FI130871B1

FI130871B1 - Transmission beam determination

Info

Publication number: FI130871B1
Application number: FI20235185A
Authority: FI
Inventors: Keeth Laddu; Andrea Bonfante; Frederick Vook
Original assignee: Nokia Technologies Oy
Priority date: 2023-02-16
Filing date: 2023-02-16
Publication date: 2024-04-30
Also published as: FI20235185A1

Abstract

Method, systems, and computer programs are provided supporting the determination of a transmission beam. A method executed by a reception entity comprises transmitting (601) at least one transmission beam condition determined based on a first measurement resource set to a transmission entity, receiving (602) a second measurement resource set, and determining (603) at least one transmission beam based on measurements performed according to the second measurement resource set.

Description

TRANSMISSION BEAM DETERMINATION

THECHNICAL FIELD

[0001] The subject disclosure generally relates to wireless communication systems and, in particular, to transmission beam determination, or to enablers thereof. Yet more particularly, the subject disclosure provides methods and apparatuses of transmission beam determination or related enablers with machine learning inference suitable, for example, for transmission beam pair prediction.

BACKGROUND

[0002] Wireless telecommunication systems are under constant development. There is a need for higher data rates and high quality of service. Reliability requirements are constantly rising and ways and means to ensure reliable connections and data traffic while keeping transmission delays minimal are constantly under development.

[0003] Communication in mobile networks can be based on beam pairs that comprise transmission and reception beams. Transmission beams are provided by a transmission entity, for example, a base station in downlink communication. Reception beams are provided by a reception entity, for example, a user equipment in downlink communication.

Machine learning for determination of transmission beams or transmission beam pairs is discussed by the 3rd Generation Partnership Project (3GPP).

[0004] By using a machine learning based beam prediction, it seems possible to predict the best transmission/reception beam pair(s) by using only a sub-set of measurements of transmission/reception beam pairs as input of the machine learning model. To ensure a high & reliability of the predication, the setups relating to the sub-set of measurements have to be defined with care. = 25 — [0005] Hence, methods and apparatuses enabling transmission beam determination with = machine learning inference for transmission beam pair prediction are herein presented. a a 3 5 SUMMARY

N

2 [0006] According to a first aspect of the subject disclosure, a method executed by a reception entity is provided. The method comprises transmitting at least one transmission beam condition determined based on a first measurement resource set to a transmission entity, receiving a second measurement resource set, and determining at least one transmission beam based on measurements performed according to the second measurement resource set.

[0007] In embodiments, determining is based on an output of a machine learning model, in response to the second measurement resource set being in line with the at least one transmission beam condition. In further embodiments, the at least one transmission beam condition is comprised by a reception entity capability report triggered in response to receiving a reception entity capability inquiry from the transmission entity, by a measurement report triggered in response to receiving a reconfiguration of the first measurement resource set from the transmission entity, or by a measurement report triggered by the reception entity during application of the machine learning model.

[0008] In embodiments, the at least one transmission beam condition comprises an indication of at least one transmission beam preferred by the reception entity. In further embodiments, the at least one transmission beam is indicated by at least one downlink reference signal. In yet further embodiments, the at least one downlink reference signal is associated to the first measurement resource set.

[0009] In further embodiments, the first measurement resource set is configured during data collection for the machine learning model. In yet further embodiments, the first measurement resource set is configured during training, validation, testing, fine-tuning, and/or monitoring of the machine learning model. In yet further embodiments, the at least one transmission beam is indicated by a set index relating to a set of downlink reference signals from multiple sets of downlink reference signals.

[0010] In embodiments, the at least one transmission beam condition comprises an & indication of at least one repetition value preferred by the reception entity, wherein a repetition value is associated with one transmission beam. In further embodiments, the at = least one repetition value is associated with a transmission beam preferred by the reception = entity and/or a transmission beam to be used by the machine learning model. In yet further

E embodiments, the at least one repetition value is indicated with a bitmap, wherein the bitmap 2 comprises indications relating to one or more sets of downlink reference signals with 2 30 different repetition values.

N [0011] In embodiments, the at least one transmission beam condition comprises an indication of at least one switching gap preferred by the reception entity, wherein a switching gap relates to a gap between consecutive repetitions of one transmission beam. In further embodiments, the at least one switching gap is associated with at least one of a transmission beam preferred by the reception entity, a transmission beam related to a repetition value preferred by the reception entity, and a transmission beam to be used by the machine learning model. In yet further embodiments, the at least one switching gap is indicated by a minimum — time duration to be fulfilled for downlink reference signal repetition per downlink reference signal. In other embodiments, the at least one switching gap is indicated by a minimum time duration to be fulfilled for repetition in transmission of a set of downlink reference signals within one downlink reception beam.

[0012] In embodiments, the method further comprises, in response to the second measurement resource set being not in line with the at least one transmission beam condition, not applying the machine learning model for determining the at least one transmission beam.

[0013] According to a second aspect, a reception entity is presented that comprises a processor and a memory with instructions which, when executed by the processor, cause the reception entity to transmit at least one transmission beam condition determined based on a — first measurement resource set to a transmission entity, receive a second measurement resource set from the transmission entity, and determine at least one transmission beam based on measurements performed according to the second measurement resource set. In embodiments, the reception entity is further caused to execute the processes as described herein.

[0014] According to a third aspect, a method executed by a transmission entity is provided, which comprises receiving at least one transmission beam condition from a reception entity relating to a first measurement resource set, and transmitting a second measurement resource set to the reception entity, wherein the second measurement resource & set 1s determined based on the at least one transmission beam condition. — [0015] According to a fourth aspect, a transmission entity is presented that comprises a = processor and a memory with instructions which, when executed by the processor, cause the = network node to receive at least one transmission beam condition from a reception entity

E relating to a first measurement resource set, and transmit a second measurement resource set 2 to the reception entity, wherein the second measurement resource set is determined based on 2 30 the at least one transmission beam condition.

N [0016] According to a fifth aspect, a non-transitory computer readable medium is provided that comprises program instructions that, when executed by an apparatus, cause the apparatus to execute the processes as described herein.

[0017] The above-noted aspects and features may be implemented in systems, apparatuses, methods, articles and non-transitory computer-readable media depending on the desired configuration. The subject disclosure may be implemented in and used with a number of different types of devices, including but not limited to cellular phones, tablet computers, wearable computing devices, portable media players, and any of various other computing devices.

[0018] This summary is intended to provide a brief overview of some of the aspects and features according to the subject disclosure. Accordingly, it will be appreciated that the above-described features are merely examples and should not be construed to narrow the scope of the subject disclosure in any way. Other features, aspects, and advantages of the subject disclosure will become apparent from the following detailed description, drawings and claims.

TERMINOLOGY

[0019] To facilitate understanding on the terminologies in the subject disclosure, the following list of terminologies is provided:

Data collection A process of collecting data by the network nodes, management entity, or UE for the purpose of artificial intelligence (AI) / machine learning (ML) model training, data analytics and inference

AI/ML Model A data driven algorithm that applies AI/ML techniques 2 to generate a set of outputs based on a set of inputs.

O

N AI/ML model training A process to train an AI/ML Model [by learning the

N input/output relationship] in a data driven manner and © obtain the trained AI/ML Model for inference

E AT/ML model Inference A process of using a trained AI/ML model to produce a > set of outputs based on a set of inputs

LO

© AI/ML model validation A subprocess of training, to evaluate the quality of an 2 AI/ML model using a dataset different from one used o for model training, that helps selecting model parameters that generalize beyond the dataset used for model training.

AI/ML model testing A subprocess of training, to evaluate the performance of a final AI/ML model using a dataset different from one used for model training and validation. Differently from AI/ML model validation, testing does not assume subsequent tuning of the model.

UE-side (AI/ML) model An AI/ML Model whose inference is performed entirely at the user equipment (UE)

Network-side (AI/ML) model | An AI/ML Model whose inference is performed entirely at the network

One-sided (AI/ML) model A UE-side (AI/ML) model or a Network-side (AI/ML) model

Two-sided (AI/ML) model A paired AI/ML Model(s) over which joint inference is performed, where joint inference comprises AI/ML

Inference whose inference is performed jointly across the UE and the network, 1.e., the first part of inference is firstly performed by UE and then the remaining part is performed by gNB, or vice versa.

