Beam management using ai / ML approaches
AI/ML techniques are employed to enhance beam management in wireless communication systems by optimizing beamforming and resource allocation, addressing inefficiencies in existing methods and improving network performance.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- NANJING XINXUNBIAO TECHNOLOGY CO LTD
- Filing Date
- 2025-01-17
- Publication Date
- 2026-07-23
Smart Images

Figure CN2025073001_23072026_PF_FP_ABST
Abstract
Description
Beam Management Using AI / ML ApproachesBRIEF DESCRIPTION OF THE DRAWINGS
[0001] Examples of several of the various embodiments of the present disclosure are described herein with reference to the drawings.
[0002] FIG. 1 illustrates an example of a mobile communication network in which embodiments of the present disclosure may be implemented.
[0003] FIG. 2A and FIG. 2B respectively illustrate a NR user plane and a NR control plane protocol stack.
[0004] FIG. 3 illustrates an example configuration of an NR frame and an example configuration of a slot in the time and frequency domain for an NR carrier.
[0005] FIG. 4 illustrates an example of bandwidth adaptation using three configured BWPs for an NR carrier as per an aspect of an example embodiment of the present disclosure.
[0006] FIG. 5 illustrates three carrier aggregation configurations with two component carriers as per an aspect of an example embodiment of the present disclosure.
[0007] FIG. 6 illustrates an example of a wireless device in communication with a base station in accordance with embodiments of the present disclosure.
[0008] FIG. 7A illustrates an example of a burst of SSBs as per an aspect of an example embodiment of the present disclosure.
[0009] FIG. 7B illustrates an example of CSI-RSs that are mapped in the time and frequency domains as per an aspect of an example embodiment of the present disclosure.
[0010] FIG. 8A, FIG. 8B and FIG. 8C illustrate examples of downlink beam management procedures as per an aspect of an example embodiment of the present disclosure.
[0011] FIG. 9 illustrates an example of a functional framework for AI / ML approaches for wireless communications as per an aspect of an example embodiment of the present disclosure.
[0012] FIG. 10 illustrates an example of beam prediction and reporting using AI / ML approaches as per an aspect of an example embodiment of the present disclosure.
[0013] FIG. 11 illustrates an example of beam prediction and inference reporting when an inference report comprises predicted beams and / or corresponding RSRPs for a plurality of time instances as per an aspect of an example embodiment of the present disclosure.
[0014] FIG. 12A and FIG. 12B illustrate examples of a single inference report indicating predicted beams and / or corresponding RSRPs for a plurality of time instances as per an aspect of an example embodiment of the present disclosure.
[0015] FIG. 13A and FIG. 13B illustrate example timing for reference signal measurement and prediction using AI / ML approaches as per an aspect of an example embodiment of the present disclosure.DETAILED DESCRIPTION
[0016] In the present disclosure, various embodiments are presented as examples of how the disclosed techniques may be implemented and / or practiced in environments and scenarios. A person having ordinary skills in the art will readily recognize that the teachings of the example embodiments in the present disclosure can be applied in a multitude of different ways. Some or all of the described examples may be implemented in any device, system or network that is capable of transmitting and receiving radio frequency (RF) signals according to the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standards, IEEE 802.15 standards, the Bluetooth standards, or the Long Term Revolution (LTE) , 3rd generation (3G) , fourth generation (4G) , fifth generation (5G) new radio (NR) standards, and those of future networks yet to be specified (e.g., a 3GPP 6G network) , or any others. The embodiments of the present disclosure will be described with reference to accompanying drawings. Limitations, features, and / or elements from the disclosed example embodiments may be combined to create further embodiments within the scope of the disclosure.
[0017] In the present disclosure, “a” , “an” , and any term that ends with the suffix “ (s) ” should be interpreted as “at least one” and “one or more. ” In this disclosure, the term “may” should be interpreted as “may, for example. ” In other words, a phrase following the term “may” is an example of one of a multitude of suitable possibilities that may, or may not, be employed by one or more of the various embodiments. The term “comprises” is interchangeable with “includes” and does not exclude unenumerated components from being included in the element being described. By contrast, “consists of” provides a complete enumeration of the one or more components of the element being described. The term “based on” , as used herein, should be interpreted as “based at least in part on” rather than, for example, “based solely on” . The term “and / or” as used herein represents any possible combination of enumerated elements. For example, “A, B, and / or C” may represent A; B; C; A and B; A and C; B and C; or A, B, and C.
[0018] In the present disclosure, the term base station may refer to and encompass a Node B (associated with UMTS and / or 3G standards) , an Evolved Node B (eNB, associated with E-UTRA and / or 4G standards) , a remote radio head (RRH) , a baseband processing unit coupled to one or more RRHs, a repeater node or relay node used to extend the coverage area of a donor node, a Next Generation Evolved Node B (ng-eNB) , a Generation Node B (gNB, associated with NR and / or 5G standards) , an access point (AP, associated with, for example, WiFi or any other suitable wireless communication standard) , and / or any combination thereof. A base station may comprise at least one gNB Central Unit (gNB-CU) and at least one gNB Distributed Unit (gNB-DU) .
[0019] In the present disclosure, the term wireless device may refer to and encompass any mobile device or fixed (non-mobile) device for which wireless communication is desired or usable. For example, a wireless device may be a telephone, smart phone, tablet, computer, laptop, sensor, meter, wearable device, Internet of Things (IoT) device, vehicle road side unit (RSU) , relay node, automobile, electric vehicle (EV) charger, extended reality (XR) glasses / goggles, and / or any combination thereof. The term wireless device encompasses other terminology, including user equipment (UE) , user terminal (UT) , access terminal (AT) , mobile station, handset, wireless transmit and receive unit (WTRU) , and / or wireless communication device.
[0020] In the present disclosure, the term configured may refer to specific settings in a device that affect the operational characteristics of the device whether the device is in an operational or non-operational state. For example, hardware, software, firmware, registers, memory values, and / or the like may be “configured” within a device, whether the device is in an operational or nonoperational state, to provide the device with specific characteristics. Terms such as “acontrol message to cause in a device” may mean that a control message has parameters that may be used to configure specific characteristics or may be used to implement certain actions in the device, whether the device is in an operational or non-operational state.
[0021] In the present disclosure, parameters (or equally called, fields, or Information elements (IEs) ) may comprise one or more information objects. In an example, the parameters may comprise the one or more information objects in a nested way. For example, a first parameter (e.g., a first IE) may comprise a second parameter (e.g., a second IE) and a third parameter (e.g., a third IE) , when the first parameter comprises the second parameter and the second parameter comprises the third parameter. In an example embodiment, one or more messages comprise a plurality of parameters may imply that a parameter of the plurality of parameters is in at least one of the one or more messages, but does not have to be in each of the one or more messages.
[0022] FIG. 1 illustrates an example of a mobile communication network 170 in which embodiments of the present disclosure may be implemented. The mobile communication network 170 may be, for example, a public land mobile network (PLMN) run by a network operator. As illustrated in FIG. 1, the mobile communication network 170 may comprise one or more wireless devices (e.g., a user equipment (UE) 100 and / or a UE 105) , a radio access network (RAN) (e.g., a base station 110 and / or a base station 115) , and a core network (CN) .
[0023] The CN may provide the one or more wireless devices with an interface to one or more data networks (DNs) (e.g., DNs 130) , such as public DNs (e.g., the Internet) , private DNs, and / or intra-operator DNs. As part of the interface functionality, the CN may set up end-to-end connections between the one or more wireless devices and the one or more DNs, authenticate the one or more wireless devices, and provide charging functionality. In an example of a new radio (NR) system in FIG. 1, a CN may comprise an Access and Mobility Management Function (AMF) and a User Plane Function (UPF) , which are shown as one component AMF / UPF 120 in FIG. 1 for ease of illustration. The UPF may serve as a gateway between the RAN and the one or more DNs. The UPF may perform functions such as packet routing and forwarding, packet inspection and user plane policy rule enforcement, traffic usage reporting, uplink classification to support routing of traffic flows to the one or more DNs, quality of service (QoS) handling for the user plane (e.g., packet filtering, gating, uplink / downlink rate enforcement, and uplink traffic verification) , downlink packet buffering, and downlink data notification triggering. The AMF may perform functions such as Non-Access Stratum (NAS) signaling termination, NAS signaling security, Access Stratum (AS) security control, inter-CN node signaling for mobility between 3GPP access networks, idle mode UE reachability (e.g., control and execution of paging retransmission) , registration area management, intra-system and inter-system mobility support, access authentication, access authorization including checking of roaming rights, mobility management control (subscription and policies) , network slicing support, and / or session management function (SMF) selection. NAS may refer to the functionality operating between a CN and a UE, and AS may refer to the functionality operating between the UE and a RAN.
[0024] The RAN may connect the CN to the one or more wireless devices through radio communications over an air interface. As part of the radio communications, the RAN may provide scheduling, radio resource management, and retransmission protocols. A communication direction from the RAN to the one or more wireless devices (e.g., from the base station 110 to the UE 100) over the air interface is known as a downlink. A communication direction from the one or more wireless devices (e.g., from the UE 100 to the base station 110) to the RAN over the air interface is known as an uplink. The downlink transmissions may be separated from the uplink transmissions using frequency division duplexing (FDD) , time-division duplexing (TDD) , and / or some combination of the two duplexing techniques. In an example of a new radio (NR) system in FIG. 1, the RAN may comprise one or more base stations. A UE of the one or more wireless devices (e.g., the UE 100 and the UE 105) and a base station of the one or more base stations (e.g., the base station 110 and the base station 115) may be connected by means of a Uu interface (e.g., an interface 140 in FIG. 1) . A first base station of the one or more base stations (e.g., the base station 110) and a second base station of the one or more base stations (e.g., the base station 115) may be connected by means of a Xn interface (e.g., an interface 150 in FIG. 1) . A base station of the one or more base stations in the RAN and the AMF / UPF in the CN may be connected by means of a NG interface (e.g., an interface 160 in FIG. 1) .
[0025] FIG. 2A illustrates a NR user plane protocol stack comprising five layers implemented in a UE 210 and a gNB 220. The NR user plane protocol stack comprises physical layers (PHYs) 211 and 221 providing transport services to the higher layers of the protocol stack and may correspond to layer 1 of the Open Systems Interconnection (OSI) model. The NR user plane protocol stack comprises media access control (sub) layers (MACs) 212 and 222, radio link control (sub) layers (RLCs) 213 and 223, packet data convergence protocol (sub) layers (PDCPs) 214 and 224, and service data application protocol (sub) layers (SDAPs) 215 and 225. The MACs (e.g., 212 and 222) , the RLCs (e.g., 213 and 223) , the PDCPs (e.g., 214 and 224) , and the SDAPs (e.g., 215 and 225) may make up layer 2 (e.g., or be known as sublayers of the layer 2) , or the data link layer, of the OSI model.
