embodiments, the prediction window and/or the measurement window may be indicated explicitly, e. g. , by configuration or preconfiguration, or implicitly, e. g. , by the timing of the reporting occasion and/or the occasions at which the measurements and/or predictions took place. In accordance with further embodiments, dependent on a temporal relationship between the reporting occasion and the occasions at which the measurements/predictions took place, the reporting module 418 may decide to include into the report at least one predicted beam and/or at least one measured beam.

In accordance with embodiments, the actual number of first and/or second occasions depends on one or more criteria, e.g., on one or more of the following: a distance to a base station, e.g. dependent on a position of the UE in a cell of the wireless communication network the UE is configured with a first number of second occasions when being at a first distance from the base station of the cell and with a second number of second occasions when being at a second distance from the base station of the cell, the first number of second occasions and the first distance being greater than the second number of second occasions and the second distance, a distance from the UE to one or more other UEs or to one or more Radio Access Network, RAN, entities,

- whether an AI/ML model is used for prediction, a prediction accuracy of the AI/ML model used, a mobility of the UE, e.g. speed with which the UE is moving, a required QoS of a service running on the UE, a required HARQ, like a number or ratio of ACKs or NACKs, a carrier frequency, e.g., FR1 or FR2 or FR3, a battery or power state, e.g. battery level or charging state, a scenario, e.g. an urban microcell, UMi, an urban macrocell, UMa, a rural microcell, RMi, a rural macrocell RMa, or indoors, a channel condition, e.g., LOS, NLOS, dependent on the delay of the strongest path, an interference level, e.g. below or above a certain threshold, a type of UE, e.g. loT device, eMBB device, a vehicular UE, V2X, or a pedestrian UE, P-UE.

In accordance with further embodiments, a minimum timer after a last CSI-RS reception may define a time window within which the UE reports rather the measurement results instead of prediction results, at least for the associated CSI-RS, i.e., for the associated beam. For example, when the UE 400 recently measured a CSI-RS, i.e., a beam, it reports the measurement result in a reporting opportunity that is associated with this CSI-RS, if the reporting opportunity is within the minimum timer after the reception of the CSI-RS. Otherwise, the UE reports the output of the AI/ML module 416, i.e., a predicted beam and/or a predicted L1-RSRP. In other words, the UE 400 may report a measured beam, if a time between a first occasion and the reporting occasion is less than a configured or preconfigured threshold, and/or report a predicted beam, if the time between a first occasion and the reporting occasion is more than a configured or pre-configured threshold. Also, the UE may predict the one or more beams at a certain second occasion and/or report the predicted one or more beams at the reporting occasion only if a time between a first occasion and the certain second occasion and/or the reporting occasion is more than a configured or pre-configured threshold.

In general, reporting is configured by a RAN entity, e.g., the gNB. Nevertheless, when using the novel concept of AI/ML models or functionality, the triggering of reporting can be enhanced. Thus, in a further embodiment, a reporting occasion can be trigged by a condition. The condition may depend on the computing process or output of the AI/ML calculation, wherein the condition can be one or more of a finished computing of an AI/ML calculation, a failed computing of an AI/ML calculation, e.g., the AI/ML module could not finish the computation within a certain time or the calculation failed completely, an indication by the AI/ML model or functionality, e.g., the AI/ML module output of AI/ML calculation, one or more results of AI/ML calculation being o above or below a configured and/or preconfigure threshold or o changed wrt. the last calculated or reported result or o changed wrt. the last calculated or reported result is above or below a configured and/or preconfigure threshold.

In accordance with embodiments, the predictions can be defined as measurement reports without measurements. In this case the measurement is running the model/prediction. In this case a prediction is configured. This can be done by reusing the measurement configuration framework by introducing new beam prediction values for measurement. In this case the measurement is redirected to a prediction model. Further it can be implicitly configured by requesting a measurement without configuring reference signal positions.

In accordance with certain embodiments, predictions are equated to measurement reports that lack actual measurements; here, the act of measurement is replaced by executing the prediction model. The predictions are configured by adapting the existing measurement configuration framework to incorporate new prediction metrics. Consequently, measurements are effectively rerouted to utilize a prediction model. Additionally, predictions can be configured implicitly by requesting measurements without setting up reference signal positions. The benefit of this is that the existing measurement reporting framework can be reused, ensuring backwards compatibility with existing 5G UEs. A single bit within the report can indicate whether the report is based on a measurement or on a prediction, which may cause the gNB to treat the measurement report differently, e.g., it may configure less measurement reporting if the results are good, or in another embodiment, it will indicate to perform “real” measurements or report “real” measurement, in case the performance of the UE using AI/ML prediction is degraded.

In one embodiment, a measurement configuration is possible wherein the "measured" parameter is essentially a pointer to a prediction model's output.

In another scenario, a measurement, like a beam measurement using Channel State Information-Reference Signals (CSI-RS), can be set up even when the network has not configured any CSI-RS. This results in the absence of a measurable signal for the User Equipment (UE), prompting the UE to carry out a prediction as an alternative.

In accordance with further embodiments, no prediction reports may be provided at occasions being within a certain range from the latest measurement reports. The prediction intervals and/or occasions may be intersected by actual measurement periods, see in Fig. 9 the measurement windows or periods 440 and the prediction intervals or windows 442. The presence of measurements actually precludes the necessity for predictions. For example, a periodic prediction is only executed or reported if it is more than a threshold in time away from a measurement. In other words, a prediction may not be performed by the AI/ML module 416 of the UE 400 in case the time instance of the prediction is within a predefined time around a measurement - either a measurement to be performed in the future (e.g., the UE assumes then that it gets more reliable values soon) or a measurement that already occurred (e.g., the UE assumes then that the earlier measured values are still valid). In accordance with other embodiments, the prediction may be performed at the mentioned time within the predefined time period, however, given the fact that measurement reports have been sent shortly before the prediction or are sent shortly after the prediction time, the reporting module 480 of the UE 400 does not generate or send a report including the predicted results. Thus, in case a previous measurement was too recent, with reference to the prediction instance, the UE either avoids the prediction or skips the reporting about the prediction. In such a situation, the UE 400 may perform a DRX and/or may wait for the next measurement or reporting occasion.

Beam measurement enhancements

Further embodiments of the second aspect of the present invention provide beam measurement enhancements. In accordance with embodiments, the predictions may be performed periodically or aperiodically as defined, for example, by a configuration or pre- configuration the UE 400. In accordance with other embodiments, the predictions may be initiated from the network side, for example responsive to a signaling provided to the UE 400 from one of the network entities, like the gNB 406 or the further UE 410. In accordance with embodiments, the UE 400 can initiate the one or more second occasions responsive to one or more conditions, e.g., responsive to one or more of the following:

- A change of a link performance, e.g., a degradation of a link performance beyond a predefined limit.

The change of the link performance may be detected by a measurement or by another indicator, e.g., a number of Hybrid Acknowledge Request, HARQ, nonacknowledgements exceeding a configured or pre-configured threshold.

A measurement being above or below a configured or pre-configured threshold.

A higher layer event, like a change of a Quality of Service, QoS, a change of a data rate, a change of a delay demand.

A geographical change, like a change of an Al zone or an Al area.

An upcoming transmission by the UE or an upcoming reception at the UE.

In accordance with embodiments, the length of the prediction windows can be tied to the link quality performance experienced by the UE. For example, predictions can run longer if the channel quality is above a configured or pre-configured threshold. In other words, a length or duration of the second occasion, like a length of a prediction window, depends on one or more performance parameters, like a link quality, experienced by the UE 400, and the length or duration of the second occasion increases or decreases if the performance parameter is above or below a configured or pre-configured threshold.

In accordance with embodiments, responsive to a change of link performance detected by a measurement or other indicator such as HARQ, the UE may start one or more aperiodic or periodic predictions to react quicker to a changing environment.

In accordance with embodiments, the predictions are placed just before planned transmissions or grants, preferably so that a prediction result can be piggybacked onto the transmission, e.g., as a MAC-CE or UCI. In accordance with embodiments, the prediction result is piggybacked only if the prediction triggers a pre-configured condition, e.g., one or more of the following: an upcoming UL transmission, e.g. a scheduled or configured PUSCH in the reporting slot, an SPS or configured grant configuration, e.g. a periodic grant that can be used if the prediction needs to be reported. a QoS of the upcoming transmission, a size or type of the upcoming transmission, remaining bits available for piggybacking, a specific location of the UE, a time since the last measurement or prediction, a time since the last report is above a configured or preconfigured threshold.

In accordance with further embodiments, the UE may only report predictions in a reporting occasion, if a minimum time between a latest measurement and the reporting occasion is fulfilled. In other words, the UE may require a certain processing time to perform a prediction. If the time between a latest measurement and the reporting occasion is too small, the UE may not be able to compute a prediction. In such a scenario, the UE may skip the report, replace the predictions with measurements or default values for which the time constraint is not fulfilled, report only measurements, or indicate an error.

In accordance with further embodiments, the UE 400 may skip a measurement occasion, if a prediction exceeds a certain threshold, for example in case it is determined that the prediction indicates that the predicted one or more beams may still be used. On the other hand, the UE may perform a measurement if the prediction triggers a pre-configured condition. In this way, a prediction-capable UE, like the UE 400, may save power by skipping one or more measurements so that, as a result, the UE 400 does not transmit a measurement report in the uplink. This allows the UE 400 to save power from not transmitting in the uplink. Moreover, the UE 400 may perform a discontinuous reception, DRX, and go into a sleep mode and just decode synchronization signals and/or control signals during DRX. For example, the UE 400 may refrain from performing measurements, for example, beam-related measurement and perform a short and/or long DRX dependent on a certain criterion, so that the UE avoids unnecessary sensing and calculation of measured parameters. Stated differently, the UE 400 may skip a measurement of a beam at a certain first occasion and/or a reporting of the measured beam at the reporting occasion if one or more first conditions apply. On the other hand, the UE 400 may perform a measurement of a beam at a certain first occasion and/or a reporting of the measured beam at the reporting occasion if one or more second conditions apply. The first condition for skipping a measurement may be one or more of the following: - A prediction at a second occasion preceding the certain first occasion exceeds a configured or pre-configured threshold, e.g., the confidence of the prediction exceeds a certain threshold making a new measurement unnecessary or the current beam is still the best in the prediction or the predicted RSRPs have a delta exceeding a threshold.

- A currently served beam, e.g., a beam quality, has not changed.