AI/ML model transfer Delivery of an AI/ML model over the air interface, either parameters of a model structure known at the receiving end or a new model with parameters.

Delivery may contain a full model or a partial model.

Model download Model transfer from the network to UE

Model upload Model transfer from UE to the network

Federated learning / federated | A machine learning technique that trains an AI/ML training model across multiple decentralized edge nodes (e.g.,

UEs, gNBs) each performing local model training using local data samples. The technique requires @ multiple interactions of the model, but no exchange of

S local data samples.

N Offline field data The data collected from field and used for offline © training of the AI/ML model z Online field data The data collected from field and used for online > training of the AT/ML model

LO

O

N Supervised learning A process of training a model from input and its corresponding labels.

Unsupervised learning A process of training a model without labelled data.

Semi-supervised learning A process of training a model with a mix of labelled data and unlabelled data

Reinforcement Learning (RL) | A process of training an AI/ML model from input (a.k.a. state) and a feedback signal (a.k.a. reward) resulting from the model’s output (a.k.a. action) in an environment the model is interacting with. enable an AI/ML model for a specific function

Model deactivation disable an AI/ML model for a specific function

Model switching Deactivating a currently active AI/ML model and activating a different AI/ML model for a specific function

Proprietary-format models ML models of vendor-/device-specific proprietary format, from 3GPP perspective

NOTE: An example is a device-specific binary executable format. Proprietary-format models are not mutually recognizable across vendors, hide model design information from other vendors when shared.

Open-format models ML models of specified format that are mutually recognizable across vendors and allow interoperability, from 3GPP perspective

NOTE: Consider “proprietary model” and “open- format model” as two separate model format categories. Open-format models are mutually recognizable between vendors, do not hide model design information from other vendors when shared

Model identification A process/method of identifying an AI/ML model for the common understanding between the network (NW) o and the UE

N

A Note: The process/method of model identification may

N or may not be applicable. © Note: Information regarding the AI/ML model may be - shared during model identification. a > Functionality identification A process/method of identifying an AI/ML functionality for

O the common understanding between the NW and the UE a Note: Information regarding the AI/ML functionality may

O be shared during functionality identification.

BRIEF DESCRIPTION OF THE DRAWINGS

[0020] A better understanding of the subject disclosure can be obtained when the following detailed description of various embodiments is considered in conjunction with the following drawings, in which:

[0021] FIG. 1 shows a schematic diagram of an example wireless network;

[0022] FIG. 2 shows a schematic diagram of an example wireless device;

[0023] FIG. 3 shows a schematic diagram of an example network node;

[0024] FIG. 4 presents a schematic overview on a determination of a transmission beam with a machine learning model.

[0025] FIG. 5 depicts a message signaling diagram of transmission beam determination with beam pair prediction according to agreements of 3GPP meetings.

[0026] FIG. 6 is a flow chart of a method for transmission beam determination at a reception entity according to the disclosure.

[0027] FIG. 7 is a flow chart of a method executed by a transmission entity for enabling transmission beam determination according to the disclosure.

[0028] FIG. 8 presents a detailed example of determination of a transmission beam with applying a machine learning model according to embodiments.

[0029] FIG. 9 shows a message signaling diagram of determination of a transmission beam triggered by a user equipment capability inquiry according to an embodiment.

[0030] FIG. 10 shows a message signaling diagram of determination of transmission beam conditions triggered by a new configuration of a first measurement set.

[0031] FIG. 11 shows a message signaling diagram of determination of transmission

J beam conditions triggered during application of the machine learning model.

S DETAILED DESCRIPTION

S 25 — [0032] The examples and embodiments set forth below represent information to enable 2 those skilled in the art to practice the subject disclosure. Upon reading the following

E: description in light of the accompanying drawing figures, those skilled in the art will 3 understand the concepts of the description and will recognize applications of these concepts lo not particularly addressed herein. It should be understood that these concepts and

O 30 applications fall within the scope of the description.

[0033] In the following description, numerous specific details are set forth. However, it is understood that embodiments may be practiced without these specific details. In other instances, well-known circuits, structures, and techniques have not been shown in detail in order not to obscure the understanding of the description. Those of ordinary skill in the art, with the included description, will be able to implement appropriate functionality without undue experimentation.

[0034] References in the specification to "one embodiment," "an embodiment," "an example embodiment," etc., indicate that the embodiment described may include a particular feature, structure, or characteristic, but every embodiment may not necessarily include the particular feature, structure, or characteristic. Moreover, such phrases are not necessarily referring to the same embodiment. Further, when a particular feature, structure, or characteristic is described in connection with an embodiment, it is submitted that it is within the knowledge of one skilled in the art to implement such feature, structure, or characteristic in connection with other embodiments whether or not explicitly described.

[0035] As used herein, "plurality" means two or more. As used herein, a "set" of items may include one or more of such items. As used herein, whether in the subject disclosure or the claims, the terms "comprising", "including", "carrying", "having", "containing", "involving", and the like are to be understood to be open-ended, i.e., to mean including but not limited to. Only the transitional phrases "consisting of" and "consisting essentially of", respectively, are closed or semi-closed transitional phrases with respect to claims. Use of ordinal terms such as "first", "second", "third", etc., in the claims or the subject disclosure to — modify an element does not by itself connote any priority, precedence, or order of one element over another or the temporal order in which acts of a method are performed, but are used merely as labels to distinguish one element having a certain name from another element having a same name (but for use of the ordinal term) to distinguish the elements. As used & herein, "and/or" and "at least one of" means that the listed items are alternatives, but the alternatives also include any combination of the listed items. = [0036] Before explaining the examples according to the subject disclosure in detail, = certain general principles of a wireless communication system are briefly explained with

E reference to FIGS. 1 to 3 to assist in understanding the technology underlying the described 2 examples. 2 30 [0037] FIG. 1 illustrates an example of a wireless network 100 that may be used for

N wireless communications. Wireless network 100 includes wireless devices, such as UEs 110 (e.g., 110A-110B), and network nodes, such as radio access nodes 120 (e.g., 120A-120B) (e.g., eNBs, gNBs, etc.), connected to one or more network nodes 130 via an interconnecting network 125. The network 100 may use any suitable deployment scenarios. UEs 110 within coverage area 115 may each be capable of communicating directly with radio access nodes 120 over a wireless interface. In some embodiments, UEs 110 may also be capable of communicating with each other via D2D communication.

[0038] As an example, UE 110A may communicate with radio access node 120A over a wireless interface. That is, UE 110A may transmit wireless signals to and/or receive wireless signals from radio access node 120A. The wireless signals may contain voice traffic, data traffic, control signals, and/or any other suitable information.

[0039] As used herein, the term "user equipment" (UE) has the full breadth of its — ordinary meaning and may refer to any type of wireless device which can communicate with a network node and/or with another UE in a cellular or mobile or wireless communication system. Examples of UE are target device, D2D UE, machine type UE or UE capable of machine-to-machine (M2M) communication, personal digital assistant, tablet, mobile terminal, smart phone, laptop embedded equipped (LEE), laptop mounted equipment (LMF), USB dongles, ProSe UE, vehicle-to-vehicle (V2V) UE, V2X UE, MTC UE, eMTC

UE, FeMTC UE, UE Cat 0, UE Cat MI, narrow band IoT (NB-IoT) UE, UE Cat NB, etc.

Example embodiments of a UE are described in more detail below with respect to FIG. 2.

[0040] In some embodiments, an area of wireless signal coverage 115 associated with a radio access node 120 may be referred to as a cell. However, particularly with respect to the fifth generation (5G) / New Radio (NR) mobile communication concepts, beams may be used instead of cells and, as such, it is important to note that concepts described herein are equally applicable to both cells and beams.