[0026] FIG. 2B illustrates an example NR control plane protocol stack. As shown in FIG. 2B, the NR control plane protocol stack comprises the PHY layers 211 and 221. The NR control plane protocol stack has the MAC (sub) layers 212 and 222, the RLC (sub) layers 213 and 223, the PDCP (sub) layers 214 and 224 (e.g., terminated in the gNB 220 on network side) . The NR control plane protocol stack has radio resource controls (RRCs) 216 and 226 (e.g., terminated in the gNB 220 on network side) . The NR control protocol stack has NAS control protocols 217 and 237 (e.g., terminated in an AMF 230 on network side) performs the functions, for instance: authentication, mobility management, security control, etc.
[0027] FIG. 3 illustrates an example configuration of an NR frame. Orthogonal frequency divisional multiplexing (OFDM) symbols are grouped into the NR frame. In NR, physical signals and physical channels may be mapped onto OFDM symbols. An NR frame may be identified by a system frame number (SFN) . The SFN may repeat with a period of 1024 frames. As illustrated, one NR frame may be 10 milliseconds (ms) in duration and may include 10 subframes that are 1 ms in duration. A subframe may be divided into slots that include, for example, 14 OFDM symbols per slot.
[0028] The duration of a slot may depend on the numerology used for the OFDM symbols of the slot. In NR, a flexible numerology is supported to accommodate different cell deployments (e.g., cells with carrier frequencies below 1 GHz up to cells with carrier frequencies in the mm-wave range) . A numerology may be defined in terms of subcarrier spacing and cyclic prefix duration. For a numerology in NR, subcarrier spacings may be scaled up by powers of two from a baseline subcarrier spacing of 15 kHz, and cyclic prefix durations may be scaled down by powers of two from a baseline cyclic prefix duration of 4.7 μs. For example, NR defines numerologies with the following subcarrier spacing / cyclic prefix duration combinations: 15 kHz / 4.7 μs; 30 kHz / 2.3 μs; 60 kHz / 1.2 μs; 120 kHz / 0.59 μs; and 240 kHz / 0.29 μs.
[0029] FIG. 3 further illustrates an example configuration of a slot in the time and frequency domain for an NR carrier. The slot includes resource elements (REs) and resource blocks (RBs) . An RE is the smallest physical resource (e.g., physical radio resource) in NR. In FIG. 3, an RE spans 1 OFDM symbol in the time domain by 1 subcarrier in the frequency domain. An RB spans 12 consecutive REs (e.g., 12 subcarriers) in the frequency domain, and one slot (e.g., 14 consecutive OFDM symbols) in the time domain. FIG. 3 illustrates an example of a single numerology being used across the entire bandwidth of the NR carrier. In other example configurations, multiple numerologies may be supported on the same carrier.
[0030] FIG. 4 illustrates an example of bandwidth adaptation (BA) using three configured BWPs for an NR carrier (e.g., a primary cell and / or a secondary cell of a UE) . With bandwidth adaptation, a transmit and / or receive bandwidth of a UE may not be as large as a bandwidth of a cell. The transmit and / or receive bandwidth of the UE may be adjusted. In an example, a width of the transmit and / or receive bandwidth may be ordered to change (e.g., to save power of the UE) . In an example, a location of the transmit and / or receive bandwidth may move in the frequency domain (e.g. to increase scheduling flexibility of the UE) . In an example, a subcarrier spacing (e.g., numerology) of the transmit and / or receive bandwidth may be ordered to change (e.g. to allow different services of the UE) . In an example, a subset of a cell bandwidth of the cell may be referred to as a Bandwidth Part (BWP) . For achieving bandwidth adaptation, one or more BWPs may be configured and / or pre-configured to a UE. Network (e.g., a base station) may indicate the UE which of the configured / pre-configured one or more BWPs is currently an active BWP.
[0031] In an example of FIG. 4, a UE may be configured with three BWPs. The BWPs comprise a BWP 402 with a bandwidth of 40 MHz and a subcarrier spacing of 15 kHz, a BWP 404 with a bandwidth of 10 MHz and a subcarrier spacing of 15 kHz, and a BWP 406 with a bandwidth of 20 MHz and a subcarrier spacing of 60 kHz. The UE may switch from a first BWP to a second BWP at a switching point, for example, based on an expiry of a BWP inactivity timer (e.g., indicating switching to a default BWP) and / or in response to receiving a downlink control information (DCI) indicating the second BWP as an active BWP. In the example of FIG. 4, the UE may switch from the BWP 402 to the BWP 404 at a switching point 408 based on an expiry of a BWP inactivity timer (e.g., indicating switching to a default BWP) , when the BWP 402 is an initial active BWP and the BWP 404 is a default BWP. In another example of FIG. 4, the UE may switch from the BWP 402 to the BWP 404 at a switching point 408 in response to receiving a DCI indicating the BWP 404 as an active BWP. The UE may switch, at a switching point 410, from the BWP 404 to the BWP 406 in response receiving a DCI indicating the BWP 406 as an active BWP. The UE may switch, at a switching point 412, from the BWP 406 to the BWP 404 in response to an expiry of a BWP inactivity timer and / or in response receiving a DCI indicating the BWP 404 as an active BWP. The UE may switch, at a switching point 414, from the BWP 404 to the BWP 402 in response to receiving a DCI indicating the BWP 402 as an active BWP.
[0032] FIG. 5 illustrates three carrier aggregation (CA) configurations with two component carriers (CCs) . Carrier aggregation for a UE may be supported for providing greater data rates. Two or more carriers can be aggregated and simultaneously transmitted to / from the same UE using carrier aggregation. The aggregated carriers in CA may be referred to as component carriers. When CA is used, there are a number of serving cells for the UE, one for a CC. The CCs may have three configurations in the frequency domain.
[0033] As shown in FIG. 5, in a first configuration for intraband CA with contiguous CCs 500, two contiguous CCs are aggregated in the same frequency band (frequency band A) and are located adjacent to each other within the frequency band. In a second configuration for intraband CA with non-contiguous CCs 510, two CCs are aggregated in the same frequency band (frequency band A) and are separated in the frequency band by a gap. In a third configuration for interband CA 520, two CCs are located in frequency bands (frequency band A and frequency band B) . In an example, up to 32 CCs may be aggregated. The aggregated CCs may have the same or different bandwidths, subcarrier spacing, and / or duplexing schemes (TDD or FDD) . A serving cell for a UE using CA may have a downlink CC. For FDD, one or more uplink CCs may be optionally configured for a serving cell.
[0034] When CA is configured to a wireless device with a plurality of aggregated cells, a first serving cell of the plurality of aggregated cells may be referred to as a Primary Cell (PCell) for the wireless device. The PCell may be the serving cell that the wireless device initially connects to. The PCell may provide, to the wireless device, NAS mobility function at RRC connection establishment / re-establishment / handover. The PCell may provide, to the wireless device, security input at RRC re-establishment / handover. A second serving cell of the plurality of aggregated cells may be referred to as a Secondary Cell (SCell) for the wireless device. The PCell and the SCell of the wireless device are different serving cells. Depending on UE capabilities of the wireless device, one or more SCells may be configured, to the wireless device, with the PCell to form a set of serving cells for the wireless device.
[0035] FIG. 6 illustrates an example of a wireless device 650 in communication with a base station 600 in accordance with embodiments of the present disclosure. The wireless device 650 and base station 600 may be part of a mobile communication network, such as the mobile communication network 170 illustrated in FIG. 1 or any other communication network. The base station 600 may connect the wireless device 650 to a core network (not shown) through radio communications over an air interface (or radio interface) . The communication direction from the base station 600 to the wireless device 650 over the air interface is known as a downlink, and the communication direction from the wireless device 650 to the base station 600 over the air interface is known as an uplink.
[0036] In the downlink, downlink data to be sent, to the wireless device 650 from the base station 600, may be provided to a processor / controller 625 of the base station 600. The downlink data may be provided to the processor / controller 625 by, for example, a core network. The processor / controller 625 may implement layer 3 and layer 2 OSI functionality to process the downlink data for transmission. Layer 3 may include an RRC layer. Layer 2 may include an SDAP layer, a PDCP layer, an RLC layer, and a MAC layer. The processor / controller 625 may provide RRC layer functionality associated with broadcasting of system information (e.g., master information block (MIB) , system information blocks (SIBs) ) , RRC connection control (e.g., RRC connection paging, RRC connection establishment, RRC connection modification, and RRC connection release) , inter radio access technology (RAT) mobility, and measurement configuration for UE measurement reporting; PDCP layer functionality associated with header compression / decompression, security (ciphering, deciphering, integrity protection, integrity verification) , and handover support functions; RLC layer functionality associated with the transfer of upper layer packet data units (PDUs) , error correction through Automatic Repeat reQuest (ARQ) , concatenation, segmentation, and reassembly of RLC service data units (SDUs) , re-segmentation of RLC data PDUs, and reordering of RLC data PDUs; and MAC layer functionality associated with mapping between logical channels and transport channels, multiplexing of MAC SDUs onto transport blocks (TBs) , demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction through Hybrid ARQ (HARQ) , priority handling, and logical channel prioritization.
[0037] In the downlink and after being processed by the processor / controller 625, the downlink data may be provided to one or more transmit (Tx) processors 620 of base station 600. The one or more Tx processors 620 may implement layer 1 OSI functionality. Layer 1 may comprise a PHY layer. The PHY layer may perform, for example, forward error correction (FEC) coding of transport channels, interleaving, rate matching, mapping of transport channels to physical channels, modulation of physical channel, multiple-input multiple-output (MIMO) or multi-antenna processing, and / or the like. The one or more Tx processors 620 may map the downlink data to coded and modulated symbols based on various modulation schemes (e.g., binary phase-shift keying (BPSK) , quadrature phase-shift keying (QPSK) , M-phase-shift keying (M-PSK) , M-quadrature amplitude modulation (M-QAM) ) and coding schemes (e.g., convolution code, turbo code, and / or low density parity check (LDPC) code) . The coded and modulated symbols of the downlink data may be split into parallel streams. Each stream of the parallel streams may be mapped to an OFDM subcarrier, multiplexed with a reference signal (e.g., pilot) in the time and / or frequency-domain, and combined using an Inverse Fast Fourier Transform (IFFT) to produce a physical channel carrying a time-domain OFDM symbol stream. The OFDM stream may be spatially precoded to produce multiple spatial streams. A channel estimator (not shown in FIG. 6) may be used to determine the modulation schemes, the coding schemes, and / or for spatial processing. Each spatial stream of the multiple spatial streams may be provided to different antenna via a separate transmitter Tx 635. Each transmitter Tx 635 may modulate a radio frequency (RF) carrier with a respective spatial stream for transmission.