- A battery status of the UE is below a configured or pre-configured threshold. Thereby, a power drainage is avoided, and the UE may rather tolerate a less optimal beam then a too high-power drainage.

- A change in a QoS requirement or in one or more higher layer criteria, like an upcoming high priority transmission. For example, the UE may have an important upcoming transmission and reduces the number of measurements prior to this transmission in order to save power so that the transmission may be safely performed.

- An indication from the network or higher layers.

The second condition may be one or more of the following: a confidence associated with the prediction is below a certain threshold, a time gap between the reporting occasion and the measurement is less than a certain threshold, a delta between one or more predicted values is too low, one or more deltas between the last measurement and the prediction exceed a threshold, a time since the last measurement report exceeds a certain threshold, an indication from the network or higher layers, a performance monitoring threshold associated to the prediction.

When entering into a sleep mode, like a Discontinuous Reception, DRX, mode, the UE may stay in the sleep mode until one or more of the following: until the next first occasion, the next second occasion, the next reporting occasion, the UE’s next uplink grant, a maximum timer is reached, e.g., based on a configured or preconfigure threshold, the UE receives a wake-up signal, WUS, e.g., a WUS transmitted by a gNB or by another UE or by another RAN or WiFi entity. Fig. 10 illustrates the skipping of a beam management reporting for power saving in accordance with embodiments. Three long DRX cycles #n, #n+1 and #n+2 are illustrated and each of which may include respective short DRX cycle occasions 450. The UE receives at the beginning of the long DRX cycle #n the PDCCH 452 triggering a DRX-inactivity timer 454. During the on-duration of the third short DRX cycle, the UE 400 performs a beam prediction using the AI/ML module 416, as is indicated at 456. In accordance with a configuration of the UE, a measurement of the beams is configured to take place at the third short DRX cycle, however, given the fact that the prediction happened just one short DRX cycle before the measurement occasion, the measurement is skipped, as is indicated at 458.

During the second long DRX cycle #n+1 , no predictions/measurements are configured, and the UE 400 is informed accordingly during the first short DRX cycle and skips all remaining short DRX cycle occasions. In the third long DRX cycle #n+2, a beam prediction 460 is performed during the first short DRX cycle. As a consequence, a measurement, which is configured to be performed at the same time, is skipped, as is indicated at 462 so that the UE 400 goes into the DRX sleep mode until the end of the long DRX cycle.

Fig. 11 illustrates an RRC measurement configuration in accordance with embodiments of the present invention which may define the one or more measurement windows, the one or more prediction windows, like a length and a position thereof. For example, function measurement windows may be defined for performing prediction in a way as it is described above with reference to Fig. 10. The measurement and/or prediction windows may be defined implicitly by the RRC configuration of the reference signal resources, and/or the report periodicity and/or the report offset. Also, trigger points, for example a maximum value or a confidence value calculated by the module, may be included in the configuration. For example, for aperiodic reporting, there may be trigger points indicated in the configuration. If a certain parameter for which a trigger point is configured, e. g. , a beam strength, a maximum beam strength, or a confidence, exceeds the configured trigger point, the UE would report a report for the reporting configuration. Otherwise, the UE would not report for that reporting configuration, or, alternatively, the UE would follow a periodic reporting. Fig. 11 illustrates at 466 that the CSI-report is allowed to include Al predictions, if the measurement configuration indicates this parameter to be enabled. Likewise, at 468, the configuration element indicates that only predicted CRIs or only predicted CRIs+RSRP may be included in the report. In another example, the allowReportAlpredictions 466 may indicate that only measurements, only predictions, or measurements and predictions are allowed to be reported in an associated report.

For example, the measurement and/or prediction windows and/or the trigger points may be indicated in the RRC configuration like the following:

• measurementwindow SEQUENCE { o Periodicity ENUMERATED { p1 , p2,...}, o Offset ENUMERATED {o1 ,o2,...}, o Length ENUMERATED {11 , I2, ...}},

• triggerPointConfidence ENUMERATED {t1 , t2, ...},

• triggerPointMaxRSRP ENUMERATED {r1 , r2,...}.

In accordance with further embodiments of the second aspect of the present invention, the UE 400 may report one or more beams in accordance with the report configuration for a measurement report which reports the one or more beams determined using the measurement module 414. In accordance with such embodiments, the UE 400 replaces one or more or all of the entries in the measurement report by respective prediction entries.

For example, the UE may indicate in the measurement report that the measurement report includes prediction entries, like a predicted CRI index or a predicted Layer 1 reference signal received power, L1-RSRP. If the measurement report includes prediction entries, the UE400 may perform one or more of the following: set a flag in the measurement report, activate a prediction indicator, add a confidence value indicating a confidence with the associated entries, wherein a first value indicates that an entry is a measured value, and a second value indicates that an entry is a predicted value, set one more bits of a multi-bit value, like a L1-RSR 7-bit value, so as to signal that prediction is used, and set the remaining bits for indicating one or more of the following: o a confidence in the AI/ML model, o an age of calculation performed by the AI/ML model, o an ID of the AI/ML model, o an ID of an the AI/ML functionality, o delta values, e.g., values indicating a delta with regard to a previous value. Further embodiments of the second aspect of the present invention provide a base station, like the gNB 406 in Fig. 4 that provides the UE 400 a configuration from which the UE determines whether measured or predicted results are to be reported. The BS 406 serves the UE 400 and configures the UE 400 to determine one or more beams, e.g., one or more strongest beams, from a plurality of beams, which are transmitted by a network entity, like the BS 406, and received at the UE, and send a report about the one or more beams, wherein the UE is to determine the one or more beams using a measurement of one or more reference signal resources, and/or at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality.

In accordance with embodiments, the BS 406 configures the UE with a configuration defining a reporting of the one or more beams to the network entity, the configuration indicating that the one or more beams determined using the measurement, and/or the one or more beams predicted using the at least one AI/ML model or functionality are to be reported.

In accordance with other embodiments, the BS 406 configures the UE to report the one or more beams at a configured or preconfigured reporting occasion, wherein, dependent on a temporal relationship between the reporting occasion and first and second occasions, the report includes the one or more beams determined using the measurement, or the one or more beams predicted using the at least one AI/ML model or functionality. At the first occasions the UE determines the one or more beams using a measurement of one or more reference signal resources, and at the second occasions the UE predicts the one or more beams using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality.

Third aspect

Embodiments of the first aspect of the present invention are now described. In accordance with embodiments of the third aspect of the present invention, enhancements for the AI/ML management and pre-processing of the input data for the AI/ML model or functionality at a UE are provided.

AI/ML model or functionality adaption

In accordance with embodiments, the UE 400, see Fig. 4, may use the AI/ML module 416 for performing one or more tasks associated with an operation of the UE. The AI/ML module 416 receives for the one or more AI/ML models or functionalities the input data 415 obtained from one or more measurements and provided to the AI/ML model by the measurement module 414. The measurement module 414 obtains the input data by performing measurements which are associated with a configured or preconfigured configuration. The UE 400 adapts the input data 415 to the AI/ML model or functionality responsive to a certain event.

In accordance with embodiments, the input data 415 may comprise a plurality of samples obtained by the measurement module 414, and the input data 415 may be adapted by fully or partially adapting the samples in the input data 415. Fully adapting the samples in the input data may comprise removing all samples from the input data 415 or resetting all samples in the input data to a configured or preconfigured value, for example, a null value or to zero. Partially adapting the samples in the input data 415 may include removing from the input data all samples obtained after the certain event, or resetting in the input data all samples received after the event to a configured or preconfigured value, like null or zero, or removing from the input data a proper subset of the samples obtained after the event and/or before the event, for example, for removing or replacing outdated samples.

Fig. 12 illustrates an embodiment of partially adapting samples in the input data of an AI/ML model. The AI/ML module 416 includes a certain AI/ML model or functionality 416a and a memory 416b holding the input data 416c for the AI/ML model or functionality 416a. Fig. 12 illustrates the AI/ML model 416 at different times, namely at a time ti and at a time t2 which is later the time ti and which is after a time tE at which certain event happened that triggers a reset 480. The input data 416c, which is stored in the memory 416b, is partially cleared, by setting the first three data elements from x1 , x2, x3 to zero and maintaining the last two data elements x4 and x5. It is assumed that the data elements x1 , x2 and x3 had values which were obtained after the time t_E at which the event, like a beam configuration change, happened so that the input data as provided by the measurement module 414 for values x1 , x2, x3 may no longer be valid as it was measured on the basis of a no longer valid configuration. On the other hand, the samples x4 and x5 are considered to be unchanged and, therefore, are maintained.

In a further embodiment, partially adapting samples in the input data of an AI/ML model or functionality can also involve removing data samples based on a function and/or replacing a set or a proper subset of samples by the one or more outputs of a function depending on the set of samples. This may be required as to remove anomalies from the set of data, e.g., peak values or values which are below a certain threshold, as to improve the results of an AI/ML calculation. Furthermore, this could reduce computational complexity, e.g., in terms of time required to compute the AI/ML function. The function described here can be configured or preconfigured by a network entity, e.g., by a gNB or by another UE.

The above described function can comprise one or more of a maximum value, e.g., based on a configured or preconfigure threshold, a minimum value, e.g., based on a configured or preconfigure threshold, arithmetic mean, geometric mean, weighted mean, moving average, a one-dimensional function, a multi-dimensional function a confidence interval, e.g., samples which are within a certain interval wrt. the average of the input data samples, or wrt. a configured or preconfigured average value.

Depending on an event, the UE may not be required to adapt the input data of the AI/ML model or functionality anymore. This could be due to an already “good” result or outcome of the AI/ML model or functionality, which may also be signaled to the UE by the network entity, e.g., gNB or by another UE..

Fig. 13 illustrates an embodiment for the handling of input samples by an AI/ML model. On the left, the input data 416c is shown as a sequence of inputs x1 to x5 which is provided to the AI/ML model 416c and suggesting a consideration of a history of the inputs during the processing. On the right of Fig. 13, it is illustrated that the AI/ML model 416c receives a reset 480 so that at least some of the previous inputs are not considered, as is illustrated by replacing x1 to x3 by zeros, thereby indicating that the AI/ML model starts processing with a fresh state from x4 and x5.

In accordance with embodiments, responsive to the event, the UE 400 performs one or more of the following: apply a default configuration, with which the UE is configured or preconfigured, apply a new configuration with which the UE is configured or preconfigured, receive from the wireless communication network, e.g., from a gNB or from the radio access network, RAN, or from another UE, a new configuration and apply the received new configuration, switch to a different AI/ML model or functionality corresponding to a new or default configuration.

