CSI reporting based on ML technology

Abstract

The present disclosure describes CSI reporting based on ML technology. The UE receives one or more CSI-RSs from a network entity to generate one or more CSI reports. When at least one CSI report is based on an ML model, the UE may compress the at least one CSI report by the selected broadband precoder using one or more subband feature vectors of one or more precoded estimated subband channels. The UE may also omit at least a portion of the at least one CSI report when the total payload size of the one or more CSI reports exceeds a threshold. The UE may be configured for parallel processing of a plurality of CSI reports based on the ML model and / or parallel processing of a first CSI report based on the ML model and a second CSI report not based on the ML model.

Description

Technical Field

[0001] The present disclosure relates generally to wireless communications, and more particularly, to channel state information (CSI) reporting based on machine learning (ML) techniques. Background Art

[0002] The 3rd Generation Partnership Project (3GPP) specifies a radio interface called fifth generation (5G) New Radio (NR) (5G NR). The architecture of a 5G NR wireless communication system may include a 5G core (5GC) network, a 5G radio access network (5G-RAN), user equipment (UE), etc. Compared to other types of wireless communication systems, the 5G NR architecture may provide increased data rates, reduced latency, and / or increased capacity.

[0003] Wireless communication systems can generally be configured to provide various telecommunication services (e.g., telephone, video, data, messaging, broadcast, etc.) based on multiple access technologies (such as orthogonal frequency division multiple access (OFDMA) technologies) that support communication with multiple UEs. With the development of mobile broadband technology, improvements in mobile broadband have helped to continue the advancement of such technologies. For example, user equipment (UE) and base stations can support more antenna configurations and multiple connections. However, one consequence is that channel state information (CSI) reports have become larger and more complex. Summary of the invention

[0004] The following presents a simplified overview of one or more aspects in order to provide a basic understanding of such aspects. This overview is not an extensive review of all contemplated aspects. This overview neither identifies the key or important elements of all aspects, nor describes the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to a more detailed description presented later.

[0005] A user equipment (UE) may receive one or more channel state information-reference signals (CSI-RS) from a network entity (such as a base station) to generate one or more channel state information (CSI) reports. In some cases, the UE may generate at least one CSI report based on a machine learning (ML) model. Thus, the UE may include a UE capability for processing multiple CSI reports based on the ML model in parallel and / or processing a first CSI report based on the ML model and a second CSI report not based on the ML model in parallel.

[0006] The UE may use the ML model to reduce the complexity of at least one CSI report by compressing the at least one CSI report using one or more subband eigenvectors. Additionally or alternatively, the UE may utilize the ML model to omit at least a portion of the at least one CSI report when a total payload size of the one or more CSI reports exceeds a threshold.

[0007] According to some aspects, a UE receives a CSI-RS from a network entity and selects a wideband precoder to compress a CSI report based on a measurement of the CSI-RS. The selection of the wideband precoder is based on the entire bandwidth used to receive the CSI-RS. The UE compresses the CSI report using one or more subband eigenvectors of one or more precoded estimated subband channels through the selected wideband precoder. The UE transmits / sends a first precoding matrix indicator (PMI) indicating a wideband precoder and a second PMI indicating one or more subband eigenvectors to the network entity.

[0008] The network entity transmits a CSI-RS to the UE to select a wideband precoder to compress a CSI report based on a measurement of the CSI-RS. The selection of the wideband precoder is based on the entire bandwidth used to transmit the CSI-RS. The network entity receives a CSI report from the UE, the CSI report compressed by the selected wideband precoder using one or more subband eigenvectors of one or more precoded subband channels, a first PMI indicating the wideband precoder, and a second PMI indicating the one or more subband eigenvectors. The network entity decompresses the CSI report using the first PMI and the second PMI as inputs to an ML model, the ML model outputting a decompressed CSI report.

[0009] According to some aspects, a UE receives one or more CSI-RS from a network entity to transmit one or more CSI reports to the network entity. At least one of the one or more CSI reports is based on an ML model. When a total payload size of the one or more CSI reports exceeds a threshold, the UE omits at least a portion of the at least one CSI report based on the ML model output. The UE transmits the one or more CSI reports to the network entity, wherein at least a portion of the at least one CSI report is omitted from the transmission.

[0010] According to some aspects, a UE receives a configuration for a CSI report processing procedure from a network entity. The configuration is for at least one of: (a) processing multiple CSI reports using an ML model, or (b) processing a first set of CSI reports using an ML model without processing a second set of CSI reports. The UE receives multiple CSI-RSs from the network entity. Measurement of the multiple CSI-RSs is based on UE capabilities for the CSI report processing procedure. The UE transmits a CSI report to the network entity based on the UE capabilities for the CSI report processing procedure.

[0011] The network entity transmits a configuration for a CSI report processing procedure to the UE. The configuration is for at least one of: (a) processing a plurality of CSI reports using an ML model, or (b) processing a first set of CSI reports using an ML model without processing a second set of CSI reports. The network entity transmits a plurality of CSI-RSs to the UE. The UE processes measurements of the plurality of CSI-RSs based on UE capabilities for the CSI report processing procedure. The network entity communicates with the UE based on the received CSI reports.

[0012] To accomplish the foregoing and related purposes, one or more aspects correspond to the features described below and particularly pointed out in the claims. One or more aspects may be implemented by any of an apparatus, a method, a component for performing the method, and / or a non-transitory computer-readable medium. The following description and the accompanying drawings set forth in detail certain illustrative features of one or more aspects. However, these features indicate only a few of the various ways in which the principles of the various aspects may be employed. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A diagram showing a wireless communication system including multiple network entities communicating over multiple cells.

[0014] Figure 2 A diagram showing an example machine learning (ML) based channel state information (CSI) compression.

[0015] Figure 3A A first timing diagram illustrating a first CSI processing unit duration associated with periodic / semi-persistent CSI reporting.

[0016] Figure 3B A second timing diagram illustrating a second CSI processing unit duration associated with aperiodic CSI reporting.

[0017] Figure 4 Diagram showing the general process of low-complexity machine learning based CSI compression.

[0018] Figure 5 is a signaling diagram illustrating the process of low-complexity machine learning (ML) based CSI reporting.

[0019] Figure 6 Method diagram showing User Equipment (UE) behavior for CSI compression and reporting.

[0020] Figure 7 Method diagram showing the behavior of network entities for CSI decompression.

[0021] FIG. 8A to FIG. 8D Table showing an example ML CSI report.

[0022] Fig. 9A A diagram illustrating the general process of CSI reporting with respect to CSI omission rules of some embodiments.

[0023] Fig. 9B A diagram illustrating alternative procedures for CSI reporting with respect to CSI omission rules of some other embodiments.

[0024] Fig.10 is based on Fig. 9A FIG. 1 is a method diagram of UE behavior for reporting artificial intelligence (AI) / non-AI based CSI using CSI omission rules.

[0025] Fig.11 is based on Fig. 9A The process in the method diagram of network entity behavior for reporting AI / non-AI based CSI using CSI omission rules.

[0026] Fig.12 is based on Fig. 9B FIG. 5 is a method diagram of UE behavior for reporting AI / non-AI based CSI using CSI omission rules.

[0027] Fig.13 is based on Fig. 9B The process in the method diagram of network entity behavior for reporting AI / non-AI based CSI using CSI omission rules.

[0028] FIG. 14A to FIG. 14C Table showing AI-based CSI reporting.

[0029] Fig.15 is a signaling diagram illustrating the process of a CSI processing unit (CPU) framework for AI-based CSI processing.

[0030] Fig.16 is a method diagram of UE behavior for a CPU framework for AI-based CSI processing.

[0031] Fig.17 is a method diagram of network entity behaviors for a CPU framework for AI-based CSI processing.

[0032] Fig.18 A timing diagram showing Type 2 CPU occupancy rules.

[0033] FIG. 19A to FIG. 19B An example timing diagram showing CPU occupancy rules for AI-based periodic / semi-persistent CSI reporting.

[0034] FIG. 20A to FIG. 20B An example timing diagram showing CPU occupancy rules for AI-based aperiodic CSI reporting.