[0038] In the downlink and at the wireless device 650, each of the one or more receivers Rx 690 receives a signal through respective antenna. The one or more receivers Rx 690 may recover received information, modulated onto the RF carrier, and provide the received information to one or more receive (Rx) processors 680 of the wireless device 650. The one or more Rx processors 680 may implement layer 1 OSI functionality. The one or more Rx processors 680 may perform spatial processing on the received information to recover one or more spatial streams desired for the wireless device 650. If multiple spatial streams are desired for the wireless device 650, the one or more Rx processors 680 may combine the multiple spatial streams into a single OFDM symbol stream. The one or more Rx processors 680 may convert the OFDM symbol stream from the time-domain to the frequency-domain using a Fast Fourier Transform (FFT) . The OFDM symbol stream may be demodulated and decoded, to recover the downlink data and / or control signal transmitted by the base station 600 on physical channel, based on a channel estimator at the wireless device 650 (not shown in FIG. 6) . The data and control signals are then provided to the processor / controller 675, which implements layer 3 and layer 2 functionality.
[0039] In the uplink, uplink data to be sent, from the wireless device 650 to the base station 600, may be provided to the processor / controller 675 of the wireless device 650. Similar to the downlink, the processor / controller 675 may implement layer 3 and layer 2 OSI functionality to process the uplink data for transmission.
[0040] In the uplink and after being processed by the processor / controller 675, the uplink data may be provided to one or more Tx processors 670 of the wireless device 650. Similar to the downlink, the one or more Tx processors 670 may implement layer 1 OSI functionality. The one or more Tx processors 670 may map the uplink data to coded and modulated symbols based on various modulation schemes and coding schemes. The coded and modulated symbols of the uplink data may be split into parallel streams. Each stream of the parallel streams may be mapped to an OFDM subcarrier, multiplexed with a reference signal (e.g., pilot) in the time and / or frequency-domain, and combined using an IFFT to produce a physical channel carrying a time-domain OFDM symbol stream. The OFDM stream may be spatially precoded to produce multiple spatial streams. A channel estimator (not shown in FIG. 6) may be used to determine the modulation schemes, the coding schemes, and / or for spatial processing. Each spatial stream of the multiple spatial streams may be provided to different antenna via a separate transmitter Tx 685. Each transmitter Tx 685 may modulate a RF carrier with a respective spatial stream for transmission.
[0041] In the uplink and at the base station 600, each of the one or more receivers Rx 640 receives a signal through respective antenna. The one or more receivers Rx 640 may recover received information, modulated onto the RF carrier, and provide the received information to one or more Rx processors 630 of the base station 600. The one or more Rx processors 630 may implement layer 1 OSI functionality. The one or more Rx processors 630 may perform spatial processing on the received information to recover one or more spatial streams desired for the base station 600. If multiple spatial streams are desired for the base station 600, the one or more Rx processors 630 may combine the multiple spatial streams into a single OFDM symbol stream. The one or more Rx processors 630 may convert the OFDM symbol stream from the time-domain to the frequency-domain using an FFT. The OFDM symbol stream may be demodulated and decoded, to recover the uplink data and / or control signal transmitted by the wireless device 650 on physical channel, based on a channel estimator at the base station 600 (not shown in FIG. 6) . The data and control signals are then provided to the processor / controller 625, which implements layer 3 and layer 2 functionality.
[0042] As shown in FIG. 6, the wireless device 650 and the base station 600 may have multiple antennas. The multiple antennas may be used to perform one or more MIMO or multi-antenna techniques, such as spatial multiplexing (e.g., single-user MIMO or multi-user MIMO) , transmit / receive diversity, and / or beamforming. In other examples, the wireless device 650 and / or the base station 600 may have a single antenna.
[0043] The processor / controller 625 and / or the processor / controller 675 may be associated with a memory 610 and a memory 660, respectively. Memory 610 and memory 660 (e.g., one or more non-transitory computer readable mediums) may store computer program instructions or code that may be executed by the processor / controller 625 and / or the processor / controller 675 to carry out one or more of the functionalities discussed in the present application. Although not shown in FIG. 6, the one or more Tx processors 620, the one or more Tx processors 670, the one or more Rx processors 630, and / or the one or more Rx processors 690 may be coupled to a memory (e.g., one or more non-transitory computer readable mediums) storing computer program instructions or code that may be executed to carry out one or more of their respective functionalities.
[0044] The processor / controller 625 and / or the processor / controller 675 may comprise one or more controllers and / or one or more processors. The one or more controllers and / or one or more processors may comprise, for example, a general-purpose processor, a digital signal processor (DSP) , a microcontroller, an application specific integrated circuit (ASIC) , a field programmable gate array (FPGA) and / or other programmable logic device, discrete gate and / or transistor logic, discrete hardware components, an on-board unit, or any combination thereof. The processor / controller 625 and / or the processor / controller 675 may perform at least one of signal coding / processing, data processing, power control, input / output processing, and / or any other functionality that may enable the wireless device 650 and the base station 600 to operate in a wireless environment.
[0045] The processor / controller 625 and / or the processor / controller 675 may be connected to one or more peripherals 605 and one or more peripherals 655, respectively. The one or more peripherals 605 and the one or more peripherals 655 may include software and / or hardware that provide features and / or functionalities, for example, a speaker, a microphone, a keypad, a display, a touchpad, a power source, a satellite transceiver, a universal serial bus (USB) port, a hands-free headset, a frequency modulated (FM) radio unit, a media player, an Internet browser, an electronic control unit (e.g., for a motor vehicle) , and / or one or more sensors (e.g., an accelerometer, a gyroscope, a temperature sensor, a radar sensor, a lidar sensor, an ultrasonic sensor, a light sensor, a camera, and / or the like) . The processor / controller 625 and / or the processor / controller 675 may receive user input data from and / or provide user output data to the one or more peripherals 605 and / or the one or more peripherals 655. The processor / controller 675 in the wireless device 650 may receive power from a power source and / or may be configured to distribute the power to the other components in the wireless device 650. The power source may comprise one or more sources of power, for example, a battery, a solar cell, a fuel cell, or any combination thereof. The processor / controller 625 and / or the processor / controller 675 may be connected to a GPS chipset 615 and a GPS chipset 625, respectively. The GPS chipset 615 and the GPS chipset 625 may be configured to provide geographic location information of the wireless device 650 and the base station 600, respectively.
[0046] In the present disclosure, a base station may communicate with various types / categories of wireless devices. Wireless devices and / or base stations may support different technologies, and / or different releases of the same technology. A downlink transmission from (e.g., by) a base station and to a wireless device may be the same as a downlink reception by the wireless device and from the base station. An uplink transmission from (e.g., by) a wireless device to a base station may be the same as an uplink reception by the base station and from the wireless device.
[0047] In the present disclosure, a wireless device may receive from a base station one or more messages (e.g. RRC messages) comprising configuration parameters of one or more cells (e.g. primary cell, secondary cell) . The wireless device may communicate with at least one base station of the one or more cells. The one or more messages (e.g. as a part of the configuration parameters) may comprise parameters of PHY, MAC, RLC, PCDP, SDAP, and / or RRC layers for configuring the wireless device. For example, the configuration parameters may comprise parameters for configuring PHY and MAC layer channels, bearers, etc. For example, the configuration parameters may comprise parameters indicating values of timers for PHY, MAC, RLC, PCDP, SDAP, RRC layers, and / or communication channels.
[0048] In the present disclosure, a wireless device and a base station may exchange control signaling (e.g., referred to as L1 / L2 control signaling and may originate from the PHY layer (e.g., layer 1) and / or the MAC layer (e.g., layer 2) ) . The control signaling may comprise downlink control signaling (e.g., MAC control element (MAC CE) and / or DCI) transmitted from the base station to the UE and / or uplink control signaling (e.g., MAC CE and / or uplink control information (UCI) ) transmitted from the UE to the base station.
[0049] In the present disclosure, an uplink configurated grant (CG) configuration may be a type 1 CG configuration. A base station may transmit an RRC message to a wireless device for configuring uplink transmission occasions / resources based on the type 1 CG configuration. In response to receiving the RRC message, the wireless device may activate the uplink transmission occasions / resources without further message / signaling for activation of the CG configuration.
[0050] In the present disclosure, an uplink CG configuration may be a type 2 CG configuration. A base station may transmit an RRC message to a wireless device for configuring uplink transmission occasions / resources based on the type 2 CG configuration. In response to receiving the RRC message, the wireless device may not activate the uplink transmission occasions / resources. The base station may further transmit a DCI to the wireless device for activating / releasing the uplink transmission occasions / resources.
[0051] In the downlink, a base station may transmit (e.g., unicast, multicast, and / or broadcast) one or more downlink reference signals (RSs) to a UE. The downlink RSs may comprise primary synchronization signal (PSS) , secondary synchronization signal (SSS) , channel state information reference signal (CSI-RS) , demodulation reference signal (DMRS) , and / or phase tracking reference signal (PT-RS) . In the uplink, the UE may transmit one or more uplink RSs to the base station. The uplink RSs may comprise DMRS, PT-RS, and / or sounding reference signal (SRS) . A PSS, a SSS together with a physical broadcast channel (PBCH) may be jointly referred to as a synchronization signal block or a SS block (SSB) .
[0052] FIG. 7A illustrates an example of a burst of SSBs. The burst of SSBs may comprise N SSBs. A SSB (e.g., SSB #1 in FIG. 7A) of the N SSBs may span 4 OFDM symbols in the time domain and 240 subcarriers in the frequency domain. A PSS of the SSB may be in the first OFDM symbol of the 4 OFDM symbols in the SSB and may occupy 127 subcarriers. The remaining subcarriers of the SSB in the first OFDM may be empty. An SSS of the SSB may be in the third OFDM symbol of the 4 OFDM symbols in the SSB and may occupy 127 subcarriers. There may be 8 and 9 empty subcarriers on each side of the SSS. A PBCH of the SSB may be in the second and the fourth OFDM symbols of the SSB. In addition, the PBCH may also use 48 subcarriers on each side of the SSS in the third OFDM symbol. One or more DMRS may be transmitted in resource elements of the PBCH for coherent demodulation of the PBCH.
[0053] In an example, the burst of SSBs may be within a beam-sweep of downlink beams of the base station and referred to as an SS burst set. A first SSB (e.g., SSB #1 in FIG. 7A) of the burst of SSBs may be transmitted via a first downlink beam of the base station (e.g., beam #1) within the beam-sweep, a second SSB (e.g., SSB #2 in FIG. 7A) of the burst of SSBs may be transmitted via a second downlink beam of the base station (e.g., beam #2) within the beam-sweep, and so on. A last SSB (e.g., SSB #N in FIG. 7A) of the burst of SSBs may be transmitted via a last downlink beam of the base station (e.g., beam #N) within the beam-sweep.