In accordance with embodiments, the certain event comprises one or more of the following: a change of a configuration associated with the one or more measurements for obtaining the input data, e.g., a beam configuration, an indication from the wireless communication network, e.g., a signaling from one or more entities in wireless communication network, a performance degradation, a change in a channel, a radio link failure, RLF, a mode switch in a vehicular UE, e.g., a UE switches from mode 1 , under control of a gNB, to mode 2, direct communication with other UEs via PC5, or vice versa, a handover, a fulfillment of a conditional handover, CHO, condition, a change in Quasi co-location, QCL, a change of a used MIMO mode, e.g., a rank drops, e.g., in case an antenna is suddenly shielded, which is more relevant in higher frequency ranges, e.g., FR2 or FR3. an AI/ML model update, e.g., a reception of new training data and/or a reception of a pre-trained AI/ML model, a state change, e.g., a transition of the UE from an inactive state to a connected state, a reception of a wake-up signal, WUS, a change in location of the UE, initiating carrier aggregation, CA, or evacuating a given carrier, e.g., if the UE switches off CA or removes a carrier from a multiband configuration, a radio link recovery, a change in signal quality, e.g., the SNR/SINR/RSSI/RSRP is improving or degrading, a successful HO or CHO, a successful beam switch or change of a TRP, a static position of a UE, e.g., a UE stops moving, a performance or confidence of the Al/M L model or functionality is above or below a certain threshold, an indication of the AI/ML model or functionality. In accordance with embodiments, the UE determines the performance degradation or the change in a channel comprise a change of one or more measured parameters, like a Signal to Noise Ratio, SNR, a Signal-to-lnterference-and-Noise Ratio, SINR, or a Reference Signal Received Power, RSRP, or a Reference Signal Received Quality, RSRQ, or a Reference Signal Strength Indicator, RSSI, or a Channel Quality Indicator, CQI), or an interference level, or more Chanel State Information, CSI, parameters dropping below a configured or preconfigured threshold or dropping during a configured or preconfigured time period by more than a configured or preconfigured amount.

In accordance with embodiments, the UE determines the change in location of the UE by one or more of: a movement of the UE from an environment for which the beam configuration applies to a new environment for which a new beam configuration applies, e.g. moving between an urban environment and a rural environment, a movement of the UE into a certain region or zone, e.g., from outdoor to indoor, or into a certain distance from the base station, e.g., with respect to a minimum required communication range, a change in longitude and/or latitude and/or height beyond a configured or preconfigured threshold.

In accordance with embodiments, the UE determines the indication from the wireless communication network by one or more of: an indication from a base station included, e.g., in Downlink Control Information, DCI, a Medium Access Control Control Element, MAC CE, or a Radio Resource Control, RRC, signaling, or through broadcast/multicast messages to multiple UEs, a signaling of a new beam configuration, e.g., when the UE moves from an environment for which the beam configuration applies to a new environment for which the new beam configuration applies, e.g. moving between an urban environment and a rural environment, a signaling of a new configuration, e.g., when the UE moves from an environment for which a localization configuration applies to a new environment for which a new localization configuration applies, e.g. moving between an urban environment and a rural environment, a signaling of new assistance information replacing current assistance information, e.g., when the UE moves from an environment for which current assistance information applies to a new environment for which the new assistance information applies, e.g. moving between an urban environment and a rural environment, a signaling from another UE, e.g., via sidelink assistance information messages, e.g., AIM. The UE may be a wearable and receives from an associated device I phone (=other UE) the indication. Furthermore, this may also be signaled using other sidelink control signaling, e.g., SCI or PC5-RRC or MAC-CE via PSSCH.

In accordance with embodiments, the input data 415 is stored in one or more of the following: the memory 416b that is part of the AI/ML module 416 implementing the AI/ML model or functionality 416a, a cloud server, another device, e.g., an associated UE, a base station, e.g., gNB or WiFi AP, a mobile edge cloud, MEC, located closely to a serving base station BS, a dedicated RAN entity, e.g., an AI/ML network function, NF.

Thus, embodiments of the third aspect of the present invention provide for a reset of an AI/ML model or functionality. For example, in the scenario of temporal domain beam prediction, a total number of M beams/pairs taken at different time instances are necessary for respective measurements in a baseline scheme. The baseline scheme may include taking all samples of the respective measurements into account as input signals for the AI/ML model or functionality. This may be because the UE is not aware of a change or simply follows the initially configured or preconfigured procedure. If there is a change in the beam configuration, it may be essential to update the samples taken for the prediction, like the input data 415 described above, to prevent a performance degradation. For optimizing the AI/ML model or functionality performance, the above described actions may be considered, namely resetting the history, for example, clearing or removing all samples in the measurement baseline scheme, or a partial clearing, for example, removing all samples from a database or memory after the time instance at which a beam configuration changed. By these actions, in accordance with the inventive approach, the AI/ML model or functionality 416a of the UE 400 is ensured to operate with up-to-date information, thereby maintaining and enhancing performance as a dynamic context of changing beam configurations. In accordance with embodiments, the reset of the history or the partial clearing means either altering the memory 416b which is part of the AI/ML module 416, for example, by resetting or nulling elements corresponding to the deleted measurements, or by altering the input into the AI/ML model. For example, if the input data 416c comprises a number of past measurement results, the UE may reset or null certain or all past measurement results that are to be input into the model.

In accordance with embodiments, a beam is identified by one or more of the following: a beam ID or by a resource ID, a Chanel State Information Reference Signal, CSI-RS, resource indicator, CRI, a beam ID. The Chanel State Information, CSI, may be one of more of the following: a Channel Quality Indicator, CQI, a precoding matrix indicator, PM I, a Chanel State Information Reference Signal, CSI-RS, resource indicator, CRI, a sounding reference signal, SRS, a CSI Reference Signal, CSI-RS, resource indicator, CRI,

Synchronization Signal/Physical Broadcast Channel, SS/PBCH, Block Resource Indicator, SSBRI, a SSB index, e.g., mapped to a certain beam, a layer indicator, LI, a spatial signature, e.g., the direction or direction of arrival or direction of departure of a signal, a rank indicator, Rl, a Layer 1 reference signal received power, L1-RSRP, a Layer 1 Signal to Interference plus Noise Ratio, L1-SINR, a Capability Index, one or more time-domain channel properties, TDCP, a received signal strength indicator, RSSI,, higher layer CSI, e.g., a Layer 3 reference signal received power, L3-RSRP, or a Layer 3 Signal to Interference plus Noise Ratio, L3-SINR training fields, e.g., as used within the Preamble in WiFi systems.

AI/ML training and model exchange

Further embodiments of the third aspect of the present invention concern procedures for AI/ML training and model exchange. In accordance with embodiments, a UE may obtain respective beams for a communication from an AI/ML model operating at the UE or at the network side, for example, at the gNB or base station. The AI/ML model may operate according to a gNB antenna configuration using as input respective beam measurements performed by the UE. The UE may be a UE 400 as depicted in Fig. 4 including the measurement module 414 and the AI/ML model 416 and, optionally, the reporting module 418. The U E 400 performs, using the measurement module 414, one or more measurements of reference signal resources associated with respective beams, which are transmitted by a network entity, like the gNB 406 or the UE 410 which are received at the UE. For example, the respective beams comprise, in accordance with embodiments, the DL Tx beams provided by the gNB 406 in a way as described above with reference to Fig. 5 or Fig. 6. The actual beams to be used by the UE 400 for a communication, for example, with the gNB 406, are determined using the at least one AI/ML model or functionality.

The UE 400 illustrated in Fig. 4 uses the at least one AI/ML model or functionality which operates in accordance with the antenna configuration of the gNB for providing the respective DL Tx beams to the UE 400. The AI/ML module 416 receives as an input 415 the measurements made by the UE’s measurement module 414. Thus, according to this embodiment, the UE 400 runs or operates the at least one AI/ML model or functionality by the AI/ML module 416 and receives from the gNB 406 the antenna configuration for the respective DL Tx beams. The one or more beams to be used by the UE 400 for a communication are determined by the AI/ML model or functionality using the received antenna configuration and the one or more measurements 415 provided by the measurement module 415.

In accordance with other embodiments, the UE 400 may not include the AI/ML module 416 or the AI/ML module 416 may not be used for determining the one or more beams. In such embodiments, the one or more beams are obtained by the UE from an AI/ML model or functionality which is located in the network entity that provides the respective beams from which the one or more beams for the communication are obtained. For example, the gNB 406 may include an AI/ML module which is operated according to the antenna configuration of the gNB 406 used for providing to the UE 400 the respective beams, like the DL Tx beams illustrated in Fig. 5 or Fig. 6. The AI/ML model or functionality at the gNB 406 receives as an input the measurements performed by the UE 400, i.e., other than depicted in Fig. 4, in such an embodiment, the UE 400 performs the respective measurements using the measurement module 414 on the one or more reference signal resources associated with the respective beams provided by the base station and transmits the input data 415 via the antenna and the llu interface 408 to the gNB 406 as input into the AI/ML model or functionality implemented at the gNB 406. The gNB based AI/ML model obtains the one or more beams for the UE 400 and signals them via the llu interface 408 to the UE 400. Thus, in accordance with embodiments, the gNB 406 may run the AI/ML model based on beam measurement reports received from the UE 400. The AI/ML model determines the beams to be used at the UE which are signaled by the gNB 406 to the UE 400.

In accordance with other embodiments, the AI/ML model may run on the UE 400, for example, in the AI/ML module 416. The gNB 406 transmits some or all of the DL Tx beams, as illustrated in Fig. 5 or Fig. 6, i.e., all DL Tx beams or a proper subset of the DL Tx beams which may be created using the antenna configuration at the gNB 406. Further, the UE receives the antenna configuration at the gNB, and the AI/ML module 416 operates in accordance with the received antenna configuration of the gNB using as an input the measurements 415 provided by the measurement module 414. In accordance with embodiments, the gNB 406 may transmit different subsets of beams, like different subsets of DL Tx beams over time, so that the UE 400 may reconstruct a radio channel using different input values, which may be considered to be similar to a sparse sensing. The advantage of this embodiment is that the necessary signaling to the UE is reduced. Thus, according to this embodiment, the UE 400 runs or operates the at least one AI/ML model or functionality by the AI/ML module 416 and receive from the gNB 406 the antenna configuration for the respective DL Tx beams. The one or more beams to be used by the UE 400 for a communication are determined by the AI/ML model or functionality using the received antenna configuration and the one or more measurements 415 provided by the measurement module 415.