[0035] Fig.21 A signaling diagram showing CSI reporting based on minimum processing delay.

[0036] Fig. 22 is a diagram illustrating an example of a hardware implementation of an example UE device.

[0037] Fig.23 is a diagram illustrating an example of a hardware implementation of one or more example network entities. DETAILED DESCRIPTION

[0038] Figure 1 A diagram 100 of a wireless communication system associated with a plurality of cells 190 is shown. The wireless communication system includes a user equipment (UE) 102 and base stations 104, some of which 104a include an aggregated base station architecture and other base stations 104b include a decomposed base station architecture. The aggregated base station architecture includes a radio unit (RU) 106, a distributed unit (DU) 108, and a centralized unit (CU) 110, which are configured to utilize a radio protocol stack physically or logically integrated within a single radio access network (RAN) node. The decomposed base station architecture utilizes a protocol stack physically or logically distributed between two or more units (e.g., RU 106, DU 108, CU 110). For example, CU 110 is implemented within a RAN node, and one or more DUs 108 may be co-located with CU 110, or alternatively, may be geographically or virtually distributed in one or more other RAN nodes. DU 108 may be implemented to communicate with one or more RUs 106. Each of the RU 106, DU 108, and CU 110 may be implemented as a virtual unit, such as a virtual radio unit (VRU), a virtual distributed unit (VDU), or a virtual central unit (VCU).

[0039] The operation and / or network design of the base station 104 can be based on the aggregation characteristics of the base station functions. For example, a decomposed base station architecture is utilized in an integrated access backhaul (IAB) network, an open radio access network (O-RAN) network, or a virtualized radio access network (vRAN) (which may also be referred to as a cloud radio access network (C-RAN)). Decomposition can include distributing functions between two or more units located at various physical locations, and virtually distributing the functions of at least one unit, which can achieve flexibility in network design. Various units of the decomposed base station architecture or decomposed RAN architecture can be configured to communicate with at least one other unit by wire or wireless communication. For example, CU 110a communicates with DU108a-108b via corresponding mid-haul links based on F1 interfaces. DU 108a-108b can communicate with RU 106a and RU 106b-106c via corresponding front-haul links, respectively. RU 106a-106c can communicate with corresponding UE 102a-102c and 102s via one or more radio frequency (RF) access links based on Uu interface. In an example, multiple RU 106 and / or base station 104 can provide services for UE 102 at the same time, such as the access link of RU 106a of cell 190a and the UE 102a of cell 190e served by base station 104a of cell 190e at the same time.

[0040] One or more CUs 110, such as CU 110a or CU 110d, may communicate directly with the core network 120 via a backhaul link. For example, CU 110d communicates with the core network 120 via a backhaul link based on a next generation (NG) interface. One or more CUs 110 may also communicate indirectly with the core network 120 via one or more decomposed base station units, such as a near real-time RAN intelligent controller (RIC) 128 via an E2 link and a service management and orchestration (SMO) framework 116 that may be associated with a non-real-time RIC 118. The near real-time RIC 128 may communicate with the SMO framework 116 and / or the non-real-time RIC 118 via an A1 link. The SMO framework 116 and / or the non-real-time RIC 118 may also communicate with an open cloud (O-cloud) 130 via an O2 link. One or more CUs 110 may further communicate with each other via a backhaul link based on an Xn interface. For example, the CU 110d of the base station 104a communicates with the CU 110a of the base station 104b via a backhaul link based on an Xn interface. Similarly, the base station 104a of the cell 190e can communicate with the CU 110a of the base station 104b via a backhaul link based on an Xn interface.

[0041] RU 106, DU 108 and CU 110 and near real-time RIC 128, non-real-time RIC 118 and / or SMO framework 116 may include (or may be coupled to) one or more interfaces configured to transmit or receive information / signals via a wired or wireless transmission medium. Base station 104 or any one of the one or more decomposed base station units may be configured to communicate with one or more other base stations 104 or one or more other decomposed base station units via a wired or wireless transmission medium. In an example, a processor, memory and / or controller associated with executable instructions of the interface may be configured to provide communication between base station 104 and / or one or more decomposed base station units via a wired or wireless transmission medium. For example, a wired interface may be configured to transmit or receive information / signals via a wired transmission medium (such as a fronthaul link between RU 106d and baseband unit (BBU) 112 of cell 190d, or more specifically, a fronthaul link between RU 106d and DU 108d). The BBU 112 includes the DU 108d and the CU 110d, which may also have a wired interface configured between the DU 108d and the CU 110d to transmit or receive information / signals between the DU 108d and the CU 110d based on a medium-range link. In a further example, a wireless interface that may include a receiver, a transmitter, or a transceiver (such as an RF transceiver) may be configured to transmit or receive information / signals via a wireless transmission medium, such as information transmitted between the RU 106a of the cell 190a and the base station 104a of the cell 190e via the cross-cell communication beam of the RU 106a and the base station 104a.

[0042] One or more high-level control functions (such as functions related to radio resource control (RRC), packet data convergence protocol (PDCP), service data adaptation protocol (SDAP), etc.) can be hosted at CU 110. Each control function can be associated with an interface for transmitting signals based on one or more other control functions hosted at CU 110. User plane functions (such as central unit-user plane (CU-UP) functions), control plane functions (such as central unit-control plane (CU-CP) functions), or a combination thereof can be implemented based on CU 110. For example, CU 110 may include one or more CU-UP processes and / or one or more CU-CP processes. When implemented in an O-RAN configuration, the CU-UP function can be based on bidirectional communication with the CU-CP function via an interface (such as an E1 interface (not shown)).

[0043] The CU 110 may communicate with the DU 108 for network control and signal transmission. The DU 108 is a logical unit of the base station 104 that is configured to perform one or more base station functions. For example, the DU 108 may control the operation of one or more RUs 106. One or more of the following may be hosted at the DU 108: a radio link control (RLC) layer, a media access control (MAC) layer, or one or more higher physical (PHY) layers, such as forward error correction (FEC) modules for encoding / decoding, scrambling, modulation / demodulation, etc. The DU 108 may host such functions based on the functional division of the DU 108. The DU 108 may similarly host one or more lower PHY layers, where each lower layer or module may be implemented based on an interface for communicating with other layers and modules hosted at the DU 108, or based on a control function hosted at the CU 110.

[0044] The RU 106 may be configured to implement lower layer functions. For example, the RU 106 is controlled by the DU 108 and may correspond to a logical node hosting RF processing functions or lower layer PHY functions, such as performing fast Fourier transform (FFT), inverse FFT (iFFT), digital beamforming, physical random access channel (PRACH) extraction and filtering, etc. The functions of the RU 106 may be based on functional partitioning, such as lower layer functional partitioning.

[0045] The RU 106 may transmit or receive over-the-air (OTA) communications with one or more UEs 102. For example, the RU 106b of the cell 190b communicates with the UE 102b of the cell 190b via the first communication beam set 132 of the RU 106b and the second communication beam set 134 of the UE 102b, which may correspond to inter-cell communication beams or inter-cell communication beams. Both real-time and non-real-time features of the control plane and user plane communications of the RU 106 may be controlled by the associated DU 108. Thus, the DU 108 and the CU 110 may be used in a cloud-based RAN architecture (such as a vRAN architecture), and the SMO framework 116 may be used to support non-virtualized and virtualized RAN network elements. For non-virtualized network elements, the SMO framework 116 may support the deployment of dedicated physical resources for RAN coverage, where the dedicated physical resources may be managed through an operation and maintenance interface (such as an O1 interface). For virtualized network elements, the SMO framework 116 can interact with a cloud computing platform (such as O-cloud 130) via an O2 link (e.g., a cloud computing platform interface) to manage the network elements. The virtualized network elements may include, but are not limited to, RU 106, DU 108, CU 110, near real-time RIC 128, etc.