[0054] In an example, a SSB may be transmitted periodically with a period that may vary from 5 ms up to 160 ms. In an example of FIG. 7A and when a period of SSB is 20 ms, the base station may transmit a second burst of SSBs, to the UE, 20 ms after the burst of SSBs. The second burst of SSBs may comprise a periodically transmitted SSB #1 which is 20 ms after the SSB #1 in the burst of SSBs, a periodically transmitted SSB #2 which is 20 ms after the SSB #2 in the burst of SSBs, and so on. The second burst of SSBs may be for a second round beam-sweep of the downlink beams (e.g., from beam #1 to beam #N in FIG. 7A) of the base station.
[0055] In an example, a SSB may have a SSB index. The UE may assume that one or more SSBs transmitted with a same SSB index are quasi co-located (QCLed) (e.g., having the same / similar Doppler spread, Doppler shift, average gain, average delay, and / or spatial Rx parameters) . The UE may not assume QCL for SSB transmissions having different SSB indices.
[0056] In an example, a base station may configure one or more CSI-RSs to a UE. The base station may transmit the one or more CSI-RSs to the UE for different purposes such as acquiring channel state information (CSI) , channel measurements, and channel sounding, etc. In response to receiving the one or more CSI-RSs, the UE may generate a CSI report and send the CSI report to the base station. The base station may perform link adaptation based on the CSI report.
[0057] In an example, a CSI-RS may correspond to up to 32 different antenna ports, each corresponding to a channel to be sounded. In an example, the CSI-RS may be a single-port CSI-RS occupying a single resource element (e.g., 1 OFDM symbol by 1 subcarrier referring to FIG. 3) . In an example, the CSI-RS may be a multi-port CSI-RS which may be seen as multiple orthogonally transmitted per-antenna-port CSI-RS sharing resource elements assigned for the multi-port CSI-RS. In an example, the CSI-RS may be a zero-power CSI-RS (ZP CSI-RS) or a non-zero-power (NZP CSI-RS) .
[0058] In an example, a NZP CSI-RS may be transmitted using one or more time-frequency resources (e.g., resource elements) . The one or more time-frequency resources for the transmission of the NZP CSI-RS may have (e.g., be associated with) a NZP CSI-RS resource identity (e.g., ID) . The NZP CSI-RS may be transmitted via a downlink beam of a base station.
[0059] In an example, a base station may configure a UE with one or more CSI-RS resource sets (e.g., NZP CSI-RS resource sets) . A CSI-RS resource set may be associated with a CSI-RS resource set ID. A CSI-RS resource set may comprise (e.g., pointers to) one or more SSB indexes and / or one or more CSI-RS resource IDs. In an example, all CSI-RS within a semi-persistently configured CSI-RS resource set maybe jointly activated / deactivated by a MAC CE commend. In an example, transmission of all CSI-RS within an aperiodic configured CSI-RS resource set may be jointly triggered by a DCI.
[0060] FIG. 7B illustrates an example of CSI-RSs that are mapped in the time and frequency domains. In an example, N CSI-RSs may be mapped to orthogonal (e.g., in time, frequency, and / or code domains) resources. A first CSI-RS of the N CSI-RSs (e.g., CSI-RS #1 in FIG. 7B) may be transmitted via a first downlink beam of a base station (e.g., beam #1 in FIG. 7B) . A second CSI-RS of the N CSI-RSs (e.g., CSI-RS #2 in FIG. 7B) may be transmitted via a second downlink beam of the base station (e.g., beam #2 in FIG. 7B) . A N-th CSI-RS of the N CSI-RSs (e.g., CSI-RS #N in FIG. 7B) may be transmitted via a N-th downlink beam of the base station (e.g., beam #N in FIG. 7B) . In an example, the N CSI-RSs may comprise one or more single-port and / or multi-port CSI-RSs. In an example, a square shown in FIG. 7B may span an RB within a bandwidth of a cell. A base station may transmit one or more RRC messages comprising CSI-RS resource configuration parameters indicating the N CSI-RSs. One or more of the following parameters may be configured by higher layer signaling (e.g., RRC and / or MAC signaling) for a CSI-RS resource configuration: a CSI-RS resource configuration identity (e.g., ID) , a number of CSI-RS ports, a CSI-RS configuration (e.g., symbol and resource element (RE) locations in a subframe) , a CSI-RS subframe configuration (e.g., subframe location, offset, and periodicity in a radio frame) , a CSI-RS power parameter, a CSI-RS sequence parameter, a code division multiplexing (CDM) type parameter, a frequency density, a transmission comb, QCL parameters (e.g., QCL-scramblingidentity, crs-portscount, mbsfn-subframeconfiglist, csi-rs-configZPid, qcl-csi-rs-configNZPid) , and / or other radio resource parameters.
[0061] In an example, a base station may configure a UE to perform measurements with corresponding reporting to network (e.g., the base station) . The base station may send to the UE an RRC message comprising configuration parameters (e.g., CSI-ReportConfig) for a configuration of a measurement and corresponding reporting. The configuration parameters may indicate one or more quantities to be reported, one or more downlink resources on which the measurement should be performed in order to derive the one or more quantities to be reported, and / or how / when the reporting is done and what uplink physical channel to use for the reporting. The one or more quantities to be reported may comprise channel quality indicator (CQI) , rank indicator (RI) , and / or precoder-matrix indicator (PMI) , which may be jointly referred to as CSI. Additionally or alternatively, the one or more quantities to be reported may comprise reference signal received power (RSRP) of one or more downlink reference signals (e.g., Layer 1 RSRP or L1-RSRP of one or more SSBs and / or CSI-RSs) . In an example, the UE may send CSI reports of measurements to the base station periodically, aperiodically, or semi-persistently. For periodic CSI reporting, the UE may be configured with a timing and / or periodicity of a plurality of CSI reports. For aperiodic CSI reporting, the base station may request a CSI report. For example, the base station may command the UE to measure a configured CSI-RS resource and provide a CSI report relating to the measurements. For semi-persistent CSI reporting, the base station may configure the UE to transmit periodically, and selectively activate or deactivate the periodic reporting. The base station may configure the UE with a CSI-RS resource set and CSI reports using RRC signaling.
[0062] In an example, beam management may comprise beam measurement, beam selection, and beam indication. A beam may be associated with one or more reference signals. For example, a beam may be identified by one or more beamformed reference signals. The UE may perform downlink beam measurement based on downlink reference signals (e.g., SSB and / or CSI-RS, etc. ) and generate a beam measurement report. The UE may perform the downlink beam measurement procedure after an RRC connection is set up with a base station. The purpose of beam management may be to establish and retain one or more beam pair links providing good connectivity. A beam pair link may comprise a transmitting (Tx) beam transmitted by the base station and a receiving (Rx) beam received by the UE. In an example, downlink beam management may comprise an initial beam establishment / paring procedure (e.g., procedure P1) , a downlink Tx side beam adjustment procedure (e.g., procedure P2) , and a downlink Rx beam adjustment procedure (e.g., procedure P3) .
[0063] FIG. 8A illustrates an example of a downlink beam management procedure (e.g., P1) for initial beam paring. In FIG. 8A, a base station may transmit one or more downlink RSs by sweeping one or more downlink Tx beams (e.g., in a counter-clockwise direction indicated by the dashed arrow at the base station side) . Beamforming at a UE may comprise an Rx beam sweep for one or more downlink Rx beams (e.g., in a clockwise direction indicated by the dashed arrow at the UE side) . Based on beam measurements of the one or more downlink RSs, the UE may send a report back to the base station indicating a downlink Tx beam (e.g., a downlink Tx beam with RS #5 in solid line) of the one or more swept downlink beams should be used for downlink transmissions. Furthermore based on Rx beam sweeping at the UE, the UE may determine to use a downlink Rx beam (e.g., in solid line at the UE side) for receiving the downlink transmissions from the base station. FIG. 8B illustrates an example of a downlink beam management procedure (e.g., P2) for downlink Tx beam adjustment. In FIG. 8B, a downlink Rx beam at a UE may be given (e.g., a downlink Rx beam in solid line) . The UE may measure one or more downlink RSs transmitted corresponding to one or more downlink Tx beams of a base station by a beam sweep. The UE may report measurement results to network (e.g., the base station) . The network (e.g., the base station) may adjust current using downlink beam (s) based on the report of the measurement results. FIG. 8C illustrates an example of a downlink beam management procedure (e.g., P3) for downlink Rx beam adjustment. In FIG. 8C, a downlink Tx beam at a base station may be given (e.g., a downlink Tx beam transmitting RS #5 in solid line) . A UE may measure a downlink RS transmitted via the downlink Tx beam by a beam sweep of one or more downlink Rx beams at the UE side. The UE may adjust current using downlink Rx beam (s) based on the measurement and the beam sweep.
[0064] A UE may initiate a beam failure recovery (BFR) procedure based on detecting a beam failure. The UE may transmit a BFR request (e.g., a preamble, a UCI, an SR, a MAC CE, and / or the like) based on the initiating of the BFR procedure. The UE may detect the beam failure based on a determination that a quality of beam pair link (s) of an associated control channel is unsatisfactory (e.g., having an error rate higher than an error rate threshold, a received signal power lower than a received signal power threshold, an expiration of a timer, and / or the like) .
[0065] The UE may measure a quality of a beam pair link using one or more reference signals (RSs) comprising one or more SS / PBCH blocks, one or more CSI-RS resources, and / or one or more demodulation reference signals (DMRSs) . A quality of the beam pair link may be based on one or more of a block error rate (BLER) , an RSRP value, a signal to interference plus noise ratio (SINR) value, a reference signal received quality (RSRQ) value, and / or a CSI value measured on RS resources. The base station may indicate that an RS resource is quasi co-located (QCLed) with one or more DM-RSs of a channel (e.g., a control channel, a shared data channel, and / or the like) . The RS resource and the one or more DMRSs of the channel may be QCLed when the channel characteristics (e.g., Doppler shift, Doppler spread, average delay, delay spread, spatial Rx parameter, fading, and / or the like) from a transmission via the RS resource to the UE are similar or the same as the channel characteristics from a transmission via the channel to the UE.
[0066] In an example, application of artificial intelligence / machine learning (AI / ML) to wireless communications may be enabled. One or more AI / ML models / functionalities may be implemented at network side (e.g., NW-side model) , at UE side (e.g., UE-side model) , and at both base station side and UE side (e.g., two-sided model) . A life cycle management (LCM) of AI / ML model (e.g., model training, model deployment, model inference, model monitoring, model updating) and AI / ML functionality may be characterized as below: - Data collection, which may also include associated assistance information, if applicable. - Model training - Functionality / model identification - Model delivery / transfer - Model inference operation - Functionality / model selection, activation, deactivation, switching, and fallback operation. The function / model selection may comprise decision by the network (either network initiated or UE-initiated and requested to the network) , decision by the UE (event-triggered as configured by the network, UE’s decision reported to the network, or UE-autonomous either with UE’s decision reported to the network or without it) - Functionality / model monitoring - Model update - UE capability
[0067] In an example, an AI / ML model may have a model identity (ID) with associated information. A functionality may be provided by one or more AI / ML operations. An AI / ML model identified by a model ID may be logical, and how it maps to physical AI / ML model (s) may be up to implementation. A logical AI / ML model may refer to a model that is identified and assigned a model ID, and physical AI / ML model (s) may refer to an actual implementation of the model.