In accordance with yet other embodiments, the UE 400 may receive from the gNB 406 a pre-trained AI/ML model which has been trained at the gNB 406 on the basis of the gNBs antenna configuration. The pre-trained AI/ML model, which may be run by the AI/ML model 416 of the UE 400 receives as input415 the measurements performed by the UE 400. Thus, according to this embodiment, the UE 400 receive from the gNB 406 the AI/ML model and/or parameters of the AI/ML model or functionality, e.g., a model trained at the gNB 406 using the antenna configuration for the respective DL Tx beams, and determines the one or more beams to be used by the UE for the communication by the AI/ML model or functionality using the one or more measurements. For example, the UE 406 may use the AI/ML model or functionality to estimate or predict a beam for a communication with the network entity, e.g., by interpolating between beams by using the reference signal resources of which are sparsely measured to predict a different beam, e.g., a more optimal beam.

In accordance with further embodiments, when the gNB executes the AI/ML model, the gNB 406 may dynamically adjust beam forming parameters based on real-time beam management reports not only received from UE 400 but also from one or more other UEs, like UE 410. The additional beam management reports may also be received from the network, for example from a network function or from a gNB, for example a target gNB in case of a handover, HO, or a potential conditional handover candidate. Using the additional beam management reports allows the AI/ML model running at the gNB to analyze signal quality indicators and locations of other UEs, for optimizing a beam direction and a beam strength for a communication with the UE 400 thereby enhancing network efficiency and reducing interference.

Stated differently, the AI/ML model or functionality analyzes one or more signal quality indicators and/or locations of the UE and the one or more further UEs to optimize a beam direction and/or a beam strength of the one or more beams to be used by the UE for the communication with the network entity. The one or more signal quality indicators may be one or more of the following: a pathloss or signal quality, e.g., using an uplink received signal strength indicator, RSSI, measured at the base station, a signal quality based on reference signals transmitted in the uplink by the UE, e.g., o based on sounding reference signals, SRS, o demodulation reference signals, DM-RS, o phase-tracking reference signals, PT-RS, a reciprocity-based signal quality, e.g., in case of a Time Division Duplex, TDD, system, based on CSI feedback provided in an uplink, feedback information sent by the UE, e.g., HARQ-ACKs or NACKs, or CBG-based ACKs or NACKs, provided by the UE, higher-layer statistics or feedback, e.g., packet delay or measured TCP slow-starts, or L3-RSRP or L3-latency.

UE AI/ML management and inference

In accordance with yet further embodiments of the third aspect of the present invention, enhancements of UE AI/ML management and inference are provided. A UE, like UE 400, may be provided with a functionality configuration referring, for example, to AI/ML-enabled features or an AI/ML-enabled feature group, FG. The feature or feature group are enabled by one or more configurations, and the one or more configurations are supported based on conditions indicated by the UE’s capability. This means that the UE indicates what it supports. For example, when considering the beam management use case, the UE may indicate a first set of beams, set A, and a second set of beams, set B, that the UE supports. The beams in set A and set B may be beams of a gNB where the UE trained the AI/ML model. Therefore, it may be required to signal details on the particular beams which need to be obtained from the gNB where the UE trained the AI/ML model. This information may also be preconfigured or configured together with the AI/ML model or functionality at the UE, i.e., meta parameters. This goes together with additional signaling overhead. Also, the gNB vendor may consider such information confidential and may not want to share detailed information about the beams provided by the gNB.

Embodiments of the third aspect of the present invention address this issue by providing efficient reporting techniques that require limited understanding of the actual beam properties at the UE. In accordance with embodiments, a UE, like UE 400 in Fig. 4, supports at least one AI/ML model or functionality. The AI/ML model or functionality uses measurements of one or more first beams from a first set of beams for determining one or more second beams from a second set of beams. The AI/ML model run by the AI/ML module 416 may have been trained by a base station or gNB different from the one currently serving the UE 400 so that a certain configuration, for example a beam configuration received from the gNB 406 currently serving the UE 400 may be different from a configuration used when training the AI/ML model at a different gNB.

In accordance with embodiments, the UE 400 determines whether the AI/ML model or functionality is applicable or compatible based on a configuration received from the network, for example from the gNB 406. For example, the UE 400 may determine a configuration of the gNB 406 on the basis of gNB specific information, like its ID or type or vendor, or based on a signaling from the gNB 406 indicating, for example, a certain beam configuration, which the UE may be provided with, e.g., responsive to specific information forwarded by the UE 400 to the gNB 406 and describing the configuration currently supported by the UE given the AI/ML being trained by a different gNB. In accordance with embodiments, the UE 400 includes the measurement module 414 for measuring one or more first beams from the first set of beams which has been received from the network entity, like a gNB. The UE 400 uses the AI/ML model or functionality to determine one or more second beams from a second set of beams from the network entity.. The AI/ML model or functionality uses the measured one or more first beams as input data 415. The UE 400 receives a signaling allowing to determine whether the AI/ML model or functionality, which has been trained by a gNB different from gNB 406 in Fig. 4, is also working for gNB 406 currently serving the UE 400. The network or gNB specific information may be one or more of the following: an identification of the network entity, e.g., a vendor ID, or a configuration ID, or a gNB ID, or a UE ID, or a target gNB ID in case of a handover, HO, or a potential conditional handover, CHO, a cell ID or a Physical Cell ID, PCI of the cell comprising the network entity, an antenna configuration of the network entity, an antenna array ID at the network entity, e.g., an ID of a transmission reception point, TRP, at the network entity comprising a multi-TRP, a beam configuration of the network entity,

In accordance with other embodiments, alternatively or in addition to the above embodiment, the UE 400 indicates to the gNB 406 information about the AI/ML model or functionality run by its AI/ML module 416. Thus, the UE may simply indicate information about its AI/ML-model or functionality, like specific information on the beams used and the like, so as to receive, responsive to this information from the gNB a corresponding configuration, like a suitable beam configuration on the basis of which the AI/ML model, although being trained by a different gNB, may be used also for the new gNB. Stated differently, the UE may be configured by the gNB 406 with respective beams, like beams for set A and beams for set B which are compatible with the sets of beams associated with the AI/ML model currently run by the AI/ML module 416 of the UE 400. Here, compatible implies that the set of beams, e.g., set A and/or set B from a gNB and a new gNB are identical or that one of the sets is a proper subset of the other set of beams. In this case, and since the gNBs are not located at the exact same position in space, identical set or subset or proper subset of beams are determined based on an identity relation. The identity relation defines whether two beams B1 and B2 are considered to be identical. The identity relation may be defined as one or a combination of the following: the beams have the same ID, the beams are spatially aligned, e.g., relative to each other, similar geometric properties, the beams have the same QCL, the beams are QCLed of at least a certain QCL Type by configuration, the beams have the same orientation in azimuth and/or elevation, relative to each other, the respective sets to which the beams belong to have shared properties, e.g., same size, same orientation, the beams are originating from the same antenna configuration, e.g., same number of antenna elements, e.g., as in a Massive MIMO array, the beams are coming from the same PLMN, the beams are coming from a same gNB configuration, e.g., wrt. gNB sectorization or TRP configuration, e.g., multi-TRP, or antenna configuration, e.g., array antenna with the same number of antenna elements, the beams are coming from the same gNB or the same gNB type or same vendor or cell or TRP or cell group.

The combination of more than one identity relation may comprise an "and", an "or", or an "xor" operation, where each combination may use the same or different operation from the previous combination. Furthermore, each combination may additionally use a "not" operation on the previous or on the combined relation.

In accordance with the above embodiments, the specific information forwarded by the UE 400 to the gNB 406 and describing the configuration currently supported by the UE may be one or more of the following:

• beam information about the set A and set B,

• an identification of the network entity with which the at least one AI/ML model or functionality is associated, e.g., a vendor ID, or a configuration ID, or a gNB ID, or a UE ID, or a target gNB ID in case of a handover, HO, or a potential conditional handover, CHO,

• a cell ID or a Physical Cell ID, PCI of the cell comprising network entity network entity with which the at least one AI/ML model or functionality is associated,

• an antenna configuration of network entity network entity with which the at least one AI/ML model or functionality is associated,

• an antenna array ID at network entity network entity with which the at least one AI/ML model or functionality is associated, e.g., an ID of a transmission reception point, TRP,

• a beam configuration of network entity network entity with which the at least one AI/ML model or functionality is associated. In accordance with embodiments, the beam information may be one or more of the following: one or more beam IDs of beams that belong to set A, one or more beam IDs of beams that belong to set B, a size of set A, e.g., a number of beams included in set A, a size of set B, e.g., a number of beams included in set B, a number of beams per dimension, like the number of beams per azimuth, or per elevation, angular differences between the beams of set A and/or set B, e.g., a phase offset between the beams of pi/4 resulting in 8 beams per dimension, amplitude differences between the beams of set A and/or set B, e.g., a , e.g., a power difference between the beams of the set A and/or set B, information about the first network entity, like a cell ID, a gNB ID, a vendor ID, or a gNB Type, from which set A and set B have been obtained for training the AI/ML model or functionality.

In accordance with embodiments, the UE 400 indicates a set of beam IDs, for example in the form of CSI resource IDs, which are contained in set A, and set of beam IDs, for example also in the form of CSI resource IDs, which are contained in set B. The UE 400 reports a source identifier, for example, a cell ID, where the UE trained the AI/ML model currently run by the AI/ML module 416. The current or serving gNB or cell, like gNB 416 in Fig. 4, uses these IDs together with the UE ID to obtain from the cell at which the AI/ML model has been trained, information about the actual beam properties. Using this information, gNB 406 configures the UE 400 with a CSI configuration that matches the configuration of the gNB at which the AI/ML model has been trained. This allows the AI/ML model to output correct predictions without knowing any details about the actual beam properties.

In accordance with other embodiments, the UE may report the one or more angular offsets, for example one per dimension, a number of beams per dimension and a set of beams belonging to set A and a set of beams belonging to set B. The gNB 406, using the one or more angular offsets, which describe the angular directions between the individual beams which are ordered according to a certain mapping, like a grid, calculates the directions of the individual beams and, using the results, configures the UE 400 with a matching CSI configuration, i.e. , a CSI configuration matching the AI/ML model currently used by the UE 400. In accordance with a further embodiment the beams may be arranged on a grid. The grid may be a two dimensional grid indicating the spatial relation of the beams to one another. The grid may be spanned over azimuth and/or elevation angles.