[0046] The SMO framework 116 may be configured to communicate directly with one or more RUs 106 using an O1 link. The non-real-time RIC 118 of the SMO framework 116 may also be configured to support the functionality of the SMO framework 116. For example, the non-real-time RIC 118 implements logic functions that are capable of controlling non-real-time RAN features and resources, features / applications of the near real-time RIC 128, and / or artificial intelligence / machine learning (AI / ML) processes. The non-real-time RIC 118 may communicate (or couple) with the near real-time RIC 128, such as through an A1 interface. The near real-time RIC 128 may implement logic functions that are capable of controlling near real-time RAN features and resources based on data collection and interaction through an E2 interface (such as an E2 interface between the near real-time RIC 128 and the CU 110a and the DU 108b).

[0047] The non-real-time RIC 118 may receive parameters or other information from an external server to generate an AI / ML model for deployment in the near-real-time RIC 128. For example, the non-real-time RIC 118 receives parameters or other information from the O-cloud 130 via the O2 link to deploy the AI / ML model to the real-time RIC 128 via the A1 link. The near-real-time RIC 128 may utilize the parameters and / or other information received from the non-real-time RIC 118 or the SMO framework 116 via the A1 link to perform near-real-time functions. The near-real-time RIC 128 and the non-real-time RIC 115 may be configured to adjust the performance of the RAN. For example, the non-real-time RIC 116 monitors patterns and long-term trends to improve the performance of the RAN. The non-real-time RIC 116 may also deploy the AI / ML model through the SMO framework 116 for implementing corrective actions, such as initiating reconfiguration of the O1 link or instructing the management process of the A1 link.

[0048] Any combination of RU 106, DU 108, and CU 110, or any reference thereto individually, may correspond to base station 104. Thus, base station 104 may include at least one of RU 106, DU 108, or CU 110. Base station 104 provides UE 102 with access to core network 120. That is, base station 104 may relay communications between UE 102 and core network 120. Base station 104 may be associated with a macro cell of a high-power cellular base station and / or a small cell of a low-power cellular base station. For example, cell 190e corresponds to a macro cell, and cells 190a-190d may correspond to a small cell. Small cells include femto cells, micro cells, micro cells, and the like. A cell structure including at least one macro cell and at least one small cell may be referred to as a "heterogeneous network."

[0049] Transmissions from the UE 102 to the base station 104 / RU 106 are referred to as uplink (UL) transmissions, while transmissions from the base station 104 / RU 106 to the UE 102 are referred to as downlink (DL) transmissions. Uplink transmissions may also be referred to as reverse link transmissions, while downlink transmissions may also be referred to as forward link transmissions. For example, the RU 106 d utilizes the antenna of the base station 104 a of the cell 190 d to transmit downlink / forward link communications to the UE 102 d, or receive uplink / reverse link communications from the UE 102 d, based on a Uu interface associated with an access link between the UE 102 d and the base station 104 a / RU 106 d.

[0050] The communication link between UE 102 and base station 104 / RU 106 can be based on multiple-input multiple-output (MIMO) antenna technology, including spatial multiplexing, beamforming and / or transmit diversity. The communication link can be associated with one or more carriers. UE 102 and base station 104 / RU 106 can utilize Y MHz (e.g., 5 MHz, 10 MHz, 15 MHz, 20 MHz, 100 MHz, 400 MHz, etc.) spectrum bandwidth allocated per carrier in up to a total of Yx MHz carrier aggregation, where x component carriers (CCs) are used for communication in each direction of the uplink direction and the downlink direction. The carriers may be adjacent to each other along the spectrum, or may not be adjacent to each other. In an example, uplink carriers and downlink carriers may be allocated in an asymmetric manner, and more or fewer carriers may be allocated for uplink or downlink. A component carrier may include a primary component carrier and one or more secondary component carriers. The primary component carrier may be associated with a primary cell (PCell), and the secondary component carrier may be associated with a secondary cell (SCell).

[0051] Some UEs 102 (such as UEs 102a and 102s) may perform device-to-device (D2D) communications via a sidelink. For example, a sidelink communication / D2D link utilizes a spectrum of a wireless wide area network (WWAN) associated with uplink and downlink communications. The sidelink communication / D2D link may also use one or more sidelink channels, such as a physical sidelink broadcast channel (PSBCH), a physical sidelink discovery channel (PSDCH), a physical sidelink shared channel (PSSCH), and / or a physical sidelink control channel (PSCCH) to transmit information between UEs 102a and 102s. Such sidelink / D2D communications may be performed via various wireless communication systems, such as wireless fidelity (Wi-Fi) systems, Bluetooth systems, long term evolution (LTE) systems, new radio (NR) systems, and the like.

[0052] The electromagnetic spectrum is often subdivided into different categories, bands, channels, etc. based on different frequencies / wavelengths associated with the electromagnetic spectrum. Fifth Generation (5G) NR is often associated with two operating bands, referred to as Frequency Range 1 (FR1) and Frequency Range 2 (FR2). FR1 ranges from 410 MHz – 7.125 GHz, and FR2 ranges from 24.25 GHz – 52.6 GHz. FR1 is often referred to as the “sub-6 GHz” band, although a portion of FR1 is actually greater than 6 GHz. In contrast, FR2 is often referred to as the “millimeter wave” (mmW) band. FR2 is not the same as the “extremely high frequency” (EHF) band, but is an approximate subset of that band, the EHF band ranges from 30 GHz – 300 GHz, and is sometimes also referred to as the “millimeter wave” band. Frequencies between FR1 and FR2 are often referred to as “mid-band” frequencies. The operating band of mid-band frequencies may be referred to as frequency range 3 (FR3), which ranges from 7.125 GHz – 24.25 GHz. The frequency bands within FR3 may include characteristics of FR1 and / or FR2. Thus, the features of FR1 and / or FR2 may be extended to mid-band frequencies. Higher operating bands have been identified to extend 5G NR communications above 52.6 GHz associated with the upper limit of FR2. Three of these higher operating bands include FR2-2 (ranging from 52.6 GHz – 71 GHz), FR4 (ranging from 71 GHz – 114.25 GHz), and FR5 (ranging from 114.25 GHz – 300 GHz). The upper limit of FR5 corresponds to the upper limit of the EHF band. Therefore, unless otherwise expressly stated herein, the term “below 6 GHz” may refer to frequencies less than 6 GHz, frequencies within FR1, or frequencies that may include mid-band frequencies. Further, unless otherwise expressly stated herein, the term "millimeter wave" or mmW refers to frequencies that may include mid-band frequencies, frequencies that may be within FR2, FR4, FR2-2 and / or FR5, or frequencies that may be within the EHF band.

[0053] UE 102 and base station 104 / RU 106 may each include multiple antennas. Multiple antennas may correspond to antenna elements, antenna panels, and / or antenna arrays that may facilitate beamforming operations. For example, RU 106b transmits a downlink beamformed signal to UE 102b based on a first beam set 132 in one or more transmit directions of RU 106b. UE 102b may receive a downlink beamformed signal from RU 106b based on a second beam set 134 in one or more receive directions of UE 102b. In a further example, UE 102b may also transmit an uplink beamformed signal to RU 106b based on a second beam set 134 in one or more transmit directions of UE 102b. RU 106b may receive an uplink beamformed signal from UE 102b in one or more receive directions of RU 106b. UE 102b may perform beam training to determine the optimal receive and transmit directions of the beamformed signal. The transmit and receive directions of the UE 102 and the base station 104 / RU 106 may be the same, or may be different. In a further example, a beamformed signal may be transmitted between the first base station 104a and the second base station 104b. For example, the RU 106a of the cell 190a may transmit a beamformed signal to the base station 104a of the cell 190e based on the RU beam set 136 in one or more transmit directions of the RU 106a. The base station 104a of the cell 190e may receive a beamformed signal from the RU 106a based on the base station beam set 138 in one or more receive directions of the base station 104a. Similarly, the base station 104a of the cell 190e may transmit a beamformed signal to the RU 106a based on the base station beam set 138 in one or more transmit directions of the base station 104a. The RU 106a may receive a beamformed signal from the base station 104a of the cell 190e based on the RU beam set 136 in one or more receive directions of the RU 106a.