[0041] With respect to a beam-based mobile communication system, the radio access

N node 120 (base station) may transmit a beamformed signal to the UE 110 in one or more a 25 — transmit directions (transmission beam, Tx beam). The UE 110 may receive the beamformed = signal from the base station 120 in one or more receive directions (reception beam, Rx = beam). The UE 110 may also transmit a beamformed signal to the base station 120 in one or 7 more directions and the base station 120 may receive the beamformed signal from the UE < 110 in one or more directions. The base station 120 and the UE 110 may determine the best 2 30 receive and transmit directions, e.g., best in the sense of these directions leading to the

N highest link guality or fulfilling other guality conditions in the most suitable manner, for each of the base station/UE pairs.

[0042] The interconnecting network 125 may refer to any interconnecting system capable of transmitting audio, video, signals, data, messages, etc., or any combination of the preceding. The interconnecting network 125 may include all or a portion of a public switched telephone network (PSTN), a public or private data network, a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a local, regional, or global communication or computer network such as the Internet, a wireline or wireless network, an enterprise intranet, or any other suitable communication link, including combinations thereof.

[0043] In some embodiments, the network node 130 may be a core network node, managing the establishment of communication sessions and other various other functionalities for UEs 110. Examples of network node 130 may include mobile switching center (MSC), MME, serving gateway (SGW), packet data network gateway (PGW), operation and maintenance (O&M), operations support system (OSS), SON, positioning node (e.g., Enhanced Serving Mobile Location Center, E-SMLC), location server node,

MDT node, etc. UEs 110 may exchange certain signals with the network node 130 using the non-access stratum (NAS) layer. In non-access stratum signaling, signals between UEs 110 and the network node 130 may be transparently passed through the radio access network. In some embodiments, radio access nodes 120 may interface with one or more network nodes 130 over an internode interface.

[0044] As used herein, the term "network node" has the full breadth of its ordinary meaning and may correspond to any type of radio access node (or radio network node) or any network node, which can communicate with a UE and/or with another network node in a cellular or mobile or wireless communication system. Examples of network nodes are & NodeB, MeNB, SeNB, a network node may belonging to MCG or SCG, base station (BS), — multi-standard radio (MSR) radio access node such as MSR BS, eNodeB, network = controller, radio network controller (RNC), base station controller (BSC), relay, donor node = controlling relay, base transceiver station (BTS), access point (AP), transmission point, 7 transmission node, RRU, RRH, node in distributed antenna system (DAS), core network < node (e.g., MSC, MME, etc.), O&M, OSS, Self-organizing Network (SON), positioning 2 30 = node(eg, E-SMLC), MDT, test equipment, etc. Example embodiments of a network node

N are described in more detail below with respect to FIG. 3.

[0045] In some embodiments, radio access node 120 may be a distributed radio access node. The components of the radio access node 120, and their associated functions, may be separated into two main units (or sub-radio network nodes) which may be referred to as the central unit (CU) and the distributed unit (DU). Different distributed radio network node architectures are possible. For instance, in some architectures, a DU may be connected to a

CU via dedicated wired or wireless link (e.g, an optical fiber cable) while in other architectures, a DU may be connected a CU via a transport network. Also, how the various functions of the radio access node 120 are separated between the CU(s) and DU(s) may vary depending on the chosen architecture.

[0046] Exemplary wireless communication systems are architectures standardized by the 3rd Generation Partnership Project (3GPP). A latest 3GPP based development is often referred to as the long-term evolution (LTE) of the Universal Mobile Telecommunications

System (UMTS) radio-access technology (RAT). The various development stages of the 3GPP specifications are referred to as releases. More recent developments of the LTE are often referred to as LTE Advanced (LTE-A). The LTE (LTE-A) employs a radio mobile architecture known as the Evolved Universal Terrestrial Radio Access Network (E-UTRAN) and a core network known as the Evolved Packet Core (EPC). Base stations of such systems are known as evolved or enhanced Node Bs (eNBs) and provide E-UTRAN features such as user plane Packet Data Convergence/Radio Link Control/Medium Access Control/Physical layer protocol (PDCP/RLC/MAC/PHY) and control plane Radio Resource Control (RRC) protocol terminations towards the communication devices. Other RAT examples comprise those provided by base stations of systems that are based on technologies such as WLAN and/or Worldwide Interoperability for Microwave Access (WiMax). A base station can provide coverage for an entire cell or similar radio service area. Core network elements include Mobility Management Entity (MME), Serving Gateway (S-GW) and Packet & Gateway (P-GW). — [0047] An example of a suitable communications system is the 5G or NR concept. = Network architecture in NR may be similar to that of LTE-A. Base stations of NR systems = may be known as next generation Node Bs (gNBs). Changes to the network architecture may 7 depend on the need to support various radio technologies and finer Quality of Service (QoS) co support, and some on-demand requirements for QoS levels to support Quality of Experience 2 30 (QoE) of user point of view. Also network aware services and applications, and service and

N application aware networks may bring changes to the architecture. Those are related to

Information Centric Network (ICN) and User-Centric Content Delivery Network (UC-CDN) approaches. NR may use multiple input-multiple output (MIMO) antennas, many more base stations or nodes than the LTE (a so-called small cell concept), including macro sites operating in co-operation with smaller stations and perhaps also employing a variety of radio technologies for better coverage and enhanced data rates.

[0048] Future networks may utilize network functions virtualization (NFV) which is a network architecture concept that proposes virtualizing network node functions into "building blocks" or entities that may be operationally connected or linked together to provide services. A virtualized network function (VNF) may comprise one or more virtual machines running computer program codes using standard or general type servers instead of customized hardware. Cloud computing or data storage may also be utilized. In radio communications this may mean node operations to be carried out, at least partly, in a server, host or node operationally coupled to a remote radio head. It is also possible that node operations will be distributed among a plurality of servers, nodes, or hosts. It should also be understood that the distribution of labour between core network operations and base station operations may differ from that of the LTE or even be non-existent.

[0049] An example 5G core network (CN) comprises functional entities. The CN is connected to a UE via the radio access network (RAN). An UPF (User Plane Function) whose role is called PSA (PDU Session Anchor) may be responsible for forwarding frames back and forth between the DN (data network) and the tunnels established over the 5G towards the UEs exchanging traffic with the data network (DN). The UPF is controlled by an SMF (Session Management Function) that receives policies from a PCF (Policy Control

Function). The CN may also include an AMF (Access & Mobility Function).

[0050] Generally, all concepts disclosed herein may be applicable to different communication networks, comprising but not limited to LTE, LTE-A, 5G, 5G advanced, 6G, & and other future or already implemented networks.

[0051] FIG. 2 is a schematic diagram of an example wireless device, UE 110, in = accordance with certain embodiments. UE 110 includes a transceiver 210, processor 220, = memory 230, and a network interface 240. In some embodiments, the transceiver 210

E facilitates transmitting wireless signals to and receiving wireless signals from radio access 2 node 120 (e.g., via transmitter(s) (Tx), receiver(s) (Rx) and antenna(s)). The processor 220 2 30 executes instructions to provide some or all of the functionalities described herein as being & provided by UE 110, and the memory 230 stores the instructions executed by the processor 220. In some embodiments, the processor 220 and the memory 230 form processing circuitry.

[0052] The processor 220 may include any suitable combination of hardware to execute instructions and manipulate data to perform some or all of the described functions of a wireless device, such as the functions of UE 110 described herein. In some embodiments, the processor 220 may include, for example, one or more computers, one or more central — processing units (CPUs), one or more microprocessors, one or more application specific integrated circuits (ASICs), one or more field programmable gate arrays (FPGAs) and/or other logic.

[0053] The memory 230 is generally operable to store instructions, such as a computer program, software, an application including one or more of logic, rules, algorithms, code, tables, etc. and/or other instructions capable of being executed by a processor 220. Examples of memory 230 include computer memory (for example, Random Access Memory (RAM) or Read Only Memory (ROM)), mass storage media (for example, a hard disk), removable storage media (for example, a Compact Disk (CD) or a Digital Video Disk (DVD)), and/or or any other volatile or non-volatile, non- transitory computer-readable and/or computer- executable memory devices that store information, data, and/or instructions that may be used by the processor 220 of UE 110. For example, the memory 230 includes computer program code causing the processor 220 to perform processing according to the methods described herein, e.g., the method of FIG. 6.