In accordance with yet other embodiments, instead of set A the UE 400 may simply signal a size of the set A and the numbering of beams of set A may be derived implicitly. For example, starting at 0 and ending at N-1 , where N is the size of the set A.

In accordance with embodiments, set B includes more beams than set A. Stated differently, set A is a proper subset of set B.

Beam subsets for measurement, reporting, performance evaluations

In accordance with embodiments, the UE 400 may be configured by the gNB 406 with a proper subset of the beams from set A and from set B, and the AI/ML model or functionality estimates or interpolates one or more characteristics of those beams which are not included in the received subsets.

Fig. 14 illustrates examples of beam subsets and their spatial relation to one another. Those beams not included in any of the subsets are illustrated as a white squares while the other squares are associated with one of the subsets A to D, labeled as number 1 , 2, 3 and 4. This embodiment is advantageous as it reduces signaling overhead as it is not necessary to signal all of the beams, rather, the non-transmitted beams, i.e., the beams not included in the subset, may be interpolated, for example, by the AI/ML model, which may also be used to estimate a quality of those beams which are not transmitted.

In accordance with further embodiments, the gNB 406 may transmit only selected subsets of beams, for example, subset A and subset C (see Fig. 14) while a spatial correlation between the subset allows the AI/ML model to interpolate the beam quality of nontransmitted subsets, such as subset B and subset D. This selective transmission further reduces the resource virtualization and computational load both at the gNB and the UE.

In accordance with yet other embodiments, the UE 400 may receive subset A and subset B (see Fig. 14) and use these subsets for an initial channel quality measurement which is reported back to the gNB. Using this report, an AI/ML model operating at the gNB may predict the channel quality of subsets B and C and adjust a beamforming strategy accordingly, thereby optimizing the network performance with a reduced or minimal signaling.

In accordance with other embodiments, the gNB may use an AI/ML model for evaluating a performance of various beam subsets by transmitting the limited number of beams periodically. For example, subsets A and B may be transmitted during one interval and their performance may be evaluated to estimate the characteristics of subsets C and D for a subsequent time interval, thereby dynamically adjusting the beamforming strategy in realtime.

In accordance with yet further embodiments, each of the beams from set A and set B may be a wide-coverage beam, like an SSB beam, having a coverage area comprising a coverage area of a plurality of narrow-coverage beams. Fig. 15 illustrates a grid overlaid by two sets of larger or wide-coverage beams, labeled as number 1 and 2. As is indicated, the beams of the different sets are arranged in an alternating pattern across the grid for covering the grid space with a fewer number of wide-coverage beams, thereby simplifying the beam measurements by reducing the number of beams to be measured. The AI/ML model, which is operated at the gNB or at the UE, may use the information from the wider beam to predict narrower transmit beams, for example, CSI-RS beams.

In accordance with further embodiments, the UE 400 may be configured with a CSI-RS- ResourceConfig or a CSI-RS resource set that is representing the set A of beams, and with another CSI-RS-ResourceConfig or CSI-RS resource set that is representing the set B of beams. The CSI-RS resources that are contained in the CSI-RS-ResourceConfig or in the CSI-RS resource set, which describe the set B beams, may only be CSI-RS resources that are also included in the CSI-RS-ResourceConfig or CSI-RS resource set that that describe the set A beams, i.e. , set B is always a subset of set A. Further, the UE 400 may implicitly derive a correspondence between the beams. In particular, the UE 400 knows, due to the fact that the same CSI-RS resource is part of set A and set B, that the sets represent the same beam with the same physical properties.

Dynamic activation and/or switching of an AI/ML model or functionality

Further embodiments of the third aspect of the present invention provide for a dynamic activation and/or switching of an AI/ML model or functionality. A UE, like UE 400 illustrated in Fig. 4, operates the AI/ML module 416 for running one or more AI/ML models or functionalities for performing a certain task. The AI/ML module 416 may hold a plurality of AI/ML models or functionalities for performing a certain task, and the UE 400 maintains one or more of currently not used AI/ML models or functionalities, which may perform the same task as the AI/ML model being currently operated, in a ready-to-use state. In the ready-to- use state the AI/ML model or functionality is fully operative, i.e. , is fully trained and also receives the input data 415, however, it is not active or used for actually performing the task, e.g., an output provided by the AI/ML model, like a prediction, may generated but not used by the UE. Any ready-to AI/ML model may be simply activated and operated by sending a short activation signal, i.e., no further operations are needed for starting the AI/ML model to use. For example, a signal may be provided allowing the output provided by the AI/ML model to be used by the UE.

The UE 400 may decide to put one or more of currently unused AI/ML models into the ready- to-use state, e.g. responsive to a certain signaling like an RRC signaling, a MAC CE, a higher layer control signaling or a signaling via a sidelink, for example, a PC5 RRC signaling or a PSCCH signaling or a control signaling embedded into a PSSCH.

In accordance with further embodiments of the third aspect of the present invention, the UE 400 may also implement any one of the above-described embodiments of the first and second aspects, using, for example, the reporting module 418 which, in accordance with embodiments of the third aspect, may also be included in the UE 400.

Network Entities

Further embodiments of the third aspect of the present invention provide a base station, like the gNB 406 in Fig. 4 that that updates input data (samples) of an AI/ML model responsive to a certain event. The gNB 406 serves the UE 400 transmits respective beams, like DL Tx beams to the UE 400. The gNB 406 receives from the UE 400 one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE 400. The gNB 406 determines from the respective beams one or more beams for a communication with the UE using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality which operates on the basis of input data including the measurements received from the UE 400. The gNB 406 adapts the input data of the AI/ML model or functionality responsive to a change of the beam configuration.

Further embodiments of the third aspect of the present invention provide a base station, like the gNB 406 in Fig. 4. The gNB 406 serves the UE 400 transmits respective beams, like DL Tx beams to the UE 400. The gNB 406 receives from the UE 400 one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE 400. The one or more beams, which are to be used by the BS for a communication with the UE, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, and the one or more beams to be used for the communication are to be obtained from at least one AI/ML model or functionality, which is operated at the gNB 406 according to an antenna configuration of the gNB 406 for the respective beams, using the one or more measurements provided by the UE, or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the gNB 406 for the respective beams, using the one or more measurements provided by the UE.

Further embodiments of the third aspect of the present invention provide a base station, like the gNB 406 in Fig. 4. The gNB 406 serves the UE400 which measures one or more first beams from a first set of beams supported by the UE, like set B above, and determining one or more second beams from a second set of beams, like set A above, supported by a first network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data. The gNB 406 transmits to the UE 400 a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for a gNB 406. In accordance with embodiments, the signaling includes gNB specific information allowing the UE 400 to identify the gNB 406 or describing beam specifics of the gNB 406. In accordance with other embodiments, the gNB 406 receives from the UE 400 the beam information about the first and second sets of beams, and the signaling includes a notification to the UE 400 whether the AI/ML model or functionality at the UE 400 is also working for the gNB 406, the notification generated by the gNB 406 using the beam information from the UE 400.

Further embodiments of the third aspect of the present invention provide a base station, like the gNB 406 in Fig. 4. The gNB 406 serves the UE 500 which is configured or preconfigured with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionalities, for performing a certain task. The gNB 406 signals to the UE 400 to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to-use state for allowing the UE 400 to activate one or more of the non-used AI/ML models or functionalities for performing the certain task. General

Embodiments of the present invention have been described in detail above, and the respective embodiments and aspects may be implemented individually or two or more of the embodiments or aspects may be implemented in combination.

In accordance with embodiments, the wireless communication system may include a terrestrial network, or a non-terrestrial network, or networks or segments of networks using as a receiver an airborne vehicle or a space-borne vehicle, or a combination thereof. Further, the wireless communication system may by a system or network different from the above described 4G or 5G mobile communication systems, rather, embodiments of the inventive approach may also be implemented in any other wireless communication network, e.g., in a private network, such as an Intranet or any other type of campus networks, or in a WiFi communication system.

In accordance with embodiments of the present invention, a user device comprises one or more of the following: a power-limited UE, or a hand-held UE, like a UE used by a pedestrian, and referred to as a Vulnerable Road User, VRU, or a Pedestrian UE, P-UE, or an on-body or hand-held UE used by public safety personnel and first responders, and referred to as Public safety UE, PS-UE, or an loT UE, e.g., a sensor, an actuator or a UE provided in a campus network to carry out repetitive tasks and requiring input from a gateway node at periodic intervals, a mobile terminal, or a stationary terminal, or a cellular loT-UE, or a vehicular UE, or a vehicular group leader (GL) UE, or a sidelink relay, or an loT or narrowband loT, NB-loT, device, or wearable device, like a smartwatch, or a fitness tracker, or smart glasses, or a ground based vehicle, or an aerial vehicle, or a drone, or a moving base station, or road side unit (RSU), or a building, or any other item or device provided with network connectivity enabling the item/device to communicate using the wireless communication network, e.g., a sensor or actuator, or any other item or device provided with network connectivity enabling the item/device to communicate using a sidelink the wireless communication network, e.g., a sensor or actuator, or a Wi-Fi device, like a station (STA), access point (AP), node or mesh node, or mesh point, or Mesh AP, or any sidelink capable network entity.

In accordance with embodiments of the present invention, a network entity comprises one or more of the following: a macro cell base station, or a small cell base station, or a central unit of a base station, an integrated access and backhaul, IAB, node, or a distributed unit of a base station, or a road side unit (RSU), or a Wi-Fi device such as an access point (AP) or mesh node (Mesh AP), or a remote radio head, or an AMF, or a MME, or a SMF, or a core network entity, or mobile edge computing (MEC) entity, or a network slice as in the NR or 5G core context, or any transmission/reception point, TRP, enabling an item or a device to communicate using the wireless communication network, the item or device being provided with network connectivity to communicate using the wireless communication network.

Although some aspects of the described concept have been described in the context of an apparatus, it is clear, that these aspects also represent a description of the corresponding method, where a block or a device corresponds to a method step or a feature of a method step. Analogously, aspects described in the context of a method step also represent a description of a corresponding block or item or feature of a corresponding apparatus.