[0054] The base station 104 may include and / or be referred to as a next generation evolved Node B (ng-eNB), a next generation NB (gNB), an evolved NB (eNB), an access point, a base transceiver station, a radio base station, a radio transceiver, a transceiver function, a basic service set (BSS), an extended service set (ESS), a transmit receive point (TRP), a network node, a network entity, a network device, or other related terms. The base station 104 or an entity at the base station 104 may be implemented as an IAB node, a relay node, a side link node, a converged (monolithic) base station having a RU 106 and a BBU including a DU 108 and a CU 110, or as a decomposed base station 104b including one or more of the RU 106, the DU 108, and / or the CU 110. The set of converged or decomposed base stations 104a-104b may be referred to as a next generation radio access network (NG-RAN).

[0055] The core network 120 may include an access and mobility management function (AMF) 121, a session management function (SMF) 122, a user plane function (UPF) 123, a unified data management (UDM) 124, a gateway mobile location center (GMLC) 125, and / or a location management function (LMF) 126. The core network 120 may also include one or more location servers (which may include the GMLC 125 and the LMF 126), as well as other functional entities. For example, the one or more location servers include one or more location / positioning servers, and in addition to one or more of a positioning determination entity (PDE), a serving mobile location center (SMLC), a mobile positioning center (MPC), etc., the one or more location / positioning servers may also include the GMLC 125 and the LMF 126.

[0056] AMF 121 is a control node that handles signal transmission between UE 102 and core network 120. AMF 121 supports registration management, connection management, mobility management, and other functions. SMF 122 supports session management and other functions. UPF 123 supports packet routing, packet forwarding, and other functions. UDM 124 supports the generation of authentication and key agreement (AKA) credentials, user identity handling, access authorization, and subscription management. GMLC 125 provides an interface for clients / applications (e.g., emergency services) to access UE positioning information. LMF 126 receives measurement and assistance information from NG-RAN and UE 102 via AMF 121 to calculate the positioning of UE 102. NG-RAN can use one or more positioning methods to determine the location of UE 102. Positioning UE 102 can involve signal measurement, positioning estimation, and optional speed calculation based on measurement. Signal measurement can be performed by UE 102 and / or serving base station 104 / RU 106.

[0057] The transmitted signal may also be based on one or more of a satellite positioning system (SPS) 114, such as a signal measured for positioning. In an example, the SPS 114 of the cell 190c may communicate with one or more UEs 102 (such as UE 102c) and one or more base stations 104 / RU 106 (such as RU 106c). The SPS 114 may correspond to one or more of a global navigation satellite system (GNSS), a global positioning system (GPS), a non-terrestrial network (NTN), or other satellite positioning / position systems. The SPS 114 may be associated with LTE signals, NR signals (e.g., based on round trip time (RTT) and / or multiple RTT), wireless local area network (WLAN) signals, ground beacon systems (TBS), sensor-based information, NR enhanced cell ID (NR E-CID) technology, downlink departure angle (DL-AoD), downlink arrival time difference (DL-TDOA), uplink arrival time difference (UL-TDOA), uplink arrival angle (UL-AoA), and / or other systems, signals, or sensors.

[0058] UE 102 may be configured as a cellular phone, a smart phone, a Session Initiation Protocol (SIP) phone, a laptop computer, a personal digital assistant (PDA), a satellite radio, a GPS, a multimedia device, a video device, a digital audio player (e.g., a Moving Picture Experts Group (MPEG) Audio Layer 3 (MP3) player), a camera, a game console, a tablet computer, a smart device, a wearable device, a vehicle, a utility meter, a gas pump, a home appliance, a healthcare device, a sensor / actuator, a display, or any other device with similar functionality. Some of UE 102 may be referred to as Internet of Things (IoT) devices, such as parking meters, gas pumps, home appliances, vehicles, healthcare equipment, etc. UE 102 may also be referred to as a station (STA), a mobile station, a subscriber station, a mobile unit, a subscriber unit, a wireless unit, a remote unit, a mobile device, a wireless device, a wireless communication device, a remote device, a mobile subscriber station, an access terminal, a mobile terminal, a wireless terminal, a remote terminal, a handheld device, a mobile client, a client, or other similar terms. The term UE may also apply to a roadside unit (RSU), which may communicate with other RSU UEs, non-RSU UEs, the base station 104, and / or entities at the base station 104, such as the RU 106.

[0059] Still refer to Figure 1In certain aspects, the UE 102 may include a UE-based channel state information (CSI) processing component 140, which is configured to receive a CSI-RS from a network entity and select a wideband precoder for compressing the CSI report based on the measurement of the CSI-RS. The selection of the wideband precoder is based on the entire bandwidth used to receive the CSI-RS. The UE-based CSI processing component 140 is further configured to compress the CSI report using one or more subband eigenvectors of one or more precoded subband channels through the selected wideband precoder. The UE-based CSI processing component 140 is further configured to transmit / send a first precoding matrix indicator (PMI) indicating the wideband precoder and a second PMI indicating the one or more subband eigenvectors to the network entity to decompress the CSI report.

[0060] In a further example, the UE-based CSI processing component 140 is configured to receive one or more CSI-RS from a network entity and transmit one or more CSI reports to the network entity. The UE generates at least one CSI report of the one or more CSI reports using the ML model. When the total payload size of the one or more CSI reports exceeds a threshold, the UE-based CSI processing component 140 is further configured to omit at least a portion of the at least one CSI report. The UE-based CSI processing component 140 is further configured to transmit one or more CSI reports to the network entity, wherein at least the portion of the at least one CSI report is omitted from the transmission.

[0061] In still further examples, the UE-based CSI processing component 140 is configured to receive a configuration for a CSI report processing procedure from a network entity. The configuration is for at least one of: (a) processing multiple CSI reports using an ML model, or (b) processing a first set of CSI reports using an ML model without processing a second set of CSI reports. The UE-based CSI processing component 140 is further configured to receive multiple CSI-RSs from the network entity. Measurements of the multiple CSI-RSs are based on UE capabilities for the CSI report processing procedure. The UE-based CSI processing component 140 is further configured to transmit a CSI report to the network entity based on the UE capabilities for the CSI report processing procedure.

[0062] In certain aspects, the base station 104 or a network entity of the base station 104 may include a network-based CSI processing component 150, which is configured to transmit a CSI-RS to a UE to select a wideband precoder that compresses the CSI report based on the measurement of the CSI-RS. The selection of the wideband precoder is based on the entire bandwidth used to transmit the CSI-RS. The network-based CSI processing component 150 is further configured to receive a CSI report from the UE, which is compressed using one or more subband eigenvectors of one or more precoded subband channels, a first PMI indicating the wideband precoder, and a second PMI indicating the one or more subband eigenvectors by the selected wideband precoder. The network-based CSI processing component 150 is further configured to decompress the CSI report using the first PMI and the second PMI as inputs to an ML model, which outputs a decompressed CSI report.

[0063] In a further example, the network-based CSI processing component 150 is configured to transmit one or more CSI-RS to the UE to obtain one or more CSI reports. At least one CSI report of the one or more CSI reports is based on the ML model. The network-based CSI processing component 150 is further configured to receive one or more CSI reports from the UE, wherein when a total payload size of the one or more CSI reports exceeds a threshold, at least a portion of the at least one CSI report is omitted.

[0064] In a further example, the network-based CSI processing component 150 is configured to transmit a configuration for a CSI report processing procedure to the UE. The configuration is used for at least one of: (a) processing multiple CSI reports using an ML model, or (b) processing a first set of CSI reports using an ML model without processing a second set of CSI reports. The network-based CSI processing component 150 is further configured to transmit multiple CSI-RSs to the UE. The measurement of multiple CSI-RSs is based on the UE capabilities for the CSI report processing procedure. The network-based CSI processing component 150 is further configured to transmit a CSI report to the UE based on the UE capabilities for the CSI report processing procedure. Although the following description may focus on 5G NR, the concepts described herein may be applicable to other similar fields, such as 5G-Advanced and future versions, LTE, LTE-advanced (LTE-A), and other wireless technologies.