[0068] FIG. 9 illustrates an example of a functional framework for AI / ML approaches for wireless communications. As shown in FIG. 9, a general framework for AI / ML approaches for wireless communications may comprise following: - Data Collection: a function that provides input data to AI / ML Model Training, Management, and Inference functions. - Training Data: data needed as input for the AI / ML Model Training function. - Monitoring Data: data needed as input for the Management of AI / ML models or AI / ML functionalities. - Inference Data: data needed as input for the AI / ML Inference function. - Model Training: a function that performs AI / ML model training, validation, and testing which may generate model performance metrics which can be used as part of model testing procedure. The Model Training function may be also responsible for data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) based on Training Data delivered by a Data Collection function, if required. - Trained / Updated Model: in case of having a Model Storage function, this may be used to deliver trained, validated, and tested AI / ML models to the Model Storage function, or to deliver an updated version of a model to the Model Storage function. - Management: a function that oversees the operation (e.g., selection / (de) activation / switching / fallback) and monitoring (e.g., performance) of AI / ML models or AI / ML functionalities. This function may be also responsible for making decisions to ensure proper inference operation based on data received from the Data Collection function (e.g., monitoring data) and the Inference function (e.g., inference output) . - Management Instruction: information needed as input to manage the Inference function. Concerning information may include selection / (de) activation / switching of AI / ML models or AI / ML-based functionalities, fallback to non-AI / ML operation (i.e., not relying on inference process) , etc. - Model Transfer / Delivery Request: used to request model (s) to the Model Storage function. - Performance Feedback / Retraining Request: information needed as input for the Model Training function, e.g., for model (re) training or updating purposes. - Inference: a function that provides outputs from the process of applying AI / ML models or AI / ML functionalities, using the data that is provided by the Data Collection function (i.e., Inference Data) as an input. The Inference function may be also responsible for data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) based on Inference Data delivered by a Data Collection function, if required. - Inference Output: data used by the Management function to monitor the performance of AI / ML models or AI / ML functionalities. - Model Storage: a function responsible for storing trained / updated models that can be used to perform the Inference function. The Model Storage function may be only intended as a reference point (if any) when applicable for protocol terminations, model transfer / delivery, and related processes. The Model Storage function does not encompass restricting actual storage locations of models. - Model Transfer / Delivery: used to deliver an AI / ML model to the Inference function.
[0069] In an example, AI / ML approaches (e.g., AI / ML models / functionalities) may be implemented for beam management (e.g., prediction in time and / or spatial domain, overhead and latency reduction, beam selection accuracy improvement, etc. ) .
[0070] In beam management case 1 (e.g., BM-Case 1) employing AI / ML approaches, one or more AI / ML models may be used for spatial-domain downlink beam prediction for a first set (e.g., set A) of beams based on measurement results of a second set (e.g., set B) of beams. The set A may be for downlink beam prediction. The set B may be a set of beams whose measurements are taken as inputs of the one or more AI / ML models. Beams in the set A and the set B may or may not be in same frequency range (e.g., sub-6GHz frequency range or millimeter wave frequency range) . The set A and the set B may be different. The set B may or may not be a subset of the set A. Model training and inference of the one or more AI / ML models may be at NW side and / or UE side. Inputs of the one or more AI / ML models may comprise: L1-RSRP measurement based on the set B, L1-RSRP measurement based on the set B and assistance information, channel impulse response (CIR) based on the set B, and / or L1-RSRP measurement based on the set B and the corresponding downlink Tx and / or Rx beam identity (ID) .
[0071] In beam management case 2 (e.g., BM-Case 2) employing AI / ML approaches, one or more AI / ML models may be used for temporal downlink beam prediction for a first set (e.g., set A) of beams based on historic measurement results of a second set (e.g., set B) of beams. The set A may be for downlink beam prediction. The set B may be a set of beams whose measurements are taken as inputs of the one or more AI / ML models. The set A and the set B may be same or different. The set B may or may not be a subset of the set A. Model training and inference of the one or more AI / ML models may be at NW side and / or at UE side. Inputs of the one or more AI / ML models may comprise measurement results of one or more latest measurement instances with following alternatives: Alt. 1) : L1-RSRP measurement based on the set B; Alt 2) : L1-RSRP measurement based on the set B and assistance information; Alt. 3) : L1-RSRP measurement based on the set B and corresponding downlink Tx and / or Rx beam ID. Predictions for one or more (e.g., N≥1) future time instances may be obtained based on outputs of the one or more AI / ML models, where each prediction of the predictions may be for each time instance of the one or more future time instances.
[0072] For both BM-Case 1 and BM-Case 2, AI / ML model output may comprise: - Alt. 1: Tx and / or Rx Beam ID (s) and / or predicted L1-RSRP of one or more predicted downlink Tx and / or Rx beams. In an example, the one or more predicted beams may be top-K predicted beams (e.g., K≥1) . - Alt. 2: Tx and / or Rx Beam ID (s) of one or more predicted downlink Tx and / or Rx beams and other information. The one or more predicted beams may be top-K predicted beams (e.g., K≥1) . - Alt. 3: Tx and / or Rx Beam angle (s) and / or predicted L1-RSRP of one or more predicted DL Tx and / or Rx beams. The one or more predicted beams may be top-K predicted beams (e.g., K≥1) . The top-K beam IDs may have been derived via post-processing of the one or more AI / ML model output.
[0073] For BM-Case 1 and BM-Case 2: - For model training, training data may be generated by UE and / or base station. - For NW-side model inference, input data may be generated by UE and terminated at base station. - For UE-side model inference, input data may be internally available at UE. - For performance monitoring at NW side, calculated performance metrics (if needed) or data needed for performance metric calculation (if needed) may be generated by UE and terminated at base station.
[0074] FIG. 10 illustrates an example of beam prediction and reporting using AI / ML approaches. Referring to FIG. 9 and in an example of FIG. 10, a base station may transmit to a wireless device one or more RRC messages comprising configuration parameters of a first set of beams, a second set of beams, and a number K indicating how many beams to report in an inference report. In an example, the first set of beams may be a beam set A for beam prediction / selection using AI / ML approaches (e.g., AI / ML models and / or AI / ML functionalities) . The second set of beams may be a beam set B for measuring reference signals (RSs) , received via the second set of beams, to predict / select one or more beams from the beam set A. The beam set B may be the same as the beam set A. The beam set B may be different from the beam set A (e.g., the beam set B may be not a subset of the beam set A) . The beam set B may be a subset of the beam set A. In an example, the one or more configuration parameters may comprise / indicate a first CSI-RS resource set associated with the beam set A. The first CSI-RS resource set may comprise one or more first CSI-RS resources (e.g., and / or one or more first SSB resources) , where each of the one or more first CSI-RS resources (e.g., and / or the one or more first SSB resources) is associated with a beam of the beam set A. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam ID of the beam) of the beam set A based on a CSI-RS resource ID and / or a CSI resource indicator (CRI) of a CSI-RS resource, of the one or more first CSI-RS resources, associated with the beam. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam ID of the beam) of the beam set A based on an SSB index and / or an SSB resource indicator (SSBRI) of a SSB resource, of the one or more first SSB resources, associated with the beam. In an example, the one or more configuration parameters may comprise / indicate a second CSI-RS resource set associated with the beam set B. The second CSI-RS resource set may comprise one or more second CSI-RS resources (e.g., and / or one or more second SSB resources) , where each of the one or more second CSI-RS resources (e.g., and / or the one or more second SSB resources) is associated with a beam of the beam set B. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam index / ID of the beam) of the beam set B based on a CSI-RS resource ID and / or a CRI of a CSI-RS resource, of the one or more second CSI-RS resources, associated with the beam. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam index / ID of the beam) of the beam set B based on an SSB index and / or an SSBRI of a SSB resource, of the one or more second SSB resources, associated with the beam. In an example, the wireless device may predict / select, from the beam set A and using AI / ML approaches, a number of K beams for downlink transmissions by the base station. The wireless device may predict / select the number of K beams based on inputting measurements of CSI-RSs (e.g., and / or SSBs) received via the one or more second CSI-RS resources (e.g., or the one or more second SSB resources) on the beam set B to the AI / ML approaches. The wireless device may transmit the inference report of the AI / ML approaches indicating the predicted / selected number of K beams of the beam set A.
[0075] In an example of FIG. 10, the base station may configure, to the wireless device, the beam set A comprising a beam 1, a beam 2, a beam 3, a beam 4, a beam 5, and a beam 6. The base station may configure, to the wireless device, the beam set B comprising the beam 3 and the beam 5. The beam set B is a subset of the beam set A. For the beam set A, the base station may further configure a RS #A1 via a first RS resource having a first RS resource ID associated with (e.g., using) the beam 1 of the beam set A, a RS #A2 via a second RS resource having a second RS resource ID and associated with (e.g., using) the beam 2 of the beam set A, and so on. For the beam set B, the base station may further configure a RS #B1 via a seventh RS resource having a seventh RS resource ID associated with (e.g., using) the beam 3 of the beam set B, and a RS #B2 via an eighth RS resource having an eighth RS resource ID associated with (e.g., using) the beam 5 of the beam set B. In an example, the seventh RS resource and / or the eighth RS resource may be same as the third RS resource and / or the fifth RS resource, respectively. In another example, the seventh RS resource and / or the eighth RS resource may be different from the third RS resource and / or the fifth RS resource, respectively. The RSs for the beam set A (e.g., RS #A1, RS #A2, RS #A3, RS #A4, RS #A5, RS #A6) and the RSs for the beam set B (e.g., RS #B1, RS #B2) may be CSI-RS and / or SSB. An RS resource ID of a CSI-RS may be a CSI-RS resource ID or a CRI. An RS resource ID of a SSB may be an SSB index or a SSBRI. For beam prediction, the wireless device may measure the RS #B1 received via the seventh RS resource on the beam 3 of the beam set B and the RS #B2 received via the eighth RS resource on the beam 5 of the beam set B. The wireless device may not measure beams (e.g., RSs on the beams) other than the beam 3 and the beam 5 based on the other beams being not in the beam set B. The wireless device may input measurement results of the RS #B1 and the RS #B2 to AI / ML models / functionalities. The wireless device may predict / select the beam 5 (e.g., in solid line in the beam set A) and the beam 2 (e.g., in solid line in the beam set A) from the beam set A based on output of the AI / ML models / functionalities. The wireless device may transmit the inference report to the base station indicating the beam 5 and the beam 2 (e.g., or beam indexes of the beam 5 and the beam 2) in the beam set A. The wireless device may indicate the beam 5 (e.g., an index of the beam 5) using the fifth RS resource ID (e.g., CSI-RS resource ID, CRI, SSB index, or SSBRI) of the RS #A5 based on the beam set A in the inference report. The wireless device may not indicate the beam 5 (e.g., an index of the beam 5) using the eighth RS resource ID (e.g., CSI-RS resource ID, CRI, SSB index, or SSBRI) of the RS #B2 based on the beam set B. The wireless device may indicate the beam 2 (e.g., an index of the beam 2) using the second RS resource ID (e.g., CSI-RS resource ID, CRI, SSB index, or SSBRI) of the RS #A2 based on the beam set A in the inference report.