Various elements and features of the present invention may be implemented in hardware using analog and/or digital circuits, in software, through the execution of instructions by one or more general purpose or special-purpose processors, or as a combination of hardware and software. For example, embodiments of the present invention may be implemented in the environment of a computer system or another processing system. Fig. 8 illustrates an example of a computer system 600. The units or modules as well as the steps of the methods performed by these units may execute on one or more computer systems 600. The computer system 600 includes one or more processors 602, like a special purpose or a general-purpose digital signal processor. The processor 602 is connected to a communication infrastructure 604, like a bus or a network. The computer system 600 includes a main memory 606, e.g., a random-access memory, RAM, and a secondary memory 608, e.g., a hard disk drive and/or a removable storage drive. The secondary memory 608 may allow computer programs or other instructions to be loaded into the computer system 600. The computer system 600 may further include a communications interface 610 to allow software and data to be transferred between computer system 600 and external devices. The communication may be in the from electronic, electromagnetic, optical, or other signals capable of being handled by a communications interface. The communication may use a wire or a cable, fiber optics, a phone line, a cellular phone link, an RF link and other communications channels 612.

The terms “computer program medium” and “computer readable medium” are used to generally refer to tangible storage media such as removable storage units or a hard disk installed in a hard disk drive. These computer program products are means for providing software to the computer system 600. The computer programs, also referred to as computer control logic, are stored in main memory 606 and/or secondary memory 608. Computer programs may also be received via the communications interface 610. The computer program, when executed, enables the computer system 600 to implement the present invention. In particular, the computer program, when executed, enables processor 602 to implement the processes of the present invention, such as any of the methods described herein. Accordingly, such a computer program may represent a controller of the computer system 600. Where the disclosure is implemented using software, the software may be stored in a computer program product and loaded into computer system 600 using a removable storage drive, an interface, like communications interface 610.

The implementation in hardware or in software may be performed using a digital storage medium, for example cloud storage, a floppy disk, a DVD, a Blue-Ray, a CD, a ROM, a PROM, an EPROM, an EEPROM or a FLASH memory, having electronically readable control signals stored thereon, which cooperate or are capable of cooperating with a programmable computer system such that the respective method is performed. Therefore, the digital storage medium may be computer readable.

Some embodiments according to the invention comprise a data carrier having electronically readable control signals, which are capable of cooperating with a programmable computer system, such that one of the methods described herein is performed.

Generally, embodiments of the present invention may be implemented as a computer program product with a program code, the program code being operative for performing one of the methods when the computer program product runs on a computer. The program code may for example be stored on a machine readable carrier.

Other embodiments comprise the computer program for performing one of the methods described herein, stored on a machine readable carrier. In other words, an embodiment of the inventive method is, therefore, a computer program having a program code for performing one of the methods described herein, when the computer program runs on a computer.

A further embodiment of the inventive methods is, therefore, a data carrier or a digital storage medium, or a computer-readable medium comprising, recorded thereon, the computer program for performing one of the methods described herein. A further embodiment of the inventive method is, therefore, a data stream or a sequence of signals representing the computer program for performing one of the methods described herein. The data stream or the sequence of signals may for example be configured to be transferred via a data communication connection, for example via the Internet. A further embodiment comprises a processing means, for example a computer, or a programmable logic device, configured to or adapted to perform one of the methods described herein. A further embodiment comprises a computer having installed thereon the computer program for performing one of the methods described herein.

In some embodiments, a programmable logic device, for example a field programmable gate array, may be used to perform some or all of the functionalities of the methods described herein. In some embodiments, a field programmable gate array may cooperate with a microprocessor in order to perform one of the methods described herein. Generally, the methods are preferably performed by any hardware apparatus.

The above-described embodiments are merely illustrative for the principles of the present invention. It is understood that modifications and variations of the arrangements and the details described herein are apparent to others skilled in the art. It is the intent, therefore, to be limited only by the scope of the impending patent claims and not by the specific details presented by way of description and explanation of the embodiments herein.

Claims

1. A user device, UE, for a wireless communication network, wherein the UE is to use at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality for performing one or more tasks associated with an operation of the UE, wherein the at least one AI/ML model or functionality operates on the basis of input data obtained from one or more measurements, and wherein the one or more measurements for obtaining the input data are associated with a configured or preconfigured configuration, and wherein the UE is to adapt the input data of the AI/ML model or functionality responsive to a certain event.

2. The user device, UE, of claim 1 , wherein the input data comprises a plurality of samples obtained from the measurements and adapting the input data of the AI/ML model or functionality comprises fully or partially adapting the samples in the input data.

3. The user device, UE, of claim 2, wherein fully adapting the samples in the input data comprises removing from the input data all samples, or resetting in the input data all samples to a configured or preconfigured value, like null.

4. The user device, UE, of claim 2, wherein partially adapting the samples in the input data comprises removing from the input data all samples obtained after the event, or resetting in the input data all samples obtained after the event to a configured or preconfigured value, like null, or removing from the input data a subset of samples obtained after the event and/or obtained before the event, e.g., remove or replace outdated samples.

5. The user device, UE, of claim 2 or 4, wherein partially adapting the samples in the input data comprises removing or not removing from the input data samples based on a function, or replacing a set of samples by the output of a function depending on the set of samples.

6. The user device, UE of claim 5., wherein the function is one or more of a maximum value, e.g., based on a configured or preconfigure threshold, a minimum value, e.g., based on a configured or preconfigure threshold, arithmetic mean, geometric mean, weighted mean, moving average, a multi-dimensional function a one-dimensional function, a confidence interval, e.g., samples which are within a certain interval wrt. the average of the input data samples, or wrt. a configured or preconfigured average value.

7. The user device, UE, of claim 5 or 6, wherein the function is configured or preconfigured by a network entity, e.g., by a gNB or another UE.

8. The user device, UE, of any one of the preceding claims, wherein, responsive to the event, the UE is to perform one or more of the following: apply a default configuration, with which the UE is configured or preconfigured, apply a new configuration with which the UE is configured or preconfigured, receive from the wireless communication network, e.g., from a gNB or from the radio access network, RAN, or from another UE, a new configuration and apply the received new configuration, switch to a different AI/ML model or functionality corresponding to a new or default configuration.

9. The user device, UE, of any one of the preceding claims, wherein the certain event comprises one or more of the following: a change of a configuration associated with the one or more measurements for obtaining the input data, e.g., a beam configuration, an indication from the wireless communication network, e.g., a signaling from one or more entities in wireless communication network, a performance degradation, a change in a channel, a radio link failure, RLF, a mode switch in a vehicular UE, e.g., a UE switches from mode 1 , under control of a gNB, to mode 2, direct communication with other UEs via PC5, or vice versa, a handover, a fulfillment of a conditional handover, CHO, condition, a change in Quasi co-location, QCL, a change of a used MIMO mode, e.g., a rank drops, e.g., in case an antenna is suddenly shielded, which is more relevant in higher frequency ranges, e.g., FR2 or FR3. an AI/ML model update, e.g., a reception of new training data and/or a reception of a pre-trained AI/ML model, a state change, e.g., a transition of the UE from an inactive state to a connected state, a reception of a wake-up signal, WUS, a change in location of the UE, initiating carrier aggregation, CA, or evacuating a given carrier, e.g., if the UE switches off CA or removes a carrier from a multiband configuration, a radio link recovery, a change in signal quality, e.g., the SNR/SINR/RSSI/RSRP is improving or degrading, a successful HO or CHO, a successful beam switch or change of a TRP, a static position of a UE, e.g., a UE stops moving, a performance or confidence of the Al/M L model or functionality is above or below a certain threshold, an indication of the Al/M L model or functionality.

10. The user device, UE, of claim 9, wherein the UE is no longer to adapt the input data of the AI/ML model or functionality according to the certain event.

11. The user device, UE, of any one of the preceding claims, wherein the UE is no longer to adapt the input data of the AI/ML model or functionality according to another certain event.

12. The user device, UE, of any one of claims 9 to 11 , wherein the performance degradation or the change in a channel comprise a change of one or more measured parameters, like a Signal to Noise Ratio, SNR, a Signal-to- Interference-and-Noise Ratio, SINR, or a Reference Signal Received Power, RSRP, or a Reference Signal Received Quality, RSRQ, or a Reference Signal Strength Indicator, RSSI, or a Channel Quality Indicator, CQI), or an interference level, or more Chanel State Information, CSI, parameters dropping below a configured or preconfigured threshold or dropping during a configured or preconfigured time period by more than a configured or preconfigured amount.

13. The user device, UE, of any one of claims 9 to 11 , wherein the change in location of the UE comprises one or more of: o a movement of the UE from an environment for which the beam configuration applies to a new environment for which a new beam configuration applies, e.g. moving between an urban environment and a rural environment, o a movement of the UE into a certain region or zone, e.g., from outdoor to indoor, or into a certain distance from the base station, e.g., with respect to a minimum required communication range, o a change in longitude and/or latitude and/or height beyond a configured or preconfigured threshold.

14. The user device, UE, of any one of claims 9 to 11 , wherein the indication from the wireless communication network comprises one or more of: o an indication from a base station included, e.g., in Downlink Control Information, DCI, a Medium Access Control Control Element, MAC CE, or a Radio Resource Control, RRC, signaling, or through broadcast/multicast messages to multiple UEs, o a signaling of a new beam configuration, e.g., when the UE moves from an environment for which the beam configuration applies to a new environment for which the new beam configuration applies, e.g. moving between an urban environment and a rural environment, o a signaling of a new configuration, e.g., when the UE moves from an environment for which a localization configuration applies to a new environment for which a new localization configuration applies, e.g. moving between an urban environment and a rural environment, o a signaling of new assistance information replacing current assistance information, e.g., when the UE moves from an environment for which current assistance information applies to a new environment for which the new assistance information applies, e.g. moving between an urban environment and a rural environment, o a signaling from another UE, e.g., via sidelink assistance information messages, e.g., AIM.

15. The user device, UE, of any one of the preceding claims, wherein the input data is stored in one or more of the following: in a memory that is part of the AI/ML model or functionality, a cloud server, another device, e.g., an associated UE, a base station, e.g., gNB or WiFi AP, a mobile edge cloud, MEC, located closely to a serving base station BS, a dedicated RAN entity, e.g., an AI/ML network function, NF.