[0065] Figure 2Figure 200 shows an example of machine learning (ML) based channel state information (CSI) compression. For a multiple-input multiple-output (MIMO) system, CSI is key information for the gNB to select a digital precoder for the UE. Typically the gNB can configure CSI reporting through RRC signaling CSI-reportConfig, where the UE uses channel measurement resources (CMRs) to measure the channel state information reference signal (CSI-RS) to estimate the downlink channel. At the same time, the gNB can configure some interference measurement resources (IMRs) in CSI-reportConfig for the UE to measure interference. With the help of CMRs and IMRs, the UE is able to identify the CSI, which may include a rank indicator (RI), a precoding matrix indicator (PMI), a channel quality indicator (CQI), and a layer indicator (LI). RI and PMI are used to determine the digital precoder (also known as the precoding matrix), CQI is used to reflect the signal-to-interference-plus-noise (SINR) state to determine the selection of the modulation and coding scheme (MCS) of the transmitter, and LI is used to identify the strongest layer, which helps with the pairing of MU-MIMO with low-rank transmission and the precoder selection of the phase tracking reference signal (PT-RS). The UE can report CSI in two parts through the physical uplink control channel (PUCCH) / physical uplink shared channel (PUSCH), where CSI part 1 may include the RI and CQI of the first transport block (TB), and CSI part 2 may include the PMI, LI, and CQI of the second TB. The payload size of CSI part 2 is determined based on CSI part 1, and both parts are transmitted using separate channel coding operations.

[0066] The gNB can configure the time domain behavior for CSI reporting in CSI-reportConfig, i.e., periodic / semi-persistent / aperiodic reporting. The gNB can activate or deactivate semi-persistent CSI reporting via MAC Control Element (CE). The gNB can trigger aperiodic CSI reporting via Downlink Control Information (DCI). The UE can report periodic CSI via PUCCH resources configured in CSI-reportConfig. The UE can report semi-persistent CSI via PUCCH resources configured in CSI-reportConfig or PUSCH resources triggered by gNB via DCI. The UE can report aperiodic CSI via PUSCH resources triggered by DCI from the gNB.

[0067] In resource element k of CSI-RS, the received signal in the frequency domain can be obtained as follows: in, Indicates the effective channel including the simulated beamforming weights, whose dimension is N Rx× N Tx , is the CSI-RS at resource element k, represents interference plus noise, N Rx is the number of receiving ports, and N Tx is the number of transmission ports.

[0068] In resource element k of the physical downlink shared channel (PDSCH), the received signal in the frequency domain may be as follows: in The precoder is indicated and should generally be the same for subcarriers within a subband (bundled physical resource block).

[0069] Starting from Rel-15, Type 2 CSI codebook is introduced for UE to measure and report CSI, in which the precoder is quantized as follows: in, The dimension is N Tx × 2L wideband precoder; is a subband precoder of dimension 2L × v; L indicates the number of beams, and v indicates the number of layers, which is RI+1.

[0070] Can be quantized based on the codebook, and The quantization can be done based on the power and angle of each element, which may result in large overhead because It is based on subbands, and one CSI report may have multiple subbands, which is determined by the bandwidth of the CSI-RS. In one example, the codebook for W1 selection can be defined as follows: in, Represents the Kronecker product; L indicates the number of beams configured by RRC signaling; N1, N2, O1, and O2 are related to the number of ports and oversampling factors configured by RRC signaling in the horizontal and vertical domains, and the candidate values ​​should be determined based on the number of CSI-RS ports. The codebook contains precoders with different values ​​of m and n. In one example, the candidate values ​​are defined as Table 5.2.2.2.1-2 in 38.214.

[0071] Machine learning is a method of performing CSI compression, where the first v columns of the feature vector of the average channel of each subband can be used as input. As used herein, the terms "machine learning" and "artificial intelligence" can be used interchangeably with each other unless explicitly stated otherwise. Figure 2An example of machine learning based CSI compression is shown after the UE receives the CSI-RS. The UE can estimate the channel based on the CSI-RS and then calculate the eigenvector of the channel in each subband. The eigenvector can be the input of the neural network for CSI compression. The UE can then report the compressed CSI to its gNB. Upon reception, the gNB can decode the compressed CSI and then decompress the CSI using a neural network to recover the eigenvector.

[0072] For each subband, a eigenvector V can be derived based on the singular vector decomposition (SVD) of the average channel in the subband as shown below. Where N indicates the number of CSI-RS resource elements of subband S; is the estimated channel based on the CSI-RS at resource element k.

[0073] For machine learning (ML) based CSI compression, the corresponding terms are defined as follows.

[0074] Data collection refers to the process of network nodes, management entities or UEs collecting data for AI / ML model training, data analysis and inference.

[0075] An AI / ML model refers to a data-driven algorithm that applies AI / ML techniques to generate a set of outputs based on a set of inputs.

[0076] AI / ML model training refers to the process of training an AI / ML model in a data-driven manner (e.g., by learning input / output relationships) and obtaining the trained AI / ML model for reasoning.

[0077] AI / ML model inference refers to the process of using a trained AI / ML model to generate a set of outputs based on a set of inputs.

[0078] AI / ML model validation refers to the training subprocess of evaluating the quality of an AI / ML model using a dataset different from the one used for model training, which helps select model parameters that generalize beyond the dataset used for model training.

[0079] AI / ML model testing refers to the training subprocess of evaluating the performance of the final AI / ML model using a dataset different from the one used for model training and validation. Unlike AI / ML model validation, testing does not assume subsequent adjustments to the model.

[0080] UE-side AI / ML models refer to AI / ML models where reasoning is performed entirely at the UE.

[0081] Network-side AI / ML models refer to AI / ML models in which reasoning is performed entirely at the network.

[0082] A one-sided AI / ML model refers to an AI / ML model on the UE side or an AI / ML model on the network side.

[0083] A two-sided AI / ML model refers to a paired AI / ML model on which joint reasoning is performed, where the joint reasoning includes AI / ML reasoning whose reasoning is performed jointly across the UE and the network, i.e., the first part of the reasoning is first performed by the UE and then the remaining part is performed by the gNB, or vice versa.

[0084] AI / ML model delivery refers to the delivery of AI / ML models, i.e., parameters of a model structure known at the receiving end, or a new model with parameters, over an air interface. The delivery can include a complete model or a partial model.

[0085] Model download refers to the transfer of models from the network to the UE.

[0086] Model upload refers to the transfer of models from UE to the network.

[0087] Federated learning / federated training refers to a machine learning technique that trains AI / ML models across multiple distributed edge nodes (e.g., UE, gNB), with each node performing local model training using local data samples. This technique requires multiple interactions of the model but does not require the exchange of local data samples.

[0088] Offline field data refers to data collected from the field and used for offline training of AI / ML models.

[0089] Online field data refers to data collected from the field and used for online training of AI / ML models.

[0090] Model monitoring refers to the process of monitoring the inference performance of AI / ML models.

[0091] Supervised learning refers to the process of training a model based on inputs and their corresponding labels.

[0092] Unsupervised learning refers to the process of training a model without the need for labeled data.

[0093] Semi-supervised learning refers to the process of training a model using a mixture of labeled and unlabeled data.

[0094] Reinforcement learning (RL) is the process of training an AI / ML model based on the inputs (aka states) and feedback signals (aka rewards) generated from the model’s outputs (aka actions) in the environment that the model interacts with.

[0095] Model activation refers to enabling an AI / ML model for a specific function.

[0096] Model deactivation refers to disabling an AI / ML model for a specific function.

[0097] Model switching refers to deactivating the currently active AI / ML model and activating a different AI / ML model for a specific function.

[0098] Figure 2 One potential problem with the current process in is that the UE complexity of the eigenvector computation places considerable demands on the processor within the UE. For each subband, the computation of the first eigenvector V1 requires a singular vector decomposition (SVD) of the average channel as shown below. Where N indicates the number of CSI-RS resource elements of subband S; is the estimated channel based on the CSI-RS at resource element k. Usually the number of transmission ports in the CSI-RS may be large, such as N Tx = 32. Therefore, the complexity of SVD may be a problem.

[0099] Therefore, a method for reducing UE complexity of ML-based CSI compression is proposed, including: reducing the complexity of feature vector calculation.