[0054] The network interface 240 is communicatively coupled to the processor 220 and — may refer to any suitable device operable to receive input for UE 110, send output from UE 110, perform suitable processing of the input or output or both, communicate to other devices, or any combination thereof. The network interface 240 may include appropriate hardware (e.g., port, modem, network interface card, etc.) and software, including protocol & conversion and data processing capabilities, to communicate through a network. — [0055] Other embodiments of UE 110 may include additional components beyond those = shown in FIG. 2 that may be responsible for providing certain aspects of the wireless device's = functionalities, including any of the functionalities described herein and/or any additional 7 functionalities (including any functionality necessary to support the mechanisms according & to the subject disclosure). As an example, UE 110 may include input devices and circuits, 2 30 output devices, and one or more synchronization units or circuits, which may be part of the

N processor 220. Input devices include mechanisms for entry of data into UE 110. For example, input devices may include input mechanisms, such as a microphone, input elements, a display, etc. Output devices may include mechanisms for outputting data in audio, video and/or hard copy format. For example, output devices may include a speaker, a display, etc.

[0056] In some embodiments, the wireless device UE 110 may comprise a series of modules configured to implement the functionalities of the wireless device described herein.

[0057] It will be appreciated that the various modules may be implemented as combination of hardware and software, for instance, the processor, memory, and transceiver(s) of UE 110 shown in FIG. 2. Some embodiments may also include additional modules to support additional and/or optional functionalities.

[0058] FIG. 3 is a schematic diagram of an example radio access node 120 or network node 130. The example radio access node 120 or network node 130 may include one or more of a transceiver 310, processor 320, memory 330, and network interface 340. In some embodiments, the transceiver 310 facilitates transmitting wireless signals to and receiving wireless signals from wireless devices, such as UE 110 (e.g, via transmitter(s) (Tx), receiver(s) (Rx), and antenna(s)). The processor 320 executes instructions to provide some or all of the functionalities described herein as being provided by the radio access node 120 or the network node 130, the memory 330 stores the instructions executed by the processor 320. In some embodiments, the processor 320 and the memory 330 form processing circuitry. The network interface 340 can communicate signals to backend network components, such as a gateway, switch, router, Internet, Public Switched Telephone

Network (PSTN), core network nodes or radio network controllers, etc.

[0059] The processor 320 can include any suitable combination of hardware to execute instructions and manipulate data to perform some or all of the described functions of the radio access node 120 or the network node 130, such as those described herein. In some & embodiments, the processor 320 may include, for example, one or more computers, one or more central processing units (CPUs), one or more microprocessors, one or more application = specific integrated circuits (ASICs), one or more field programmable gate arrays (FPGAs) = and/or other logic. 7 [0060] The memory 330 is generally operable to store instructions, such as a computer < program, software, an application including one or more of logic, rules, algorithms, code, 2 30 tables, etc. and/or other instructions capable of being executed by a processor 320. Examples

N of memory 330 include computer memory (for example, Random Access Memory (RAM) or Read Only Memory (ROM)), mass storage media (for example, a hard disk), removable storage media (for example, a Compact Disk (CD) or a Digital Video Disk (DVD)), and/or or any other volatile or non-volatile, non- transitory computer-readable and/or computer- executable memory devices that store information. For example, the memory 330 includes computer program code causing the processor 320 to perform processing according to the methods described herein, e.g., the method of FIG. 7.

[0061] In some embodiments, the network interface 340 is communicatively coupled to the processor 320 and may refer to any suitable device operable to receive input for the radio access node 120 or the network node 130, send output from the radio access node 120 or the network node 130, perform suitable processing of the input or output or both, communicate to other devices, or any combination of the preceding. The network interface 340 may include appropriate hardware (e.g., port, modem, network interface card, etc.) and software, including protocol conversion and data processing capabilities, to communicate through a network.

[0062] Other embodiments of the radio access node 120 or the network node 130 can include additional components beyond those shown in FIG. 3 that may be responsible for — providing certain aspects of the node’s functionalities, including any of the functionalities described herein and/or any additional functionalities (including any functionality necessary to support the solutions described herein). The various different types of radio access nodes or network nodes may include components having the same physical hardware but configured (e.g., via programming) to support different radio access technologies, or may represent partly or entirely different physical components.

[0063] Processors, interfaces, and memory similar to those described with respect to

FIG. 3 may be included in other nodes (such as UE 110, radio access node 120, etc.). Other nodes may optionally include or not include a wireless interface (such as the transceiver & described in FIG. 3). — [0064] In some embodiments, the radio access node 120 or the network node 130 may = comprise a series of modules configured to implement the functionalities of the radio access = node 120 or the network node 130 described herein. 7 [0065] It will be appreciated that the various modules may be implemented as ce combination of hardware and software, for instance, the processor, memory, and 2 30 transceiver(s) of the radio access node 120 or the network node 130 shown in FIG. 3. Some

N embodiments may also include additional modules to support additional and/or optional functionalities.

[0066] Before referring to FIGS. 6 to 10 and describing methods of determining a transmission beam, e.g., by predicting a transmission/reception beam pair with the best connection conditions, according to this disclosure, some background information and aspects related to the subject disclosure will be provided. It should be noted that all concepts described herein, although described, e.g., for one communication direction, e.g., for downlink communication, are applicable for the other direction as well, e.g., in the uplink (UL) communication. Moreover, concepts described for one entity, e.g., a UE 110, are applicable to another entity, e.g., a base station or network node, when considering for example another communication direction or another network setting as will be apparent to the skilled person.

[0067] Release-18 3GPP of 5G started the study on Artificial Intelligence (Al)/Machine

Learning (ML) for NR Air Interface. It has been defined that the goal is to explore the benefits of augmenting the air interface with features enabling improved support of AI/ML- based algorithms for enhanced performance, reduced complexity and/or reduced overhead. — Several use cases are currently considered to enable the identification of a common AI/ML framework, including functional reguirements of AI/ML architecture, which could be used in subseguent projects.

[0068] One relevant area, in which AI/ML-based algorithms can be applied is beam management (BM). For AI/ML enhancements related to beam management, two sub-use cases are identified: beam prediction in the spatial domain (BM-Casel) and beam prediction in the time domain (BM-Case2). The primary motivation is to support a reduced overhead and lower beam measurements and reporting latency.

[0069] It was already agreed in 3GPP meetings and discussion that BM-Casel & comprises spatial-domain downlink (DL) beam prediction for a set A of beams based on — measurement results of set B of beams. AI/ML inference can be done at the UE or the = network side. The determination of DL transmission (Tx) or reception (Tx) beams to be used = for mobile communication or signaling can be achieved by predicting the DL Tx beam, the 7 DL Rx beam, and/or the DL Tx-Rx beam pair. < [0070] For the DL Tx-Rx beam pair prediction at the UE side, a trained ML model at 2 30 the UE 110 may be applied. The model may be different for different kinds of UEs and may

N further depend on antenna panel configuration and other aspects which are specific to UEs 110. Such ML models for predicting the best beam pair help to avoid P2-P3 beam refinement procedures.

[0071] The 5G beam management procedures in 5G has have been divided into three phases. P1 procedure relates to that, before a data flow is enabled in the scheduler, periodic

SSB beam scanning is implemented on the base station side in a certain interval (the SSB periodicity). At the same time, wide beam scanning is implemented on the UE side to determine the optimal receive wide beam. P2 procedure relates to performing a beam sweep in a narrower angular sector than in P1. The narrow beams closest to the wide beam in the beam grid are selected to be examined using Channel State Information (CSI)- Reference

Signal (RS), which is followed by CSI-report. P3 procedure relates to that the optimal narrow beam is selected from P2 and CSI-RSs are transmitted to the UE, which updates its Rx beam.

In the data transmission, the base station uses the best BS Tx beam found during P2 and the

UE uses the best UE Rx beam found during P3.

[0072] Moreover, when the ML model inference is on the UE side, UE 110 does not need to report its own Rx beam selection to the network (NW). Additionally, for measurements, it is up to UE 110 to schedule its Rx beam operation for receiving the DL Tx beams.