[0076] In existing technologies, for BM-Case 1 using AI / ML approaches, a base station may configure, to a wireless device, a number indicating how many beams to report in an inference report. In response to receiving the number, the wireless device may predict / select the number of beams from a beam set A based on measurements of received RSs via a beam set B. The wireless device may send, to the base station, the inference report indicating the number of beams and / or corresponding RSRPs of the number of beams.
[0077] Implementing the existing technologies, for BM-Case 2 using AI / ML approaches, a base station may configure, to a wireless device, a number of time instances to be covered in a single inference report and a number indicating how many beams to be reported for each time instance of the number of time instances in the single inference report. The wireless device may predict / select the number of beams for the each time instance of the number of time instances. The wireless device may transmit, to a base station, the single inference report (e.g., comprising inference results) indicating the number of beams and / or corresponding RSRPs for the each time instance of the number of future time instances. A length of the single inference report for BM-Case 2 may be multiple times longer than a length of an inference report for BM-Case 1. Signaling overhead reduction may be necessary for BM-Case 2 inference report.
[0078] Embodiments of the present disclosure may enable a wireless device to predict / select a smaller number of beams for each time instance of a plurality of time instances to be reported in an inference report. In an example, a wireless device may receive, from a base station, one or more RRC messages comprising a first parameter and a second parameter. The first parameter may indicate a number of time instances to be reported in an inference report using AI / ML approaches. The number of time instances may comprise a first time instance and a second time instance. The second parameter may indicate a first number of predicted downlink beams (e.g., a maximum number of predicted / selected beams for a time instance) , for a time instance of the number of time instances, for the inference report using the AI / ML approaches. The wireless device may predict / select, using the AI / ML approaches, a second number of downlink beams for the first time instance and a third number of downlink beams for the second time instance. The wireless device may predict / select, from a first set of beams, the second number of downlink beams and the third number of downlink beams based on measurement of received reference signals via a second set of beams. When the second number and the third number are smaller than or equal to the first number, the wireless device may transmit, to the base station, the inference report for the number of time instances indicating the predicted second number of downlink beams for the first time instance and the predicted third number of downlink beams for the second time instance. In another example, when the second number and / or the third number are greater than the first number, the wireless device may transmit, to the base station, the inference report for the number of time instances indicating a smaller number, between the first number and the second number, of the predicted second number of downlink beams for the first time instance and a smaller number, between the first number and the third number, of the predicted third number of downlink beams for the second time instance. In another example, the wireless device may transmit, to the base station, the inference report for the number of time instances indicating one or more first predicted beams for the first time instance and one or more second predicted beams for the second time instance, where a number of beams in the one or more first predicted beams is the same as a number of beams in the one or more second predicted beams. For example, the number of beams may be the first number. Embodiments of the present disclosure may enable a wireless device to transmit an inference report with variable length. In an example, a wireless device may transmit, to a base station, an inference report comprising different numbers of beams for each of a number of time instances. The inference report may comprise a first part with a fixed length (e.g., for blind decoding) and a second part with a variable length (e.g., based on the different numbers of beams for he each of the number of time instances) . The first part of the inference report may indicate a reference quantized RSRP and a length indication of the variable length of the second part of the inference report. The second part of the inference report may indicate differential RSRPs, compared with the reference quantized RSRP in the first part of the inference report, of the different numbers of beams for the each of the plurality of time instances.
[0079] Implementing the embodiments of the present disclosure may reduce signaling overhead of an inference report comprising predicted / selected beams for a plurality of future time instances. Implementing the embodiments of the present disclosure may increase flexibility for an inference report comprising predicted beams for a plurality of future time instances. Implementing the embodiments of the present disclosure may enabling blind decoding of a first part of an inference report with a fixed length. Implementing the embodiments of the present disclosure may simplify signaling design for an inference report comprising predicted beams for a plurality of future time instances.
[0080] FIG. 11 illustrates an example of beam prediction and inference reporting when an inference report comprises predicted beams and / or corresponding RSRPs for a plurality of time instances. In an example of FIG. 11 and referring to Fig. 10, a base station may transmit to a wireless device one or more RRC messages comprising one or more first configuration parameters. The one or more first configuration parameters may configure / indicate a first set of beams and a second set of beams. In an example, the one or more first configuration parameters may comprise / indicate a first CSI-RS resource set associated with the first set of beams. The first CSI-RS resource set may comprise one or more first CSI-RS resources (e.g., and / or one or more first SSB resources) , where each of the one or more first CSI-RS resources (e.g., and / or the one or more first SSB resources) is associated with (e.g., transmitted on / mapped to) a beam of the first set of beams. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam index / ID of the beam) of the first set of beams based on a CSI-RS resource ID and / or a CRI of a CSI-RS resource, of the one or more first CSI-RS resources, associated with the beam. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam index / ID of the beam) of the first set of beams based on an SSB index and / or an SSBRI of a SSB resource, of the one or more first SSB resources, associated with the beam. In an example, the one or more first configuration parameters may comprise / indicate a second CSI-RS resource set associated with the second set of beams. The second CSI-RS resource set may comprise one or more second CSI-RS resources (e.g., and / or one or more second SSB resources) , where each of the one or more second CSI-RS resources (e.g., and / or the one or more second SSB resources) is associated with (e.g., transmitted on / mapped to) a beam of the second set of beams. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam index / ID of the beam) of the second set of beams based on a CSI-RS resource ID and / or a CRI of a CSI-RS resource, of the one or more second CSI-RS resources, associated with the beam. The base station and the wireless device may identify / indicate / recognize / determine a beam (e.g., a beam index / ID of the beam) of the second set of beams based on an SSB index and / or an SSBRI of a SSB resource, of the one or more second SSB resources, associated with the beam. In an example, the first set of beams may be a beam set A for beam prediction / selection using AI / ML approaches (e.g., AI / ML models and / or AI / ML functionalities) . The second set of beams may be a beam set B for measuring reference signals, received via the second set of beams, to predict / select one or more beams from the beam set A. The beam set B may be the same as the beam set A. The beam set B may be different from the beam set A (e.g., the beam set B may be not a subset of the beam set A) . The beam set B may be a subset of the beam set A. In an example, one or more first CSI-RSs via the one or more first CSI-RS resources may be aperiodic / periodic / semi-persistent CSI-RSs. In an example, one or more second CSI-RSs via the one or more second CSI-RS resources may be aperiodic / periodic / semi-persistent CSI-RSs.
[0081] In an example of FIG. 11 and referring to Fig. 10, the one or more RRC messages may comprise one or more second configuration parameters configuring / indicating a number of Nmax time instances (e.g., Nmax future time instances) to be reported / covered in an inference report (e.g., a single report of inference results) of the AI / ML approaches, where Nmax is an integer number and Nmax≥1. The one or more second configuration parameters may configure / indicate a number of Kmax beams indicating a maximum number of beams to be reported for a time instance (e.g., each time instance of a plurality of time instances) in the inference report (e.g., a single inference report) , where Kmax is an integer number and Kmax≥1.
[0082] In an example of FIG. 11 and referring to Fig. 10, the wireless device may determine a number of N time instances (e.g., N future time instances) to be included / reported / covered in the (e.g., single) inference report, based on 1<N≤Nmax. The number of N time instances may at least comprise a first time instance and a second time instance. The wireless device may determine the number of N, where N≤Nmax, for reducing a number of time instances to be reported in the inference report for signal overhead reduction. In another example, the wireless device may determine the number of Nmax time instances to be included / reported / covered in the (e.g., single) inference report in response to the receiving the one or more RRC messages. The Nmax time instances may at least comprise a first time instance and a second time instance.
[0083] In an example of FIG. 11 and referring to Fig. 10, the wireless device may receive the one or more second CSI-RSs (e.g., and / or the one or more second SSBs) via the one or more second CSI-RS resources (e.g., and / or the one or more second SSB resources) on the beam set B during a past time duration (e.g., in an observation window) . The wireless device may measure the one or more second CSI-RSs (e.g., and / or the one or more second SSBs) . The wireless device may input measurement results into the AI / ML models / functionalities. The wireless device may predict / select, from the beam set A and based on output of the AI / ML models / functionalities, a number of K1 beams for downlink transmissions by the base station during the first time instance, where K1 is an integer and K1≥1. The wireless device may predict / select, from the beam set A and based on the output of the AI / ML models / functionalities, a number of K2 beams for downlink transmissions by the base station during the second time instance, where K2 is an integer and K2≥1.
[0084] In an example of FIG. 11 and referring to Fig. 10, the one or more RRC messages may comprise / indicate an RSRP threshold. The wireless device may predict / select the number of K1 beams based on predicted RSRPs, for / during the first time instance, of CSI-RSs / SSBs associated with the number of K1 beams in the beam set A being higher than or equal to the RSRP threshold. The wireless device may predict / select the number of K2 beams based on predicted RSRPs, for / during the second time instance, of CSI-RSs / SSBs associated with the number of K2 beams in the beam set A being higher than or equal to the RSRP threshold.
[0085] In an example of FIG. 11 and referring to Fig. 10, the one or more RRC messages may comprise / indicate an X decibel (dB) RSRP gap. The wireless device may predict / select the number of K1 beams based on predicted RSRPs, for / during the first time instance, of CSI-RSs / SSBs associated with the number of K1 beams in the beam set A being within the X dB gap compared with a strongest RSRP of the beam set A. The wireless device may predict / select the number of K2 beams based on predicted RSRPs, for / during the second time instance, of CSI-RSs / SSBs associated with the number of K2 beams of the beam set A being within the X dB gap compared with a strongest RSRP of the beam set A.