[0100] The UE may be configured with multiple CSI-reportConfigs, and some may be based on ML configuration, i.e., AI-based CSI reporting, and some may be based on legacy codebook configuration, i.e., non-AI-based CSI reporting. The UE may be triggered to report CSI from multiple CSI-reportConfigs via PUSCH or PUCCH. Then the total payload size of the CSI may exceed the maximum payload size of CSI part 2 in PUSCH / PUCCH. Some parts need to be omitted. For multiple AI-based CSI reports or mixed AI-based and non-AI-based CSI reports, how to define the CSI omission rules may be a second issue.

[0101] Therefore, a method for AI-based CSI reporting is proposed, including: a CSI omission rule for multiple AI-based and non-AI-based CSI reports; and a CSI omission rule for a single AI-based CSI report.

[0102] Figure 3A A first timing diagram 300 is shown of a first CSI processing unit duration 311 associated with periodic / semi-persistent CSI reporting. Figure 3BA second timing diagram 350 is shown of a second CSI processing unit duration 313 associated with an aperiodic CSI report. The UE 102 may be configured with multiple CSI-reportConfig information elements (IEs) for multiple CSI measurements and reports. Thus, multiple CSI processing units may be used to process the received CSI-RS in parallel to create multiple CSI measurements and reports. The UE 102 may transmit UE capability information to the base station 104 indicating the number of CSI processing units supported by the UE 102. If the CSI report requested by the base station 104 requires more parallel CSI processing at the UE 102 than the number of CSI processing units available at the UE (i.e., indicated in the UE capability report), the UE 102 may report non-current information for a low priority CSI report. In some examples, the UE 102 may determine the priority of a particular CSI report based on a predefined protocol. For example, the CSI processing unit occupancy rules for periodic / semi-persistent CSI reports 307 and aperiodic CSI reports 309 may be based on a predefined protocol.

[0103] The first CSI processing unit duration 311 of the periodic / semi-persistent CSI report 307 corresponds to the CSI processing unit occupied starting from the first symbol of the earliest resource of the CMR 303 or IMR 305 used for measurement by the UE 102. For example, the UE 102 may perform one or more measurements of CSI-RS, CSI interference measurement (CSI-IM), synchronization signal block (SSB), etc. The first CSI processing unit duration 311 of the periodic / semi-persistent CSI report 307 continues through the last resource of the CMR 303 and IMR 305 used for measurement by the UE 102, and ends at the last symbol of the PUSCH / PUCCH used by the UE 102 to transmit the periodic / semi-persistent CSI report 307 to the base station 104.

[0104] The second CSI processing unit duration 313 of the aperiodic CSI report 309 corresponds to the CSI processing unit occupied starting at the first symbol after receiving the physical downlink control channel (PDCCH) 301 that triggers the aperiodic CSI report 309. The second CSI processing unit duration 313 of the aperiodic CSI report 309 continues through the last resource of the CMR 303 and IMR 305 measured by the UE 102, and ends at the last symbol of the PUSCH used by the UE 102 to transmit the aperiodic CSI report 309 to the base station 104. If the PDCCH 301 that triggers the aperiodic CSI report 309 corresponds to two PDCCH candidates in two corresponding search space sets, the PDCCH candidate that ends later in time is used to determine the second CSI processing unit duration 313 of the aperiodic CSI report 309.

[0105] If the PDCCH 301 triggers an initial semi-persistent CSI report on the PUSCH, the CSI processing unit duration of the initial semi-persistent CSI report may not be the same as the first CSI processing unit duration 311 of the periodic / semi-persistent CSI report 307. Instead, the CSI processing unit duration of the initial semi-persistent CSI report may correspond to the second CSI processing unit duration 313 of the aperiodic CSI report 309. That is, after the PDCCH 301, the CSI processing unit duration of the initial semi-persistent CSI report transmitted on the PUSCH starts at the first symbol after the PDCCH 301 and ends at the last symbol of the PUSCH carrying the initial semi-persistent CSI report. If the PDCCH 301 that triggers the initial semi-persistent CSI report corresponds to two PDCCH candidates in two corresponding search space sets, the PDCCH candidate that ends later in time is used to determine the CSI processing unit duration of the initial semi-persistent CSI report.

[0106] The CSI reporting by the UE 102 may be based on a minimum processing delay time. For example, the scheduling of periodic / semi-persistent CSI reporting 307 or aperiodic CSI reporting 309 may include minimum processing delays Z and Z'. The values ​​of Z and Z' for different types of CSI reports may be based on one or more predefined protocols. If the scheduling offset does not indicate a minimum processing delay Z and Z', the UE 102 may report the minimum processing delay associated with the non-current CSI, or ignore the triggering DCI if there is no other signal (e.g., data or hybrid automatic repeat request (HARQ)-acknowledgement (ACK) (HARQ-ACK)) to be transmitted on the PUSCH triggered by the DCI (e.g., PDCCH 301). If the CSI request field of the DCI is used to trigger one or more CSI reports from the UE 102 on the PUSCH, the first uplink symbol associated with the one or more CSI reports is used as no earlier than symbol Z. ref UE 102 may transmit one or more CSI reports to base station 104 if the timing of the start of the nth CSI report is advanced by 0.1% or more. For example, if the first uplink symbol associated with the nth CSI report is not earlier than symbol Z′ ref( n), the UE 102 may transmit the nth triggered CSI report to the base station 104.

[0107] Compared with non-AI-based CSI, AI-based CSI reporting can be implemented by the UE based on different hardware (e.g., neural processing unit (NPU)). Then, AI-based and non-AI-based CSI may not share the CPU. Therefore, how to manage the parallel processing of multiple AI-based CSI reports and mixed AI-based and non-AI-based CSI reports may be a problem. In addition, the complexity of AI-based CSI reporting will also be determined by the ML model. Then, how to define the minimum processing delay of AI-based CSI reporting may be another problem.

[0108] Therefore, a method for a parallel processing method for AI-based CSI measurement and reporting is proposed, including: a CSI processing unit (CPU) management framework; and minimum processing delay requirements. Low-complexity ML-based CSI compression

[0109] Figure 4 A diagram 400 shows the general process of low-complexity machine learning based CSI compression. Figure 2 Compared to the traditional method in , the input of the neural network is calculated as follows:

[0110] The first v columns of the second feature vector V2 may be inputs to a neural network, where v indicates the number of layers. W1 is a wideband precoder that may be quantized based on a predefined codebook or a codebook configured by a network entity, such as a Type 1 codebook defined in Section 5.2.2.2.1 of 3GPP TS 38.214 v17.2.0, or a Type 2 codebook defined in Section 5.2.2.2.3 of 3GPP TS 38.214 v17.2.0. In one example, the codebook selected for W1 may be defined as follows: in, Represents the Kronecker product; L indicates the number of beams configured by RRC signaling; N1, N2, O1, and O2 are related to the number of ports and oversampling factors configured by RRC signaling in the horizontal and vertical domains, and the candidate values ​​should be determined based on the number of CSI-RS ports. The codebook contains precoders with different values ​​of m and n. In one example, the candidate values ​​are defined as Table 5.2.2.2.1-2 in 38.214.

[0111] The dimension can be N Rx × 2L, where L is configured by the RRC parameters. In one example, L can be 2. Then, the dimension of the input matrix of the SVD can be smaller than Figure 2 The process in is much smaller.

[0112] Figure 5 is a signaling diagram 500 illustrating a procedure for low complexity ML based CSI reporting. The gNB may provide the configuration of ML based CSI reporting through RRC signaling, such as CSI-reportConfig. The gNB may trigger CSI reporting through MAC CE or DCI. For periodic CSI reporting, this message may be skipped. In one example, the gNB may trigger semi-persistent CSI reporting through MAC CE, while the gNB may trigger aperiodic CSI reporting through DCI. The UE may perform CSI measurements based on reception of CSI-RS. The UE may compress the CSI based on low complexity ML based compression operations, such as Figure 6 As shown, where the second eigenvector is used as input to the ML model compared to the conventional method (the first eigenvector is used as input). After CSI compression, the UE can report the compressed CSI to the gNB. The gNB can receive the CSI and decompress the CSI based on ML. Figure 6 Method diagram 600 illustrates UE behavior for CSI compression and reporting. Figure 7 Method diagram 700 showing gNB behavior for CSI decompression.