[0073] The best DL Tx-Rx beam pair that identifies the DL Tx beam to be used for communication/signaling may be the Tx-Rx beam pair that results in the largest Layer 1 (L1)-Reference Signal Received Power (RSRP) over all Tx and Rx beams. Alternatively, the best DL Tx-Rx beam pair may be the Tx-Rx beam pair that results in the largest L1-

RSRP over all Tx beams with specifically considered Rx beams, i.e., a subset of all Rx beams.

[0074] In BM-Casel, the UE 110 may use a limited set of beam measurements (set B) as the input of an ML model. The ML model may then be used to predict the best beams

N from a bigger (or at least different) set of beams (set A). At least some beams of the set A a 25 — will not be measured by the UE 110. The input to the model may be DL Tx-Rx beam pair = measurements and the output of the model may be the predicted DL Tx-Rx beam pair(s). = With this approach, the UE 110 is able to avoid the lengthy P2-P3 beam refinement

E procedures and determine the best DL Tx and DL Rx beam via the ML-based beam pair & prediction. 2 30 [0075] A schematic overview of such a determination of a transmission beam with a

N machine learning model 400 is depicted in FIG. 4. For an ML-based beam prediction applied, e.g., at the UE 110, it is possible to use the ML model 400 to predict the best DL Tx-Rx beam pair(s) among all or a large set of Tx-Rx beam pairs, also named herein set A, by using a sub-set of measurements of DL Tx-Rx beam pairs, also named herein set B, as the input of the ML model 400.

[0076] The machine learning model 400 receives an input 410 and produces an output 420. The input 410 comprises measurements of DL Tx-Rx beam pairs. This is shown on the left-hand side of FIG. 4. Each matrix 411, 412, 413, and 414 correspond to one Rx beam, e.g., matrix 411 to Rx beam #1, matrix 412 to Rx beam #2, matrix 413 to Rx beam #3, and matrix 414 to Rx beam #4. If the UE 110 provides more Rx beams, there can (but not must) be more Rx beams that are considered. Each position in a matrix correspond to a Tx beam, i.e., all positions/rectangles form a larger set of Tx-Rx beam pairs, which corresponds to set

A The black markings on the left-hand side correspond to those Tx beams that are measured according to a measurement resource set for a respective Rx beam (e.g., RSRP of such a pair) and that are used as input of the machine learning model 400. The black markings, thus, form set B and the input to the machine learning model 400.

[0077] The output 420 of the machine learning model 400 may be a prediction of RSRP of all beam pairs of set A and/or the predicted best beam pair set in set A. The best Tx-Rx beam pair, e.g., that one with the predicted highest RSRP value, predicted highest SINR value, or predicted best signal quality, is indicated with the black marking 421 on the right- hand side of FIG. 4. A selection of the best predicted beam pair, e.g., the one with the estimated highest signal quality or the one estimated to be the most suitable one for communication according to other conditions, may also depend on considering further values, such as a confidence value of the estimation or the like.

[0078] Although the concept of using machine learning for transmission or reception beam prediction, e.g., by predicting a best beam pair, has potential, some issues arise. For & example, at frequency range 2 (FR2) of 5G, there may only be a small number of dominant — directions of departure from the Tx side and a correspondingly small number of arrivals with = respect to the RX side. In this case there is likely only a small number of DL preferred beams = along with a small number of preferred Rx beams for each Tx beam. 7 [0079] Moreover, it may be complicated to design an ML model 400 that can take < random DL Tx-Rx beam pair measurements at the input of the model 400 and have good 2 30 beam prediction accuracy. It may also be complicated for a multi-panel UE 110 to have all & the panels activated at the same time. The UE 110 may require repeating the DL Tx/Rx beam pair measurements corresponding to the DL Tx beam for each panel. It may also be complicated to switch Rx beams in a faster manner at the UE 110 in order to get DL Tx beam pair measurements corresponding to different Rx beams in a single instance.

Therefore, getting the latest measurements related to Set B becomes complicated. The Rx beam information (switching gaps or which beams are used at the input and the output of the model 400) may not be available at the gNB.

[0080] Now turning to Fig. 5 that presents a message signaling diagram of transmission beam determination with beam pair prediction according to agreements of 3GPP, which is faced with the above-mentioned issues. This example starts with message 1 send from a NW 120, e.g., a base station, to a UE 110. In this example, the message is a UE capability inquiry.

The UE 110 then determines in process 2 ML parameters for supporting DL Tx-Rx beam — pair prediction. In process 3 the UE 110 generates a UE capability report for the UE feature group (FG) on FL Tx-Rx beam pair prediction. This UE capability report is then as message 4 transmitted to the NW 120. The UE capability report, hence, comprises a UE FG information element for DL Tx-Rx beam pair prediction.

[0081] The NW 120 then determines in process 5 DL Tx-Rx beam pair prediction related parameters that are transmitted to the UE 110 as message 6, which comprises DL Tx-

Rx beam pair prediction configuration. The DL Tx-Rx beam pair prediction configuration may comprise a measurement resource set (for set B as explained before). Based on the UE capability report of message 4, the NW 120 can optionally further trigger machine learning model monitoring, validation, and testing procedures prior to using the ML model 400 for inference (not shown in Figure 5).

[0082] In further one or more signals 7, the NW 120 transmits the DL Tx beams that correspond to the DL Tx-Rx beam pair prediction configuration. In process 8, the DL beam pair measurements with different Rx beams assumptions, e.g., for different Rx beams, are & performed on the signals 7 at the UE 110, which then executes DL Tx-Rx beam pair a 25 predictions in process 9. This means, the UE 110 determines the best DL Tx-Rx beam pairs, = e.g., those with the best signal guality for communication with the NW 120. This = determination is done with a machine learning model 400 according to the agreements of the 7 3GPP. The UE 110 afterwards transmits message 10 to the NW 120 that comprises a report ce on DL Tx beam indices that correspond to the best DL Tx-Rx beam pair(s) determined in 2 30 process 9. Thereby, the NW 120 is informed about the Tx beam to be used for & communication with the UE 110. Box 11 indicates that the machine learning model 400 as configured is applied further on the UE side.

[0083] As can be seen in FIG. 5, the DL Tx-Rx beam pair prediction configuration is only based on capability of the UE 110 with respect to the machine learning beam prediction.

This leads to the above-mentioned issues, which are overcome by the methods and systems as described with respect to FIGs. 6 to 10.

[0084] FIG. 6 is a flow chart of a method for transmission beam determination at a reception entity according to the disclosure.

[0085] The method starts with transmitting at least one transmission beam condition in box 601. The at least one transmission beam condition is based on a first measurement resource set and/or measurements performed according to the first measurement resource — set. The at least one transmission beam condition is transmitted to a transmission entity.

[0086] In most examples given herein, the reception entity is a user eguipment 110 and the transmission entity is a network apparatus 120. However, the reception entity may also be a network apparatus 120 and the transmission entity a user eguipment 110, the reception entity may be a user eguipment 110 and the transmission entity a user eguipment 110, or the reception entity may be a network apparatus 120 and the transmission entity is a network apparatus 120.

[0087] The first measurement resource set may correspond to the above-mentioned set

A (i.e, that of all possible beam pairs) but may also be a different measurement resource set that is used to determine the transmission beam condition(s). It may, thus, be larger or smaller than the set A. The first measurement resource set is also named set L in this disclosure.

[0088] The method proceeds with receiving a second measurement resource set in box 602. The second measurement resource set may be based on the at least one transmission & beam condition and/or be received from the transmission entity. In other embodiments, there may also be other entities involved such as intermediate transmission entities or core network = entities, which may transmit the second measurement resource set to the reception entity. = The second measurement resource set may correspond to set B. It, hence, is a subset of a

E measurement resource set comprising all possible beam pairs, such as set A. In some 2 embodiments, the reception entity receives more than one second measurement resource set, 2 30 e.g, for different services or transmission conditions.

N [0089] The method then comprises determining at least one transmission beam based on measurements performed on at least one of the at least one second measurement set, which is depicted in box 603. Hence, not all possible beam pairs are measured for determining at least one transmission beam that may be used for communication between the transmission entity and the reception entity but only a subset of beam pairs is measured.