16. The user device, UE, of any one of the preceding claims, wherein the one or more of tasks comprise one or more of the following:

- AI/ML model based access to a RAN,

- AI/ML model based network energy saving,

- AI/ML model based load balancing, an AI/ML model based mobility optimization,

- AI/ML model based modulation and coding scheme, MCS, selection,

- AI/ML model based synchronization,

- AI/ML model based encoding and/or decoding and/or precoding,

- AI/ML model based modulation and/or demodulation,

- AI/ML model based positioning or ranging,

- AI/ML model based joint communication and sensing, JSAC,

- AI/ML model based feedback calculation, e.g., CSI/CQI/PMI/RI feedback,

- AI/ML model based interference management, AI/ML model based quality of experience, QoE, and/or quality of service, QoS, predictions,

AI/ML model based network traffic forecasting.

17. The user device, UE, of claim 16, wherein the at least one AI/ML model or functionality performs beam management, and the UE is to determine one or more beams for a communication with one or more network entities of the wireless communication system using the at least one AI/ML model or functionality.

18. The user device, UE, of claim 17, wherein the UE is to obtain the one or more measurements from measuring the one or more beams.

19. The user device, UE, of claim 18, wherein a beam is identified by one or more of the following: a beam ID, a resource ID, a channel state information, CSI, or a signal derived from the CSI, a time index, e.g., the UE is configured or preconfigured with a certain timing and can derive from this, when certain beams can be decoded, a multi-TRP identifier, a physical cell ID, PCI, of a base station or BSSID of a WiFi access point.

20. The user device, UE, of claim 19, wherein the Chanel State Information, CSI, comprises one of more of the following: a Channel Quality Indicator, CQI, a precoding matrix indicator, PM I, a Chanel State Information Reference Signal, CSI-RS, resource indicator, CRI, a sounding reference signal, SRS,

21. A user device, UE, for a wireless communication network, wherein the UE is to perform one or more measurements of reference signal resources associated with respective beams, which are transmitted by a network entity of the wireless communication system, like a base station serving the UE, wherein one or more beams, which are to be used by the UE for a communication with the network entity, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the UE is to obtain the one or more beams to be used by the UE for the communication with the network entity or to be used by the network entity for the communication with the UE from at least one AI/ML model or functionality, which is operated at the network entity according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE, and/or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE.

22. The user device, UE, of claim 21 , wherein the at least one AI/ML model or functionality determines the one or more beams in accordance with the antenna configuration for the respective beams and is operated by the network entity, and wherein the UE is to report the one or more measurements to the network entity.

23. The user device, UE, of claim 21 , wherein the at least one AI/ML model or functionality determines the one or more beams in accordance with the antenna configuration for the respective beams and is operated by the network entity, and wherein the UE is to receive from the network entity feedback, which indicates one or more beams to be used by the UE for communication with the network entity.

24. The user device, UE, of claim 21 , wherein the at least one AI/ML model or functionality determines the one or more beams in accordance with the antenna configuration for the respective beams, and wherein the UE is to transmit to the network entity feedback, which indicates one or more beams to be used by the network entity for communication with the UE.

25. The user device, UE, of any one of claims 22 to 24, wherein the at least one AI/ML model or functionality operated by the network entity receives one or more measurements of reference signal resources associated with respective beams from one or more further UEs, e.g., for dynamically adjusting at least one AI/ML model or functionality using beamforming parameters based on one or more beam management reports.

26. The user device, UE, of claim 25, wherein the network entity is to receive the beam management reports from one or more of the following: one or more further UEs of the wireless communication system, the network, e.g., from a network function, NF, or from a neighboring gNB, e.g., a target gNB in case of a HO or potential CHO gNB candidate.

27. The user device, UE, of claim 25 or 26, wherein at least one AI/ML model or functionality analyzes one or more signal quality indicators and/or locations of the UE and the one or more further UEs to optimize a beam direction and/or a beam strength of the one or more beams to be used by the UE for the communication with the network entity.

28. The user device, UE, of claim 27, wherein the one or more signal quality indicators comprises one or more of the following: a pathloss or signal quality, e.g., using an uplink received signal strength indicator, RSSI, measured at the base station, a signal quality based on reference signals transmitted in the uplink by the UE, e.g., o based on sounding reference signals, SRS, o demodulation reference signals, DM-RS, o phase-tracking reference signals, PT-RS, a reciprocity-based signal quality, e.g., in case of a Time Division Duplex, TDD, system, based on CSI feedback provided in an uplink, feedback information sent by the UE, e.g., HARQ-ACKs or NACKs, or CBG-based ACKs or NACKs, provided by the UE, higher-layer statistics or feedback, e.g., packet delay or measured TCP slow-starts, or L3-RSRP or L3-latency.

29. The user device, UE, of any of the preceding claims, wherein the UE is to operate the at least one AI/ML model or functionality, receive from the network entity the antenna configuration for the respective beams, and determine the one or more beams to be used by the UE for the communication with the network entity by the at least one Al/M L model or functionality using the received antenna configuration for the respective beams and the one or more measurements.

30. The user device, UE, of claim 29, wherein the respective beams received by the UE comprise a proper subset of respective beams.

31. The user device, UE, of claim 30, wherein the UE is to receive different proper subsets of respective beams over time and use them as input data, e.g., to reconstruct a radio channel using different input values.

32. The user device, UE, of claim 30, wherein the UE is to receive the proper subset of respective beams together with specific configuration parameters.

33. The user device, UE, of any of the preceding claims, wherein the UE is to receive from the network entity at least one AI/ML model and/or parameters of the AI/ML model or functionality, e.g., a model trained at the network entity using the antenna configuration for the respective beams, and determine the one or more beams to be used by the UE for the communication with the network entity by the AI/ML model or functionality using the one or more measurements.

34. The user device, UE, of claim 16, wherein the UE is to use the AI/ML model or functionality to estimate or predict a beam for a communication with the network entity, e.g., by interpolating between beams by using the reference signal resources of which are sparsely measured to predict a different beam, e.g., a more optimal beam.

35. A user device, UE, for a wireless communication network, wherein the UE supports at least one AI/ML model or functionality, that uses the measurements of one or more first beams from a first set of beams to determine one or more second beams from a second set of beams, wherein the UE is to determine whether the at least one AI/ML model or functionality is applicable based on a configuration received from a network entity, or wherein the UE is to indicate to a network entity information about the at least one AI/ML model or functionality.

36. The user device, UE, of claim 35 wherein the UE is to measure one or more first beams from a first set of beams received from a network entity, wherein the UE is to determine one or more second beams from a second set of beams received from the network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data.

37. The user device, UE, of claim 35 or 36, wherein the UE is to receive a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for another network entity of the wireless communication system.

38. The user device, UE, of any one of claims 35 or 37, wherein the network entity is a serving base station, or a target base station, or a TRP, e.g., as in multi-TRP.

39. The user device, UE, of claim any one of claims 35 or 38, wherein the signaling includes network entity specific information allowing the UE to identify the network entity or the another network entity or describing beam specifics of the network entity.

40. The user device, UE, of claim 39, wherein network entity specific information comprises one or more of the following: an identification of the network entity or the another network entity, e.g., a vendor ID, or a configuration ID, or a gNB ID, or a UE ID, or a target gNB ID in case of a handover, HO, or a potential conditional handover, CHO, a cell ID or a Physical Cell ID, PCI of the cell comprising the network entity or the another network entity, an antenna configuration of the network entity or the another network entity, an antenna array ID at the network entity or the another network entity, e.g., an ID of a transmission reception point, TRP, at the network entity comprising a multi-TRP, a beam configuration of the network entity or the another network entity.

41. The user device, UE, of any one of claims 35 to 40, wherein the information about the at least one AI/ML model or functionality indicated to the network entity or the another network entity comprises one or more of the following:

• beam information about the first and second set of beams,

42. The user device, UE, of claim 41 , wherein the beam information comprises one or more of the following: one or more beam IDs of beams that belong to the first set of beams, one or more beam IDs of beams that belong to the second set of beams, a size of the first set of beams, e.g., a number of beams included in the first set of beams, a size of the second set of beams, e.g., a number of beams included in the second set of beams, a number of beams per dimension, like the number of beams per azimuth, or per elevation, angular differences between the beams of the first and/or second set, e.g., a phase offset between the beams of pi/4 resulting in 8 beams per dimension, amplitude differences between the beams of the first and/or second set, e.g., a , e.g., a power difference between the beams of the first and/or the second set, information about the network entity, like a cell ID, a gNB ID, a vendor ID, or a gNB Type, from which the first and second sets of beams have been obtained for training the AI/ML model or functionality.

43. The user device, UE, of any one of claims 35 to 42, wherein the second set of beams includes more beams than the first set of beams.

44. The user device, UE, of any one of claims 35 to 43, wherein the first set of beams is a proper subset of the second set of beams.

45. The user device, UE, of any one of claims 35 to 44, wherein, if the UE determines from the signaling that the AI/ML model or functionality is not working for the another network entity, the UE is to perform one or more of the following: modify or update the AI/ML model or functionality so as to meet the requirements for with the another network entity, replace the AI/ML model or functionality by a new AI/ML model or functionality meeting the requirements for with the another network entity, e.g., by selecting the new AI/ML model or functionality from a set of AI/ML models or functionalities stored in the UE, or by obtaining the new AI/ML model or functionality from the wireless communication network or from an external storage to which the UE is connectable via the wireless communication network, switching to a default AI/ML model or functionality, stop using the AI/ML model or functionality.

46. The user device, UE, of any one of claims 35 to 45, wherein, for a communication with the another network entity, the UE is configured by the another network entity with the first and/or second set of beams that is compatible with the first and/or second set of beams of the network entity that is associated with the at least one AI/ML model or functionality.

47. The user device, UE of claim 46, wherein compatible means the first and/or second set of beams from both network entities are identical, or the first and/or second set of beams from the another network entity is a proper subset of the first and/or second of beams from the network entity.

48. The user device, UE of claim 46 or 47, wherein beams are defined as identical if one or more of the following applies: the beams have the same ID, the beams are spatially aligned, e.g., relative to each other, the beams have the same QCL, the beams have the same orientation in azimuth and/or elevation, relative to each other, the beams are originating from the same antenna configuration, e.g., same number of antenna elements, e.g., as in a Massive MIMO array, the beams are coming from the same PLMN, the beams are coming from a same gNB configuration, e.g., wrt. gNB sectorization or TRP configuration, e.g., multi-TRP, or antenna configuration, e.g., array antenna with the same number of antenna elements, the same are coming from the same gNB or the same gNB type or same vendor.

49. The user device, UE, of claim 48, wherein the UE is to receive from a network entity the first set of beams, and use the AI/ML model or functionality to estimate or interpolate one or more corresponding characteristics, like the RSRP, the RSSI, the RSRQ, the SINR, the SNR of one or more beams of the second set which are not received at the UE.