[0113] In an embodiment, the input to the neural network of the UE-side CSI encoder should be based on the selected wideband precoder and the channel estimated from the CSI-RS. In one example, the input can be based on the matrix The first v columns of the second eigenvector V2 calculated by the SVD of W1. The number of beams of W1 and the codebook structure to be searched can be configured by higher layer signaling such as RRC signaling. In one example, the codebook can be based on the Type 1 or Type 2 CSI codebook defined in Section 5.2.2.2.1 or 5.2.2.2.3 of 38.214. In one example, the codebook for W1 selection can be defined as follows: in, represents the Kronecker product; L indicates the number of beams that can be configured by RRC signaling such as numberOfBeams; N1, N2, O1, and O2 are related to the number of ports and oversampling factors configured by RRC signaling such as n1-n2 in the horizontal and vertical domains, and the candidate values ​​should be determined based on the number of CSI-RS ports. The codebook contains precoders with different values ​​of m and n. In one example, the candidate values ​​are defined as Table 5.2.2.2.1-2 in 38.214.

[0114] The ML model of the CSI encoder may be configured through higher layer signaling (e.g., RRC signaling) of the gNB, or may be predefined or preconfigured. In one implementation, different ML models may be used for different codebook structures. In one example, one ML model should be used for a codebook configured with a certain number of beams. In another example, one ML model should be used for a codebook configured with a certain number of beams N1, N2, O1, and O2. In another example, one ML model should be used for a codebook configured with a certain number of subbands, a certain number of beams N1, N2, O1, and O2.

[0115] In some implementations, prior to RRC configuration of the codebook for ML based CSI reporting, the UE may report its capability of supporting the codebook structure for ML based CSI reporting.

[0116] In an embodiment, for AI-based CSI reporting, the output of the ML for CSI compression, which may be reported as a first precoder matrix index (PMI) of W1 and a second PMI for each subband, which may be represented as a compressed feature vector or a second PMI for each subband, may be reported by the UE to the gNB.

[0117] In one example, the first PMI may be reported as two indexes: one for indicating the horizontal beam index m, and the other for indicating the vertical beam index n. The second PMI is the output of the ML model for CSI compression.

[0118] FIG. 8A to FIG. 8D Tables 800-830 illustrate example ML CSI reports. In one implementation, for short PUCCH based CSI reporting, PMI and other CSI may be reported in a single part. Fig. 8A Table 800 shows an example of the MLCSI reporting format in the short PUCCH.

[0119] In one implementation, for CSI based on PUSCH or long PUCCH, both PMIs may be reported in CSI part 2, where the bit width of CSI part 2 is determined by the information reported in CSI part 1, such as CRI / RI and CQI of the first transport block (TB). Figure 8B Table 810 shows another example of the ML CSI reporting format as part 2 CSI in long PUCCH and PUSCH.

[0120] In another implementation, for CSI based on PUSCH or long PUCCH, both PMIs may be reported in CSI part 2, with the first PMI reported in CSI part 1 and the second PMI reported in CSI part 2. Figure 8C Table 820 and Fig.8DTable 830 shows another example of ML CSI reporting format as part 1 and part 2 CSI in long PUCCH and PUSCH.

[0121] In an embodiment, low complexity based CSI compression may be enabled through RRC signaling of the gNB, such as RRC parameters in CSI-reportConfig, which RRC signaling should be based on UE capability report. In one implementation, the codebook configuration with codebookType is 'ML' or 'Type3'. The RRC signaling of codebook ML may include at least one of the following: the number of beams (L), the number of ports in the horizontal and vertical domains (N1, N2), the number of oversampling factors in the horizontal and vertical domains (O1, O2), the number of subbands used for CSI compression, the number of subbands per CQI calculation, which indicates the number of subbands for CSI compression used for CQI calculation, the RI limit for indicating the candidate rank for precoder selection, or the ML model indicating the ML model used for CSI compression.

[0122] In addition, for low-complexity based ML CSI compression and traditional ML, a subtype may be indicated in the codebook configuration of ML to indicate whether it is low-complexity based ML, such as typeIII-hybrid, or traditional ML, such as typeIII.

[0123] Unless otherwise specified herein, RRC signaling may indicate an RRC reconfiguration message from the gNB to the UE, or a system information block (SIB), where the SIB may be an existing SIB (e.g., SIB1) or a new SIB transmitted by the gNB. In addition, the gNB may obtain the UE capabilities via UE capability report signaling or from the core network (e.g., access and mobility management function (AMF)).

[0124] Furthermore, in the present invention, the described solution is based on 5G NR technology. It should be noted that the solution can be applied to other wireless technologies such as 6G. "gNB" can be summarized as a base station or a radio access network (RAN) node. CSI Omission for AI-Based CSI Reporting

[0125] Fig. 9AA diagram 900 showing a general process of CSI reporting with respect to CSI omission rules of some embodiments. When the CSI overhead of multiple CSI reports exceeds the maximum payload size of the PUSCH / PUCCH of the CSI reports, a certain CSI in part 2 needs to be omitted based on the priority rule so that the overall payload size of the CSI reports fits the maximum payload size that can be supported by the PUSCH / PUCCH for the CSI reports. In the present invention, CSI with higher priority values ​​should be omitted, for example, CSI with priority value = 1 needs to be omitted compared to CSI with priority value = 0.

[0126] First, the gNB can send RRC signaling to the UE to configure at least one ML-based CSI-reportConfig. If the codebookType in the CSI-reportConfig is set to a specific value, such as 'ai-Ml' or 'type3', or the reportQuantity in the CSI-reportConfig is set to a specific value, such as 'ri-compressedPmi-cqi', the CSI report can be regarded as AI-based CSI. In addition, the gNB can also configure some non-AI-based CSI reports, which can be based on a specific codebook, such as Type1 or Type2 codebook.

[0127] Second, the gNB may trigger at least one AI-based CSI report via MAC CE or DCI. In one example, the gNB may send a MAC CE to activate semi-persistent CSI reporting. In another example, the gNB may send a DCI to trigger aperiodic CSI reporting. In another example, the UE should periodically report periodic CSI reports at uplink resources configured by the gNB via RRC signaling.

[0128] Third, the UE may measure the CSI-RS corresponding to the triggered or configured CSI report to derive the CSI. If the gNB configures or triggers multiple CSI reports and the total payload size exceeds the maximum payload size of PUCCH / PUSCH for CSI reporting, the UE may determine to apply CSI omission rules to omit some CSI reports and / or some parts of some CSI reports. The UE needs to first determine the priority of AI-based and non-AI-based CSI reports, and then if a certain CSI omission of AI-based CSI reporting is applied, the UE may decide to omit a part of the AI-based CSI report. For CSI omission of AI-based reporting, the following three embodiments of AI-based CSI report content are considered: Scenario 1 corresponds to that the AI-based CSI report may include a PMI for indicating compressed characteristic vectors of channels in all subbands; Scenario 2 corresponds to that the AI-based CSI report may include a first PMI for indicating the selected wideband precoder and a second PMI for indicating compressed characteristic vectors of precoded channels with the selected wideband precoder in all subbands; and Scenario 3 corresponds to that the AI-based CSI report may include a first PMI for indicating the selected wideband precoder, a second PMI for indicating compressed characteristic vectors of precoded channels with the selected wideband precoder in even subbands, and a third PMI for indicating compressed characteristic vectors of precoded channels with the selected wideband precoder in odd subbands.

[0129] Fourth, the UE sends a CSI report with the remaining contents of the AI-based and non-AI-based CSI reports that have been triggered or configured to be reported by the gNB. The CSI report can be transmitted by PUCCH or PUSCH. For PUCCH, the CSI report can be reported in the long PUCCH format. The gNB can determine the remaining CSI of the CSI report based on the CSI omission rules and then decode the CSI report accordingly.