[0090] In embodiments, determining the at least one transmission beam is based on an output of a machine learning model 400 in response to the second measurement resource set being in line with the at least one transmission beam condition. Hence, if the second measurement resource set is in line with the at least one transmission beam condition, a machine learning model 400 for determining a transmission beam can be applied because the previously transmitted transmission beam conditions are fulfilled. In embodiments, this means that the UE 110 reports to the network 120 (e.g., gNB) that it can support ML-based

DL Tx-Rx beam pair prediction (BM-Casel with UE-sided model with DL Tx-Rx beam pairs at the input of the model 400) and includes in this report one or more DL Tx beam related assumptions/conditions, i.e., the transmission beam conditions, to initiate/enable (applicable before model inference) or continue (applicable during the model inference) the beam prediction at the UE side. — [0091] The at least one transmission beam condition may, thus, in some examples be comprised by a reception entity capability report, e.g., a UE capability report, triggered in response to receiving a reception entity capability inquiry, e.g., a UE capability inquiry, from the transmission entity. In other examples, the at least one transmission beam condition may in be comprised by a measurement report triggered in response to receiving a reconfiguration of the first measurement resource set from the transmission entity. In yet other examples, the at least one transmission beam condition may be comprised by a measurement report triggered by the reception entity during application of the machine learning model 400. Of course, the reception entity and the transmission entity may support one or more of these & triggering conditions.

[0092] As indicated above, the at least one transmission beam condition may be a = condition that enables the reception entity to determine at least one transmission beam with = a machine learning model 400. Examples for such conditions are an indication of at least

E one transmission beam preferred by the reception entity, an indication of at least one 2 repetition value preferred by the reception entity, wherein a repetition value is associated 2 30 — with one transmission beam, and/or an indication of at least one switching gap preferred by

N the reception entity, wherein a switching gap relates to a gap between consecutive repetitions of one transmission beam.

[0093] The indication of at least one transmission beam preferred by the reception entity may relate to at least one UE-preferred DL Tx beam to enable DL Tx-Rx beam pair prediction at the UE side. Hence, one DL Tx beam that is, e.g., required by the machine learning model 400 to achieve a good prediction, e.g., relating to the black markings in FIG. 4.

[0094] In embodiments, the at least one transmission beam is indicated by at least one downlink reference signal. The at least one downlink reference signal may in some examples be associated to the first measurement resource set. The first measurement resource set may in some embodiments then be configured during data collection for the machine learning model 400. In some embodiments, the first measurement resource set may be configured during training, validation, testing, fine-tuning, and/or monitoring of the machine learning model 400.

[0095] In other words and in some embodiments, the UE 110 may report one or more

DL RSs (in the form of CRI, SSBRI, and/or in a pre-defined ordering) which are preferred — to be received by the UE 110 (in order to apply the ML model 400). These DL RSs may correspond to a separately configured first measurement set for the UE 110 (in addition to the set A and set B used for model inference, namely set L). Such a separately configured first measurement set may also be considered as part of the data collection process. In further variants, this separately configured first measurement set may be configured as a part of model training, validation, testing, fine-tuning, or model monitoring. In one example, this means that the UE 110 is allowed to report DL RSs that are observed to provide better model performance instead of reporting the best L1-RSRP beams of a previous configured first measurement set. & [0096] In further embodiments, the at least one transmission beam is indicated by a set index relating to a set of downlink reference signals from multiple sets of downlink reference = signals. The UE 110 may, thus, indicate a set of DL RSs (in the form of indicating a set = index) from multiple sets of DL RSs (e.g., selecting a set of DL RSs from pre-configured 7 multiple sets of DL RSs). < [0097] The indication of at least one repetition value preferred by the reception entity 2 30 may relate to an indication of at least one UE-preferred repetition value associated with at

N least one DL Tx beam, wherein the repetition value provides information about the number of repetitions that gNB shall transmit DL Tx beam(s) to enable DL Tx-Rx beam pair prediction at the UE side. Therefore, the repetition value preferred by the reception entity may be associated with a transmission beam preferred by the reception entity and/or a transmission beam to be used by the machine learning model 400.

[0098] The at least one repetition value may be indicated with a bitmap, wherein the bitmap comprises indications relating to one or more sets of downlink reference signals with different repetition values. In some embodiments, the UE 110 may report a bitmap (in the form of CRI/SSBRI, or indices from the preferred DL RSs indicated in the above process), where the bitmap may consist of indications related to a first set of DL RSs may consider having a single repetition, a second set of DL RSs maybe consider having two repetitions, ..., Xth set of DL RSs may be considered to have X repetitions.

[0099] The indication of at least one switching gap preferred by the reception entity may relate to an indication of at least one UE-preferred switching gap between consecutive repetitions of a given DL Tx beam, wherein the switching gap provides information about a minimum time duration that gNB shall consider during repetition of a given DL Tx beam.

[0100] In some embodiments, the at least one switching gap may be associated with at least one of a transmission beam preferred by the reception entity, a transmission beam related to a repetition value preferred by the reception entity, and a transmission beam to be used by the machine learning model 400. That is, the UE-preferred switching gap can be associated with the UE-preferred DL Tx beams, DL beams related to UE-preferred repetition value or any other DL Tx beam being used for the DL Tx-Rx beam pair prediction.

[0101] The at least one switching gap may be indicated by a minimum time duration to be fulfilled for downlink reference signal repetition per downlink reference signal.

Alternatively, the at least one switching gap may be indicated by a minimum time duration to be fulfilled for repetition in transmission of a set of downlink reference signals within one & downlink reception beam. Hence, in one variant, the UE 110 may indicate a minimum time — duration that shall be fulfilled in DL RS repetitions for each reported DL RS (per DL RS); = in another variant, the UE 110 may indicate a minimum time duration that shall be fulfilled = in the transmission of a set of DL RSs (per set of DL RS), where the set contains DL RSs

E that can be received with a one Rx beam assumption (without beam switch) at the UE). 2 [0102] If second measurement resource set is not in line with the at least one 2 30 transmission beam condition, the machine learning model 400 for determining the at least

N one transmission beam, which is applied when the second measurement resource set is in line with the at least one transmission beam condition, may not be applied. Other methods or another machine learning model 400 may then be applied that possibly requires more input, 1.e., measuring more transmission/reception beam pairs, or the like.

[0103] Following the disclosure of FIG. 6, the reception entity may comprise means for transmitting at least one transmission beam condition determined based on a first measurement resource set to a transmission entity, means for receiving a second measurement resource set from the transmission entity, and means for determining at least one transmission beam based on measurements performed according to the second measurement resource set. Other means may be provided as well that are required to execute the methods and processes as described herein.

[0104] Also following the disclosure of FIG. 6, the reception entity comprises a processor and a memory with instructions which, when executed by the processor, cause the reception entity to transmit at least one transmission beam condition determined based on a first measurement resource set to a transmission entity, to receive a second measurement resource set from the transmission entity, and to determine at least one transmission beam — based on measurements performed according to the second measurement resource set. Other instructions may be provided as well that cause the reception entity to execute the methods and processes as described herein.

[0105] FIG. 7 is a flow chart of a method executed by a transmission entity for enabling transmission beam determination according to the disclosure. In box 701, at least one transmission beam condition relating to a first measurement set is received from a reception entity, such as a reception entity, which may execute the method of Fig. 6. And in box 702, at least one second measurement resource set is transmitted to the reception entity. The second measurement resource set is determined based on the at least one transmission beam & condition and may be used according to the method of Fig. 6.

[0106] This means, based on the at least one transmission beam condition, e.g., = comprised by a UE report, the NW 120, e.g., the gNB, uses and/or configures, i.e., indicates = a measurement resource set(s) to the UE (e.g., set B), according to the UE-preferred DL Tx

E beam(s), UE-preferred repetition value(s), and UE-preferred switching gap(s). This enables 2 the UE to execute the ML model 400 for DL Tx-Rx beam pair prediction. Accordingly, if 2 30 changes on measurement resource set(s) are made by the gNB that are not in line with the & reported preferences, the UE 110 may not apply the specific ML model 400 for DL Tx-Rx beam pair prediction.