50. The user device, UE, of claim 48, wherein the UE is to receive from a network entity only a plurality of proper subsets of the beams in the first or second set, wherein the plurality of proper subsets comprises a first subset and a second subset, use the first and second subsets for one or more initial measurements of one or more characteristics, like the RSRP, the RSSI, the RSRQ, the SINR, the SNR, and o report the one or more initial measurements to the network entity or o use an AI/ML model or functionality to predict one or more corresponding characteristics of

■ at least a third subset of beams or

■ the first or second set of beams.

51. The user device, UE, of claim 50, wherein the UE is to receive the plurality of proper subsets periodically or at configured or preconfigured intervals, wherein the one or more corresponding characteristics of the at least third subset or first or second set of beams are predicted for a following period or interval.

52. The user device, UE, of any one of claims 35 to 51 , wherein each of the first or second beams of the first or second set is a wide-coverage beam, e.g., SSB beams having a coverage area comprising coverage areas of a plurality narrow-coverage beams, and wherein the UE is to measure one or more of the wide-coverage beams.

53. The user device, UE, of claim 52, wherein the UE is to predict from the one or more measurements of the wide-coverage beams one or more of the narrow-coverage beams, e.g. CSI-RS beams.

54. The user device, UE, of any one of claims 49 to 53, wherein the UE is to receive for each of the plurality of proper subsets a reference signal resource configuration, like a CSI- RS resource config or a CSI-RS resource set.

55. The user device, UE, of claim 54, wherein a first reference signal resource configuration of a first subset of beams is a proper subset of a second reference signal resource configuration of a second subset of beams.

56. A user device, UE, for a wireless communication network, wherein the UE is configured or preconfigured with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionality, for performing a certain task, wherein the UE is to use one or more of the AI/ML models or functionality for performing the certain task, wherein the UE is to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to-use state allowing the UE to activate one or more of the non-used AI/ML models for performing the certain task, and wherein the UE is to put the one or more of the of the AI/ML models, which are currently not used for performing the certain task, into a ready-to-use state responsive to a first signaling.

57. The user device, UE, of claim 56, wherein the first signaling comprises one or more of an RRC signaling, a MAC CE, or a higher layer control signaling or signaling via sidelink, e.g., PC5 RRC or PSCCH or control signaling embedded into PSSCH.

58. The user device, UE, of claim 56 or 57, wherein the UE is to activate one or more of the non-used AI/ML models or functionalities for performing the certain task responsive to a second signaling.

59. The user device, UE, of claim 58, wherein the UE is to deactivate one or more of the used AI/ML models or functionalities responsive to the second signaling.

60. The user device, UE, of claim 58 or 59, wherein the second signaling comprises a DCI indicating a switch, an activation or a deactivation, e.g., using an index or a bitmap indicating used and non-used AI/ML models or functionalities to be deactivated/activated.

61. The user device, UE, of any of the preceding claims, wherein the UE comprise one or more of a power-limited UE, or a hand-held UE, like a UE used by a pedestrian, and referred to as a Vulnerable Road User, VRU, or a Pedestrian UE, P-UE, or an on-body or hand-held UE used by public safety personnel and first responders, and referred to as Public safety UE, PS-UE, or an loT UE or Ambient loT UE, e.g., a sensor, an actuator or a UE provided in a campus network to carry out repetitive tasks and requiring input from a gateway node at periodic intervals, or a mobile terminal, or a stationary terminal, or a cellular loT-UE, an industrial loT-UE, I loT, or a SL UE, or a vehicular UE, or a vehicular group leader UE, GL-UE, or a scheduling UE, S-UE, or an loT or narrowband loT, NB-loT, device, a NTN UE, or a WiFi device or WiFi station, STA, or a ground based vehicle, or an aerial vehicle, or a drone, or a moving base station, or road side unit, RSU, or a building, or any other item or device provided with network connectivity enabling the item/device to communicate using the wireless communication network, e.g., a sensor or actuator, or any other item or device provided with network connectivity enabling the item/device to communicate using a sidelink the wireless communication network, e.g., a sensor or actuator, or any sidelink capable network entity or just a network entity.

62. A base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, wherein the BS is to transmit respective beams to the UE, wherein the BS is to receive from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, wherein the BS is to determine from the respective beams one or more beams for a communication with the UE using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the AI/ML model or functionality operates on the basis of input data, the input data comprising the measurements received from the UE, and wherein the BS is to adapt the input data of the AI/ML model or functionality responsive to a change of the beam configuration.

63. A base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, wherein the BS is to transmit respective beams to the UE respective beams, wherein the BS is to receive from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, wherein one or more beams, which are to be used by the BS for a communication with the UE, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the one or more beams to be used for the communication are to be obtained from at least one AI/ML model or functionality, which is operated at the BS according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE, or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE.

64. A base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, the UE measuring one or more first beams from a first set of beams supported by the UE and determining one or more second beams from a second set of beams supported by a first network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data, and wherein the BS is to transmit to the UE a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for a BS.

65. The base station, BS, of claim 64, wherein the signaling includes BS specific information allowing the UE to identify the BS or describing beam specifics of the BS.

66. The base station, BS, of claims 64, wherein the BS is to receive from the UE the beam information about the first and second set of beams, and the signaling includes a notification to the UE whether the AI/ML model or functionality at the UE is also working for the BS, the notification generated by the BS using the beam information from the UE.

67. A base station, BS, for a wireless communication network, wherein the BS is to serve a user device, UE, of the wireless communication network, wherein the UE is configured or preconfigured with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionalities, for performing a certain task, wherein the BS is to signal to the UE to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to- use state allowing the UE to activate one or more of the non-used AI/ML models or functionalities for performing the certain task.

68. The base station, BS, of any one of claims 62 to 67, wherein the BS comprises one or more of a macro cell base station, or a small cell base station, or a central unit of a base station, or a distributed unit of a base station, or an Integrated Access and Backhaul, IAB, node, or a road side unit, RSU, or a WiFi access point, AP, or a UE, or a SL UE, or a group leader UE, GL-UE, or a relay or a remote radio head, or an AMF, or an SMF, or a core network entity or a network entity, or mobile edge computing, MEC, entity, or a network slice as in the NR or 5G core context, or any transmission/reception point, TRP, enabling an item or a device to communicate using the wireless communication network, the item or device being provided with network connectivity to communicate using the wireless communication network.

69. A wireless communication network, like a 3^rd Generation Partnership Project, 3GPP, system, comprising a one or more user devices, UEs, of any one of claims 1 to 61 and/or one or more base stations, BSs, of any one of claims 62 to 68.

70. A method for operating a user device, UE, for a wireless communication network, the method comprising: using, by the UE, at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality for performing one or more tasks associated with an operation of the UE, wherein the at least one AI/ML model or functionality operates on the basis of input data obtained from one or more measurements, and wherein the one or more measurements for obtaining the input data are associated with a configured or preconfigured configuration, and adapting, by the UE, the input data of the AI/ML model or functionality responsive to a certain event.

71. A method for operating a user device, UE, for a wireless communication network, the method comprising: performing, by the UE, one or more measurements of reference signal resources associated with respective beams, which are transmitted by a network entity of the wireless communication system, like a base station serving the UE, determining, by the UE, one or more beams, which are to be used by the UE for a communication with the network entity, using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, obtaining, by the UE, the one or more beams to be used by the UE for the communication with the network entity or to be used by the network entity for the communication with the UE from at least one AI/ML model or functionality, which is operated at the network entity according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE, and/or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the network entity for the respective beams, using the one or more measurements provided by the UE.

72. A method for operating a user device, UE, for a wireless communication network, the method comprising: supporting, by the UE, at least one AI/ML model or functionality, that uses the measurements of one or more first beams from a first set of beams to determine one or more second beams from a second set of beams, determining, by the UE, whether the at least one AI/ML model or functionality is applicable based on a configuration received from a network entity, or indicating, by the UE, to a network entity information about the at least one AI/ML model or functionality.

73. A method for operating a user device, UE, for a wireless communication network, the method comprising: configuring por preconfiguring the UE with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionality, for performing a certain task, using, by the UE, one or more of the AI/ML models or functionality for performing the certain task, maintaining, by the UE, one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to-use state allowing the UE to activate one or more of the non-used AI/ML models for performing the certain task, and putting, by the UE, the one or more of the of the AI/ML models, which are currently not used for performing the certain task, into a ready-to-use state responsive to a first signaling.

74. A method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, transmitting, by the BS, respective beams to the UE, receiving, by the BS, from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, determining, by the BS, from the respective beams one or more beams for a communication with the UE using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, wherein the AI/ML model or functionality operates on the basis of input data, the input data comprising the measurements received from the UE, and adapting, by the BS, the input data of the AI/ML model or functionality responsive to a change of the beam configuration.

75. A method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, transmitting, by the BS, respective beams to the UE respective beams, receiving, , by the BS, from the UE one or more measurements of the respective beams obtained in accordance with a configured or preconfigured beam configuration at the UE, determining, by the BS, one or more beams, which are to be used by the BS for a communication with the UE, are determined using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, obtaining the one or more beams to be used for the communication from at least one AI/ML model or functionality, which is operated at the BS according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE, or at least one AI/ML model or functionality, which is operated at the UE according to an antenna configuration of the BS for the respective beams, using the one or more measurements provided by the UE.

76. A method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, the UE measuring one or more first beams from a first set of beams supported by the UE and determining one or more second beams from a second set of beams supported by a first network entity using at least one Artificial Intelligence/Machine Learning, AI/ML, model or functionality, the AI/ML model or functionality using the measured one or more first beams as input data, and transmitting, by the BS, to the UE a signaling allowing the UE to determine whether the AI/ML model or functionality is also working for a BS.

77. A method for operating a base station, BS, for a wireless communication network, the method comprising: serving, by the BS, a user device, UE, of the wireless communication network, wherein the UE is configured or preconfigured with a plurality of Artificial Intelligence/Machine Learning modes, AI/ML models or functionalities, for performing a certain task, signaling, by the BS, to the UE to maintain one or more of the AI/ML models or functionalities, which are currently not used for performing the certain task, in a ready-to- use state allowing the UE to activate one or more of the non-used AI/ML models or functionalities for performing the certain task.

78. A non-transitory computer program product comprising a computer readable medium storing instructions which, when executed on a computer, perform the method of one of claims 70 to 77.