[0086] In an example of FIG. 11 and referring to Fig. 10, the one or more RRC messages may comprise / indicate a probability / confidence threshold. The wireless device may predict / select the number of K1 beams based on predicted probabilities, for / during the first time instance, of the number of K1 beams to be strongest beams (e.g., with highest RSRPs) in the beam set A being higher than or equal to the probability / confidence threshold. The wireless device may predict / select the number of K1 beams based on a summation of predicted probabilities, for / during the first time instance, of the number of K1 beams to be strongest beams (e.g., with highest RSRPs) in the beam set A being higher than or equal to the probability / confidence threshold. The wireless device may predict / select the number of K2 beams based on predicted probabilities, for / during the second time instance, of the number of K2 beams to be strongest beams (e.g., with highest RSRPs) in the beam set A being higher than or equal to the probability / confidence threshold. The wireless device may predict / select the number of K2 beams based on a summation of predicted probabilities, for / during the second time instance, of the number of K2 beams to be strongest beams (e.g., with highest RSRPs) in the beam set A being higher than or equal to the probability / confidence threshold. For example, and for the first time instance, the wireless device may predict / select a beam 1 and beam 2 based on a predicted probability of the beam 1 to be the strongest beam in the beam set A being higher than or equal to the probability / confidence threshold and a predicted probability of the beam 2 to be the strongest beam in the beam set A being higher than or equal to the probability / confidence threshold. For the second time instance, the wireless device may predict / select a beam 3 based on a predicted probability of the beam 3 to be the strongest beam in the beam set A being higher than or equal to the probability / confidence threshold. For example, and for the first time instance, the wireless device may predict / select a beam 1 and beam 2 based on a summation, of a first predicted probability / confidence of the beam 1 to be the strongest beam and a second predicted probability / confidence of the beam 2 to be the strongest beam, being higher than or equal to the probability / confidence threshold. For the second time instance, the wireless device may predict / select a beam 3 based on a predicted probability / confidence of the beam 3 to be the strongest beam in the beam set A being higher than or equal to the probability / confidence threshold.
[0087] In an example of FIG. 11 and referring to Fig. 10, the wireless device may transmit the inference report (e.g., a report of inference results) of the AI / ML models / functionalities (e.g., AI / ML approaches) indicating the predicted / selected number of K1 beams of the beam set A for the first time instance and the predicted / selected number of K2 beams of the beam set A for the second time instance. The wireless device may transmit the inference report of the AI / ML models / functionalities (e.g., AI / ML approaches) indicating the predicted / selected number of K1 beams of the beam set A for the first time instance and the predicted / selected number of K2 beams of the beam set A for the second time instance, based on (e.g., if / when) K1≤Kmax and K2≤Kmax. The inference report may comprise / indicate CRIs and / or SSBRIs associated with the number of K1 beams of the beam set A for the first time instance. The inference report may further comprise / indicate predicted RSRPs of the number of K1 beams of the beam set A for the first time instance. The inference report may comprise / indicate CRIs and / or SSBRIs associated with the number of K2 beams of the beam set A for the second time instance. The inference report may further comprise / indicate predicted RSRPs of the number of K2 beams of the beam set A for the second time instance.
[0088] In an example of FIG. 11 and referring to Fig. 10, the wireless device may further select the number of Kmax predicted / selected beams from the number of K1 beams for the first time instance based on (e.g., when / if) K1≥Kmax and / or the number of Kmax predicted / selected beams having highest RSRPs and / or probability / confidence to have highest RSRPs among the number of K1 beams. The wireless device may further select the number of Kmax predicted / selected beams from the number of K2 beams for the second time instance based on (e.g., when / if) K2≥Kmax and / or the number of Kmax predicted / selected beams having highest RSRPs and / or probability / confidence to have highest RSRPs among the number of K2 beams.. The wireless device may transmit the inference report of the AI / ML models / functionalities (e.g., AI / ML approaches) indicating the predicted / selected number of Kmax beams for the first time instance and the predicted / selected number of Kmax beams for the second time instance. The inference report may comprise / indicate CRIs and / or SSBRIs associated with the number of Kmax beams of the number of K1 beams for the first time instance. The inference report may further comprise / indicate predicted RSRPs of the number of Kmax beams for the first time instance. The inference report may comprise / indicate CRIs and / or SSBRIs associated with the number of Kmax beams of the number of K2 beams for the second time instance. The inference report may further comprise / indicate predicted RSRPs of the number of Kmax beams for the second time instance.
[0089] In an example of FIG. 11 and referring to Fig. 10, the wireless device may transmit the inference report of the AI / ML models / functionalities (e.g., AI / ML approaches) indicating the predicted / selected number of min [K1, Kmax] beams of the beam set A for the first time instance and the predicted / selected number of min [K2, Kmax] beams of the beam set A for the second time instance. The inference report may comprise / indicate CRIs and / or SSBRIs associated with the number of min [K1, Kmax] beams of the beam set A for the first time instance. The inference report may further comprise / indicate predicted RSRPs of the number of min [K1, Kmax] beams of the beam set A for the first time instance. The inference report may comprise / indicate CRIs and / or SSBRIs associated with the number of min [K2, Kmax] beams of the beam set A for the second time instance. The inference report may further comprise / indicate predicted RSRPs of the number of min [K2, Kmax] beams of the beam set A for the second time instance.
[0090] FIG. 12A and FIG. 12B illustrate examples of a single inference report indicating predicted beams and / or corresponding RSRPs for a plurality of time instances. In an example of FIG. 12A, an inference report (e.g., a report of inference results) for a plurality of time instances (e.g., for BM-Case 2 using AI / ML approaches) may comprise / indicate a number of predicted / selected beams for each time instance of the plurality of time instances in the inference report. The inference report may comprise / indicate one or more beams (e.g., beam indexes / IDs) of the number of predicted / selected beams for the each time instance of the plurality of time instances. The inference report may comprise / indicate one or more RSRPs where each of the one or more RSRPs is associated with a beam of the number of predicted / selected beams for the each time instance of the plurality of time instances. In an example of FIG. 11 and referring to FIG. 12A, the inference report may comprise / indicate the number of min [K1, Kmax] , CRIs / SSBRIs of the predicted / selected number of min [K1, Kmax] beams in the beam set A, and / or corresponding predicted RSRPs of the predicted / selected number of min [K1, Kmax] beams for the first time instance. The inference report may further comprise / indicate the number of min [K2, Kmax] , CRIs / SSBRIs of the predicted / selected number of min [K2, Kmax] beams in the beam set A, and / or corresponding predicted RSRPs of the predicted / selected number of min [K2, Kmax] beams for the first time instance. In an example, predicted RSRPs in an inference report may be quantized by one or more quantization steps and within a quantization range for reducing signaling overhead to deliver the predicted RSRPs in the inference report. In an example of FIG. 12A, the one or more RSRPs associated with the number of predicted / selected beams using AI / ML approaches may be quantized RSRPs.
[0091] In an example of FIG. 12B, an inference report (e.g., for a plurality of time instances for BM-Case 2 using AI / ML approaches) may comprise (e.g., be split into) a first part and a second part. The first part of the inference report may have a fixed length (e.g., message size) . A base station may be able to blind decode the first part of the inference report based on the fixed length of the first part. The second part of the inference report may have a variable length. The variable length of the second part may be based on how much information to be delivered using the second part.
[0092] In an example of FIG. 12B, a first part of an inference report for a plurality of time instances may comprise / indicate a reference quantized RSRP of a predicted beam. For example, a wireless device may determine / select the reference quantized RSRP based on the predicted beam, associated with the selected reference quantized RSRP, being for an earlier time instance among the plurality of time instances. For example, a wireless device may determine / select the reference quantized RSRP based on the predicted beam, associated with the selected reference quantized RSRP, having a lowest / highest beam index / ID among a number of predicted beams for a time instance of the plurality of time instances. For example, a wireless device may determine / select the reference quantized RSRP based on the selected reference quantized RSRP having a biggest value among the quantized RSRPs of a number of predicted beams for a time instance of the plurality of time instances. For example, a wireless device may determine / select the reference quantized RSRP based on the selected reference quantized RSRP having a biggest value among the quantized RSRPs of a number of predicted beams for the plurality of time instance (e.g., for all of the plurality of time instances) . For example, a wireless device may determine / select the reference quantized RSRP based on the predicted beam, associated with the selected reference quantized RSRP, having highest confidence / probability to be a top-1 beam (e.g., with strongest RSRP) among a number of predicted beams for a time instance of the plurality of time instances. For example, a wireless device may determine / select the reference quantized RSRP based on the predicted beam, associated with the selected reference quantized RSRP, having highest confidence / probability to be a top-1 beam (e.g., with strongest RSRP) among the quantized RSRPs of a number of predicted beams for the plurality of time instance (e.g., for all of the plurality of time instances) . In an example of FIG. 12B, the first part of the inference report for the plurality of time instances may comprise / indicate a length of a second part of the inference report. In an example of FIG. 12B, the second part of the inference report may comprise / indicate differential RSRPs, compared with the selected reference quantized RSRP in the first part, for one or more predicted beams for the plurality of time instances.
[0093] In an example and referring to FIG. 11, FIG. 12A and FIG. 12B, a base station may configure, to a wireless device, two (e.g., Nmax=2) time instances to be reported / included / covered in a single inference report. The base station may further configure, to the wireless device, up to three (e.g., Kmax=3) predicted beams to be reported for a time instance of the two time instances in the single inference report. The wireless device may predict / select a beam 1 (e.g., K1=1) for a first time instance of the two time instances based on a first predicted RSRP of the beam 1 being higher than or equal to a RSRP threshold during / for the first time instance. The wireless device may predict / select a beam 2, a beam 3, a beam 4, and a beam 5 (e.g., K2=4) for a second time instance of the two time instances based on predicted RSRPs of the beam 2 to the beam 5 being higher than or equal to the RSRP threshold during / for the second time instance. The wireless device may transmit the single inference report to the base station. The single inference report may comprise / indicate the beam 1 and / or the first predicted RSRPs for the first time instance based on min [K1, Kmax] =1. The single inference report may comprise / indicate the beam 2, the beam 3, and the beam 4, and / or corresponding predicted RSRPs for the second time instance but not the beam 5 and / or corresponding predicted RSRPs of the beam 5 based on min [K2, Kmax] =3. Referring to FIG. 12B, the wireless device may quantize predicted RSRPs. For example, the wireless device may select / determine a quantized RSRP of the first predicted RSRP of the beam 1 as a reference quantized RSRP, based on the quantized RSRP of the first predicted RSRP being the strongest beam (e.g., having highest RSRP) for the first time instance. For example, the wireless device may select / determine a quantized RSRP of a third predicted RSRP of the beam 3 as a reference quantized RSRP, based on the quantized RSRP of the third predicted RSRP being the strongest beam (e.g., having highest RSRP) during the two time instances. A first part of the inference report may comprise / indicate the reference quantized RSRP. The first part of the inference report may further comprise / indicate a length of a second part of the inference report. The second part of the inference report may comprise / indicate differential RSRPs compared with the reference quantized RSRP for the beam 1 for the first time instance and for the beam 2, the beam 3 and the beam 4 for the second time instance. Beam indexes / IDs of the predicted beam 1, beam 2, beam 3, and / or beam 4 may be indicated only in the first part of the inference report and / or only in the second part of the inference report. The beam indexes / IDs of the predicted beam 1, beam 2, beam 3, and / or beam 4 may be duplicated in both the first part of the inference report and the second part of the inference report.