[0107] Following the disclosure of FIG. 7, the transmission entity may comprise means for receiving at least one transmission beam condition from a reception entity relating to a first measurement resource set, and transmitting a second measurement resource set to the reception entity, wherein the second measurement resource set is determined based on the at least one transmission beam condition. Other means may be provided as well that are required to execute the methods and processes as described herein.

[0108] Also following the disclosure of FIG. 7, the transmission entity may comprise a processor and a memory with instructions which, when executed by the processor, cause the network node to receive at least one transmission beam condition from a reception entity — relating to a first measurement resource set, and transmit a second measurement resource set to the reception entity, wherein the second measurement resource set is determined based on the at least one transmission beam condition. Other instructions may be provided as well that cause the transmission entity to execute the methods and processes as described herein.

[0109] Such an implementation for DL Tx-Rx beam pair prediction with a machine learning is depicted in FIG. 8. The implementation diagram consists of three blocks, which are the same as already depicted FIG. 4, namely, the machine learning model 400, the input 410, and the output 420.

[0110] The input 410 may consist of three-dimensional vectors formed by the measurements of the Tx-Rx beam pairs for different Rx beams. A pattern of Tx beams — measured from one to another Rx beam, as shown in the matrices 411, 412, ..., 41N with the black markings, may be the same or different. As shown in FIG: 8, for Rx beam #1, the measurements correspond to the Tx beams on the diagonal elements of the Tx beam 2D codebook (which is depicted in matrix 411). Then for Rx beam #2 and #N, the measurements & may correspond to the Tx beams for elements of the Tx beam 2D codebook different from — the ones taken for Rx beam #1 as is depicted in matrices 412 and 41N. N may be an integer = greater than 2. For the Tx-Rx beams not measured, the vector is filled with zeros. = [0111] The measurements of the Tx-Rx beam pairs for different Rx beams may have

E the same dimension and can be concatenated together along the third dimension of the input 2 vector. Such a process in shown in box 810. In another example, the history of the Tx-Rx 2 30 beam measurements may be taken into account, and sequences of three-dimensional vectors

N may be stacked together along a fourth dimension representing the time.

[0112] The ML model 400 for DL Tx-Rx beam prediction may have multiple layers, e.g., M layers 801, 802, ..., 80M. The number M may be an integer greater or equal to 0.

Among this, there may be one layer (e.g., dense neural network (DNN), convolutional neural network (CNN)) or multiple layers for one or multiple feedforward branches, which propagate the input data towards the output layer. Different branches may consider processing the input with different settings of parameters. For instance, for convolutional neural network (CNN), layers of parallel branches may operate with different kernel sizes.

Different types of layers may be mixed together, for instance, one or multiple CNN layers may be concatenated to one or more DNN layers.

[0113] In embodiments, when the history of the Tx-Rx beam measurements is considered in the input vector, the ML model 400 for DL Tx-Rx beam prediction may be formed by one or more 2D Convolutional Long short-term memory (LSTM) layers where the input and recurrent transformations are convolutional to extracts features from both the

Tx-Rx pattern of measurements and their variations over time.

[0114] The ML model 400 for DL Tx-Rx beam prediction may be trained from scratch or updated with supervised learning for the prediction of the best Tx/Rx beam pair index (classification task) or for the prediction of the Tx/Rx beam pair RSRP values (regression task). In the first variant (classification task), a soft-max function may be used at the output layer to provide the probability distribution over the complete set of Tx-Rx beam pairs. A selection function as shown in box 820 is then utilized for sorting and providing the index of the best Tx-Rx beam pair 421 or the indexes of the top-K Tx-Rx beam pairs. In another — variant (regression task), the selection function 820 may be used to sort the predicted RSRP values of the Tx-Rx beam pairs and provide the index of the best Tx-Rx beam pair 421 or the indexes of the top-K Tx-Rx beam pairs. The above machine learning models are mere examples. Any other method that adapts similarly, for example, for the above mentioned

O linear or non-linear regression, or for a classification task, may also be used as machine a 25 — learning model 400 according to the current application. = [0115] FIG. 9 shows a message signaling diagram of determination of a transmission = beam triggered by a user equipment capability inquiry according to an embodiment.

E [0116] After the NW 120 (e.g., gNB) inguires for a UE 110 capability on DL Tx-Rx ce beam prediction with message 1, the UE 110 then determines in process 2 ML parameters 2 30 for supporting DL Tx-Rx beam pair prediction. Afterwards, the generation of the UE

N capability report may contain one or more of the further processes 4, 5, and 6. The UE 110 may consider a default measurement set L to identify the DL Tx beams as shown in process 3. In one example, a CSI resource configuration may be defined to the UE 110 such that it can associate the capability reporting with it. The UE 110 may use set L when it derives the indexing for a preferred DL Tx beam (e.g., CRI reporting considering the set L).

[0117] In process 4, the UE 110 determines the UE-preferred-DL Tx beam(s) by considering the applicable DL Tx-Rx patterns at the machine learning model 400, wherein applicability may be determined by considering which patterns were used for model training and/or fine-tuning and/or updating. In process 5, the UE 110 determines the UE-preferred- repetition value(s) by considering the applicable DL Tx-Rx patterns at the machine learning model 400, wherein applicability may be determined by considering which patterns were used for model training and/or fine-tuning and/or updating. In one example, one DL Tx beam may be considered more than one time (received with different Rx beam assumptions), while another DL Tx beam may not be considered more than one time (only one Rx may be applied). So, the repetition values could be different for different DL Tx beams. In process 6, the UE 110 determines the UE-preferred-switching-gap(s) by considering the applicable

DL Tx-Rx patterns at the machine learning model 400, wherein applicability may be — determined by considering which patterns were used for model training and/or fine-tuning and/or updating. In one example, if one DL Tx beam is associated with two Rx beams, there will be a switching gap associated with that. In another variant, DL Tx beams may be grouped and indicated in the group level where a given Rx beam may receive the group of

DL Tx beams. Within a group of Tx beams, the switching delay may be considered zero.

[0118] The above aspects are considered in process 9 relating to the generation of a UE capability report. Thee UE capability report is in message 8 transmitted back to the NW 120 such that NW 120 could use it to set proper set B as well as configurations for set A for the

UE 110, which are transmitted back in message 10. Messages and processes 11 — 15 & correspond to the messages and processes 7 to 11 in FIG. 5. — [0119] FIG. 10 shows a message signaling diagram of determination of transmission = beam conditions triggered by a new configuration of a first measurement set. In this example, = the machine learning model 400 is already applied for DL Tx-Rx beam pair prediction as

E depicted with box 1. The additional procedures in the figure are mainly for optimizing the co RS overhead, improving model performance, enabling switching and/or enabling updating 2 30 the model 400.

N [0120] In process 2, the NW 120 (e.g., gNB) determines a measurement resource set L as shown with process 2, configures it to the UE 110 with message 3, and optionally transmit the DL RSs associated with set L as shown with arrow 4. This set L is different from set A and set B, and is used mainly to identify the DL Tx beams. In process 5, the UE 110 considers the new set L for determining the UE-preferred-DL Tx beam(s), UE-preferred-repetition value(s), and UE-preferred-switching-gap(s) in processes 6, 7, and 8, which are similar to the processes 4, 5, and 6 explained with respect to FIG. 9.

[0121] In process 9, the UE 110 prepares a report according to reporting configuration associated with the measurement set L in the UL resources scheduled/configured by the NW 120, resulting in message 10. The NW 120 considers in process 11 the reported details when adjusting the inference parameters of the machine learning model 400.

[0122] FIG. 11 shows a message signaling diagram of determination of transmission beam conditions triggered during application of the machine learning model. In this example, the machine learning model 400 is already applied for DL Tx-Rx beam pair prediction, i.e., with a configured measurement set B and predication set A, as depicted with box 1. This is similar to box 1 of FIG. 10.

[0123] In process 2, the UE 110 that uses the machine learning model 400 for inference, — performs model monitoring (performance monitoring to identify model degradations or the like). In process 3, it is assumed that the UE 110 determines that the performance of machine learning model 400 is poor, e.g, a performance identifier falls below a threshold, the inference fails to predict the best beam pairs, or the model needs more processing resources.

Then, there is a need of updating the model parameters (e.g., DL Tx-Rx beam pairs used as the input 410).

[0124] In process 4, the UE 110 triggers an update to the model parameters by sending a message to the NW 120, e.g., a MAC-CE message which is dedicated for this purpose.

Processes/messages 5 to 14 are similar to the processes/messages 2 to 11 of FIG. 10. & [0125] Processes 4 to 7 are not mandatory. For example, processes/messages 4 to 7 may — be executed if a new set L is needed or if an update of set L is considered to be senseful. = Hence, network 120 may then determine a measurement Set L (process 5), configure it to = the UE (message 6), and optionally transmit the DL RSs associated with set L (message 7). 7 In another example, processes 4 to 7 may not be needed, e.g, if set L is already known to ce the UE 110 and/or no need for update is determined. In such cases, the UE-initiated report 2 30 as shown as message 13 may include the information of message 4.

N [0126] The herein described procedures may be applied per model or per functionality level (identified by an ID) or across models or functionalities of a given entity, e.g., UE feature. It should be understood that the apparatuses described herein may comprise or be coupled to other units or modules etc., such as radio parts or radio heads, used in or for transmission and/or reception. Although the apparatuses have been described as one entity, different modules and memory may be implemented in one or more physical or logical entities.

[0127] It is noted that whilst embodiments have been described in relation to LTE and 5G NR, similar principles can be applied in relation to other networks and communication systems where enforcing fast connection re-establishment is reguired. Therefore, although certain embodiments were described above by way of example with reference to certain example architectures for wireless networks, technologies and standards, embodiments may be applied to any other suitable forms of communication systems than those illustrated and described herein.

[0128] It is also noted herein that while the above describes exemplary embodiments, there are several variations and modifications which may be made to the disclosed solution without departing from the scope of the subject disclosure.

[0129] In general, the various exemplary embodiments may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. Some aspects of the subject disclosure may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although the subject disclosure is not limited thereto. While various aspects of the subject disclosure may be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, technigues or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special & purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof. = [0130] Example embodiments of the subject disclosure may be implemented by = computer software executable by a data processor of the mobile device, such as in the

E processor entity, or by hardware, or by a combination of software and hardware. Computer 2 software or program, also called program product, including software routines, applets 2 30 and/or macros, may be stored in any apparatus-readable data storage medium and they

N comprise program instructions to perform particular tasks. A computer program product may comprise one or more computer- executable components which, when the program is run,

are configured to carry out embodiments. The one or more computer-executable components may be at least one software code or portions of it.

[0131] Further in this regard it should be noted that any blocks of the logic flow as in the figures may represent program processes, or interconnected logic circuits, blocks and functions, or a combination of program processes and logic circuits, blocks and functions.

The software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example DVD and the data variants thereof, CD. The physical media is a non-transitory media.

[0132] The memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor- based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory. The data processors may be of any type suitable to the local technical environment, and may comprise one or more of general- — purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASICs), FPGA, gate level circuits and processors based on multi-core processor architecture, as non-limiting examples.

[0133] Example embodiments of the subject disclosure may be practiced in various components such as integrated circuit modules. The design of integrated circuits is by and large a highly automated process. Complex and powerful software tools are available for converting a logic level design into a semiconductor circuit design ready to be etched and formed on a semiconductor substrate.

[0134] The foregoing description has provided by way of non-limiting examples a full & and informative description of the exemplary embodiment of the subject disclosure.

However, various modifications and adaptations may become apparent to those skilled in = the relevant arts in view of the foregoing description, when read in conjunction with the = accompanying drawings and the appended claims. However, all such and similar

E modifications of the teachings of this disclosure will still fall within the scope of the subject 2 disclosure as defined in the appended claims. Indeed, there is a further embodiment 2 30 comprising a combination of one or more embodiments with any of the other embodiments

N previously discussed.

Claims

CLAIMS:

1. A method executed by a reception entity comprising: — transmitting (601) at least one transmission beam condition determined based on a first measurement resource set to a transmission entity; — receiving (602) a second measurement resource set; and — determining (603) at least one transmission beam based on measurements performed according to the second measurement resource set.

2. The method of claim 1, wherein, in response to the second measurement resource set being in line with the at least one transmission beam condition, the determining (603) of the at least one transmission beam is based on an output of a machine learning model.

3. The method of claim 2, wherein the at least one transmission beam condition is comprised by a reception entity capability report triggered in response to receiving a reception entity capability inquiry from the transmission entity, by a measurement report triggered in response to receiving a reconfiguration of the first measurement resource set from the transmission entity, or by a measurement report triggered by the reception entity during application of the machine learning model.

4. The method of claim 2 or claim 3, wherein the at least one transmission beam condition D comprises an indication of at least one downlink transmission beam preferred by the N reception entity. 2 3 I 25 5. The method of claim 4, wherein the at least one downlink transmission beam preferred > by the reception entity is indicated by indicating at least one downlink reference signal = preferred to be received by the reception entity. &

&

6. The method of claim 5, wherein the at least one downlink reference signal is associated to the first measurement resource set.

7. The method of claim 6, wherein the first measurement resource set is configured during data collection for the machine learning model.

8. The method of claim 6 or claim 7, wherein the first measurement resource set is configured during training, validation, testing, fine-tuning, and/or monitoring of the machine learning model.

9. The method of claim 8, wherein the at least one downlink transmission beam is indicated by a set index relating to a set of downlink reference signals from multiple sets of downlink reference signals.

10. The method of any one of claims 2 to 9, wherein the at least one transmission beam condition comprises an indication of at least one repetition value preferred by the reception entity, wherein a repetition value is associated with one transmission beam.

11. The method of claim 10, wherein the at least one repetition value is associated with a transmission beam preferred by the reception entity and/or a transmission beam to be used by the machine learning model. &

12. The method of claim 10 or claim 11, wherein the at least one repetition value is a indicated with a bitmap, wherein the bitmap comprises indications relating to one or 3 more sets of downlink reference signals with different repetition values. O E LO 25

13. The method of any one of claims 2 to 12, wherein the at least one transmission beam = condition comprises an indication of at least one switching gap preferred by the S reception entity, wherein a switching gap relates to a gap between consecutive repetitions of one transmission beam.

14. The method of claim 13, wherein the at least one switching gap is associated with at least one of a transmission beam preferred by the reception entity, a transmission beam related to a repetition value preferred by the reception entity, and a transmission beam to be used by the machine learning model.

15. The method of claim 13 or claim 14, wherein the at least one switching gap is indicated by a minimum time duration to be fulfilled for downlink reference signal repetition per downlink reference signal.

16. The method of claim 13 or claim 14, wherein the at least one switching gap is indicated by a minimum time duration to be fulfilled for repetition in transmission of a set of downlink reference signals within one downlink reception beam.

17. The method of any one of claims 2 to 16 further comprising, in response to the second measurement resource set being not in line with the at least one transmission beam condition, not applying the machine learning model for determining (603) the at least one transmission beam.

18. A reception entity comprising a processor and a memory with instructions which, when executed by the processor, cause the reception entity to: — transmit (601) at least one transmission beam condition determined based on a first measurement resource set to a transmission entity; & — receive (602) a second measurement resource set from the transmission entity; a and 3 25 — determine (603) at least one transmission beam based on measurements = performed according to the second measurement resource set. a a © o

19. The reception entity of claim 18 being further caused to execute the method of any one N of claims 2 to 17. N

20. A method executed by a transmission entity comprising:

— receiving (701) at least one transmission beam condition from a reception entity relating to a first measurement resource set; and — transmitting (702) a second measurement resource set to the reception entity, wherein the second measurement resource set is determined based on the at least one transmission beam condition.

21. A transmission entity comprising a processor and a memory with instructions which, when executed by the processor, cause the transmission entity to: — receive (701) at least one transmission beam condition from a reception entity relating to a first measurement resource set; and — transmit (702) a second measurement resource set to the reception entity, wherein the second measurement resource set is determined based on the at least one transmission beam condition.

22.A non-transitory computer readable medium comprising program instructions that, when executed by an apparatus, cause the apparatus to execute the method of any one of claims 1 to 17 or the method of claim 20. ™ N O N O <t O I [an a LO 00 LO 0 N O N