[0094] FIG. 13A and FIG. 13B illustrate example timing for reference signal measurement and prediction using AI / ML approaches. Referring to FIG. 11, a wireless device may transmit an inference report in uplink, for a number of N time instances (e.g., slot interval) , in slot n, where N≥1. Each of the number of N time instances may comprise a number of d consecutive time slots, where d≥1. In an example, N∈ {1, 2, 4, 8} . In an example, the number d may be configured by higher layer parameter (e.g., parameter in a RRC message) . If the wireless device is configured with a periodic or semi-persistent CSI-RS resource set for channel measurement, the number of d may be equal to a periodicity of CSI-RS resource (e.g., RS resource (s) in FIG. 13A) of the periodic / semi-persistent CSI-RS resource set. A start timing of the number of N time instances may be slot L, where L=n+δ, and a slot offset δ∈ {-nCSI_ref, 0, 1, 2} may be configured by higher layer parameter delta. The wireless device may generate the inference report transmitted in slot n based on measurements of RS resources before the slot (n-nCSI_ref) , which may be a CSI reference resource. A duration (e.g., an observation window) for measuring RS resources for the inference report may end at the slot (n-nCSI_ref) (e.g., CSI reference resource) .
Claims
1.A method comprising:receiving, by a wireless device from a base station, one or more radio resource control (RRC) messages comprising:a first parameter indicating a number of time instances to be reported in an inference report of an artificial intelligence machine learning (AI / ML) model, wherein the number of time instances comprise a first time instance and a second time instance; anda second parameter indicating a first number of predicted downlink beams, for a time instance of the number of time instances, for the inference report of the AI / ML model;predicting, using the AI / ML model:a second number of downlink beams for the first time instance, wherein the second number being smaller than or equal to the first number; anda third number of downlink beams for the second time instance, wherein the third number being smaller than or equal to the first number; andtransmitting, to the base station, the inference report for the number of time instances indicating the predicted second number of downlink beams for the first time instance and the predicted third number of downlink beams for the second time instance.2.A method comprising:receiving, by a wireless device from a base station, one or more configuration parameters indicating:a number of time instances comprising a first time instance and a second time instance; anda first number of beams for a time instance of the number of time instances;predicting based on the first number:a second number of beams for the first time instance; anda third number of beams for the second time instance; andtransmitting an inference report for the number of time instances indicating the second number of beams and the third number of beams.3.The method of claim 2, further comprising receiving one or more radio resource control (RRC) messages comprising the one or more configuration parameters.4.The method of claim 3, wherein the one or more configuration parameters comprises a first parameter indicating the number of time instances to be reported in the inference report of an artificial intelligence machine learning (AI / ML) model.5.The method of claim 4, wherein the one or more configuration parameters comprises a second parameter indicating the first number of beams for the time instance of the number of time instances for the inference report of the AI / ML model.6.The method of claim 5, wherein the predicting the second number of beams is based on the second number being smaller than or equal to the first number.7.The method of claim 6, wherein the predicting the third number of beams is based on the third number being smaller than or equal to the first number.8.The method of claim 7, wherein the transmitting further comprising transmitting, to the base station, the inference report for the number of time instances indicating the predicted second number of beams for the first time instance and the predicted third number of beams for the second time instance.9.The method of claim 8, wherein at least one of the first number of beams, the second number of beams and the third number of beams are downlink beams.10.The method of any one of claims 1 to 9, wherein the number of time instances are future time instances.11.The method of claim 2, wherein the one or more configuration parameters further indicates:a first beam set comprising one or more first beams; anda second beam set comprising one or more second beams.12.The method of claim 11, further comprising receiving one or more reference signals via the first beam set, wherein each of the one or more reference signals is received via a corresponding beam of the first beam set.13.The method of claim 12, wherein the predicting further comprising predicting, from the second beam set, the second number of beams based on a measurement of the one or more reference signals via the first beam set within a time duration.14.The method of claim 13, wherein the predicting further comprising predicting, from the second beam set, the third number of beams based on the measurement of the one or more reference signals via the first beam set within the time duration.15.The method of claim 14, wherein the time duration is a past time duration.16.The method of claim 11, wherein the first beam set is same as the second beam set.17.The method of claim 11, wherein the first beam set is not a subset of the second beam set.18.The method of claim 11, wherein the first beam set is a subset of the second beam set.19.The method of claim 14, wherein the one or more configuration parameters further indicates a reference signal received power (RSRP) threshold.20.The method of claim 19, wherein the predicting is further based on:predicted RSRPs of the second number of beams being higher than or equal to the RSRP threshold; andpredicted RSRPs of the third number of beams being higher than or equal to the RSRP threshold.21.The method of claim 14, wherein the one or more configuration parameters further indicates a gap value in decibel (dB) compared with a reference signal received power (RSRP) of a beam.22.The method of claim 21, wherein the predicting is further based on:predicted RSRPs of the second number of beams being within the gap value in dB compared with the RSRP of the beam; andpredicted RSRPs of the third number of beams being within the gap value in dB compared with the RSRP of the beam.23.The method of claim 22, wherein the beam is a beam with the highest predicted RSRP among the second number of beams for the first time instance.24.The method of claim 22, wherein the beam is a beam with the highest predicted RSRP among the third number of beams for the second time instance.25.The method of claim 22, wherein the beam is a beam with the highest predicted RSRP in the second beam set.26.The method of claim 14, wherein the one or more configuration parameters further indicates a probability threshold.27.The method of claim 26, wherein the predicting is further based on:predicted probabilities of the second number of beams, to be a top-1 beam during the first time instance, being higher than or equal to the probability threshold; andpredicted probabilities of the third number of beams, to be a top-1 beam during the second time instance, being higher than or equal to the probability threshold.28.The method of claim 27, wherein the top-1 beam during the first time instance is a beam with the highest RSRP in the second beam set during the first time instance.29.The method of claim 27, wherein the top-1 beam during the second time instance is a beam with the highest RSRP in the second beam set during the second time instance.30.The method of claim 2, wherein the predicting further comprising predicting reference signal received powers (RSRPs) of the second number of beams for the first time instance and RSRPs of the third number of beams for the second time instance.31.The method of claim 30, further comprising quantizing, based on one or more quantization steps and a quantization range, the RSRPs of the second number of beams and the RSRPs of the third number of beams.32.The method of claim 31, wherein the inference report further indicates the quantized RSRPs of the second number of beams and the quantized RSRPs of the third number of beams.33.The method of claim 32, further comprising selecting a quantized RSRP as a reference quantized RSRP based on at least one of:a beam, associated with the selected reference quantized RSRP, being for an earlier time instance of the first time instance and the second time instance;a beam, associated with the selected reference quantized RSRP, being for the earliest time instance in time among the number of time instances;a beam, associated with the selected reference quantized RSRP, having a lowest beam index among the second number of beams;a beam, associated with the selected reference quantized RSRP, having a highest beam index among the second number of beams;the selected reference quantized RSRP having a biggest value among the quantized RSRPs of the second number of beams;the selected reference quantized RSRP having a biggest value among the quantized RSRPs of the second number of beams and the quantized RSRPs of the third number of beams;a beam, associated with the selected reference quantized RSRP, having highest confidence to be a top-1 beam among the second number of beams;a beam, associated with the selected reference quantized RSRP, having highest confidence to be a top-1 beam among the second number of beams and the third number of beams;a beam, associated with the selected reference quantized RSRP, having highest probability to be a top-1 beam among the second number of beams; anda beam, associated with the selected reference quantized RSRP, having highest probability to be a top-1 beam among the second number of beams and the third number of beams.34.The method of claim 33, wherein the inference report comprises a value of the reference quantized RSRP and differential values, compared with the value of the reference quantized RSRP, for the quantized RSRPs of the second number of beams and the quantized RSRPs of the third number of beams.35.The method of claim 34, wherein the inference report indicates:for the first time instance, at least one of:channel state information resource indicators (CRIs) of the second number of beams;synchronization signal block resource indicators (SSBRIs) of the second number of beams; andthe quantized RSRPs of the second number of beams; andfor the second time instance, at least one of:CRIs of the third number of beams;SSBRIs of the third number of beams; andthe quantized RSRPs of the third number of beams.36.The method of claim 35, wherein the inference report further indicates a fourth number indicating a subset of the number of time instances, wherein the fourth number of time instances comprises the first time instance and the second time instance.37.The method of claim 35, wherein the inference report comprises:a first part with a fixed length for blind decoding; anda second part with a length based on at least one of:the number of time instances to be reported in the inference report;the first number of beams;the second number of beams; andthe third number of beams.38.The method of claim 37, wherein the first part of the inference report indicates the reference quantized RSRP and the length of the second part.39.The method of claim 38, wherein the second part of the inference report indicates the differential values of the quantized RSRPs of the second number of beams and the quantized RSRPs of the third number of beams.40.The method of claim 2, wherein the second number equals the first number based on the first time instance being a earliest time instance of the number of time instances.41.A wireless device comprising:one or more processors; andmemory storing instructions that, when executed by the one or more processors, cause the wireless device to perform the method of any one of claims 1 to 40.42.A non-transitory computer-readable storage medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform the method of any one of claims 1 to 40.43.A system comprising:a base station comprising one or more first processors; and first memory storing first instructions that, when executed by the one or more first processors, cause the base station to:transmit one or more configuration parameters indicating:a number of time instances comprising a first time instance and a second time instance; anda first number of beams for a time instance of the number of time instances; andreceive an inference report for the number of time instances indicating a second number of beams and a third number of beams; anda wireless device comprising one or more second processors; and second memory storing second instructions that, when executed by the one or more second processors, cause the wireless device to:predict based on the first number,the second number of beams for the first time instance; andthe third number of beams for the second time instance; andtransmit the inference report.44.A method comprising:receiving, by a wireless device from a base station, one or more configuration parameters indicating:a number of time instances comprising a first time instance and a second time instance; anda first number of beams for a time instance of the number of time instances;predicting based on the first number:a second number of downlink beams for the first time instance; anda third number of downlink beams for the second time instance; andtransmitting, to the base station, an inference report for the number of time instances indicating:for the first time instance, a smaller number, of the first number and the second number, of the predicted second number of downlink beams; andfor the second time instance, a smaller number, of the first number and the third number, of the predicted third number of downlink beams.45.A method comprising:receiving, by a wireless device from a base station, one or more configuration parameters indicating:a number of time instances comprising a first time instance and a second time instance; anda first number of beams for a time instance of the number of time instances;predicting based on the first number:a second number of downlink beams for the first time instance; anda third number of downlink beams for the second time instance;selecting, from the second number and the third number, a number based on comparison of the second number and the third number; andtransmitting, to the base station, an inference report for the number of time instances indicating the selected number of the predicted second number of downlink beams for the first time instance and the selected number of the predicted third number of downlink beams for the second time instance.46.The method of claim 45, wherein the selecting the number is based on the number being a smaller number of the second number and the third number.