[0130] Fig. 9B A diagram 950 illustrates an alternative process for CSI reporting with respect to CSI omission rules according to other embodiments. Fig. 9A The difference is that for ML-based CSI reporting, if CSI omission is required, the UE may replace the AI-based CSI with non-AI-based CSI. The UE may then perform CSI omission on some parts of the new non-AI-based CSI report. Fig. 9A compared to, Fig. 9B The process in can be more flexible so that the UE can report more effective CSI. Fig. 9B The process in may require higher UE complexity because for AI-based CSI reporting, in addition to the AI-based CSI, the UE may also need to measure non-AI-based CSI.

[0131] Fig.10 is based on Fig. 9A The process in method diagram 1000 shows UE behavior for reporting AI / non-AI based CSI using CSI omission rules. Fig.11 is based on Fig. 9A The process in Figure 1100 shows the method behavior of gNB for reporting AI / non-AI based CSI using CSI omission rules. Fig.12 is based on Fig. 9B The process in method diagram 1200 shows UE behavior for reporting AI / non-AI based CSI using CSI omission rules. Fig.13 is based on Fig. 9B The process in Figure 1300 shows the method behavior of gNB for reporting AI / non-AI based CSI using CSI omission rules.

[0132] Unless otherwise specified, RRC signaling may indicate an RRC reconfiguration message from the gNB to the UE, or a system information block (SIB), where the SIB may be an existing SIB (e.g., SIB1) or a new SIB transmitted by the gNB. In addition, the gNB may obtain UE capabilities via UE capability report signaling or from the core network (e.g., access and mobility management function (AMF)).

[0133] Furthermore, in the present invention, the described solution is based on 5G NR technology. It should be noted that the solution can be applied to other wireless technologies such as 6G. "gNB" can be summarized as a base station or a radio access network (RAN) node.

[0134] In an embodiment, whether AI-based CSI reporting can be multiplexed and reported through PUSCH / PUCCH can be reported by the UE through UE capabilities and predefined or configured by the gNB through higher layer signaling (e.g., RRC signaling). AI-based CSI reporting, i.e., CSI measured using an ML model-based CSI encoder and non-AI-based CSI reporting, such as CSI measured based on Type 1 / Type 2 codebooks.

[0135] In an embodiment, the priority of AI-based CSI reporting or non-AI-based reporting may be different. The gNB and UE may determine the priority of CSI reporting by the time domain behavior of CSI reporting, i.e., periodic / semi-persistent / aperiodic reporting, CSI-reportConfigId, serving cell ID, number of reports, and whether the measurement is based on AI or non-AI methods.

[0136] In the example, the priority can be based on , where if the priority value is lower, the actual priority is higher (e.g., priority 0 corresponds to a higher priority than priority 1). A value of k=0 may indicate a CSI report including L1-RSRP or L1-SINR information, a value of k=1 may indicate a non-AI based CSI report not including L1-RSRP or L1-SINR information, and a value of k=2 may indicate an AI based CSI report not including L1-RSRP or L1-SINR information. In addition, Indicates aperiodic CSI reporting transmitted on PUSCH, Indicates a semi-persistent CSI report transmitted on PUSCH, indicates a semi-persistent CSI report transmitted on the PUCCH, and Indicates the periodic CSI report transmitted on the PUCCH. c corresponds to the serving cell index, and corresponds to the value of the higher layer parameter maxNrofServingCells. s corresponds to reportConfigID, and Corresponds to the value of the higher layer parameter maxNrofCSI-ReportConfigurations.

[0137] Alternatively, for CSI reports carrying L1-RSRP or L1-SINR, k = 0, and for AI-based CSI reports not carrying L1-RSRP or L1-SINR, k = 1, and for non-AI-based CSI reports not carrying L1-RSRP or L1-SINR, k = 2. In another implementation, the gNB and UE may always assume that AI-based CSI reports have a higher or lower priority than non-AI-based CSI reports.

[0138] FIG. 14A to FIG. 14C Tables 1400-1420 of AI-based CSI reporting are shown. In an embodiment, within the AI-based CSI reporting, it may be assumed that the wideband PMI and subband PMI have different priorities. This may only apply to scenarios 2 and 3. The priority value of the wideband PMI is less than the subband PMI, that is, when it is necessary to omit the subband PMI, the subband PMI can be omitted. For scenarios 2 and 3, the priorities of the first PMI and the second / third PMI should be different. In one example, the gNB and the UE may determine the priority of the first PMI and the second / third PMI according to the following example: Fig.14A Table 1400 determines N Rep The priority of the AI-based CSI reports is as follows, where as in the previous embodiment, CSI report n corresponds to the CSI report with the nth smallest Pri i,CSI CSI report of (y,k,c,s) values.

[0139] In another embodiment, the gNB and UE assume that a single priority value is assigned for AI-based CSI reporting. This can be applicable to scenarios 1, 2, and 3. Then, only the priority value should be assigned based on Fig. 14B Table 1410 is based on Pri i,CSI (y, k, c, s) to determine the CSI omission priority.

[0140] In another embodiment, the gNB and UE assume that AI-based CSI can be divided into 3 priority groups. In one implementation, for scenario 3, the groups can be: Group 0 for wideband PMI (first PMI), Group 1 for subband PMI for even subbands (second PMI), and Group 2 for subband PMI for odd subbands (third PMI). For AI-based CSI reporting, ML-based CSI compression can be performed twice, the first time for even subbands and the second time for odd subbands.

[0141] In another implementation, for scenario 2, the groups may be: Group 0 for wideband PMI (first PMI), Group 1 for high priority bits for subband PMI (part of second PMI), and Group 2 for low priority bits for subband PMI (another part of second PMI). With the high priority bits, the gNB is able to recover the first eigenvector with some loss. Which bits should be high priority bits should depend on the ML model and may be predefined or configured by the gNB through higher layer signaling, such as RRC signaling in CSI-reportConfig. The bits in Group 1 and Group 2 may be generated based on a single AL / ML model or separate ML models.

[0142] In another implementation, for scenario 1, the groups may be: Group 0: high priority bits of PMI in CSI part 2, Group 1: medium priority bits of PMI in CSI part 2, and Group 2: low priority bits of PMI in CSI part 2. Based on the bits with different priorities, the gNB may be able to recover the first eigenvector with some loss. Which bits should be high / medium / low priority bits should depend on the neural network architecture and may be predefined or configured by the gNB through higher layer signaling, such as RRC signaling in CSI-reportConfig. The bits in Group 1 / 2 / 3 may be generated based on a single AL / ML model or separate ML models. Therefore, the gNB may be able to recover the first eigenvector with some loss based on the bits with different priorities. Fig. 14C Table 1420 defines the CSI omission priority.

[0143] In an embodiment, if a portion of the AI-based CSI needs to be omitted, the UE may fall back to report non-AI-based CSI based on a fallback codebook, such as a Type2 CSI codebook or an eType2 CSI codebook. The fallback codebook may be configured through RRC signaling, such as RRC parameters in CSI-reportConfig, or predefined or reported through UE capability signaling. The UE may perform CSI omission based on the non-AI-based CSI rules defined in 5.2.3 of 38.214.

[0144] In some implementations, the gNB may apply the same CSI omission rule to determine whether the UE reports AI-based CSI or non-AI-based CSI for AI-based CSI reporting. In some other implementations, for CSI reporting, the UE may report an additional indicator to report whether the CSI is measured based on AI or non-AI. In one implementation, such an indicator may be explicitly transmitted in the CSI report, for example, an indicator for reporting the codebook type used for CSI reporting. In another implementation, the gNB may configure or trigger 2 PUCCH / PUSCH resources for CSI reporting, one for AI-based reporting and the other for non-AI-based reporting. By selecting the corresponding PUCCH / PUSCH resource, the UE may implicitly report whether the CSI is based on AI or non-AI. In another implementation, such an indicator may be implicitly reported based on the scrambling sequence of the DMRS or PUCCH / PUSCH, one of which is used to indicate AI-based CSI reporting and the other is used to indicate non-AI-based CSI reporting. By receiving such an indicator, the gNB can determine whether to decode the CSI as AI-based CSI or non-AI-based CSI.

[0145] In an embodiment, the UE may use multiple ML models with different compression rates for CSI compression, where the ML models may be configured or predefined via RRC signaling. Similarly, the gNB may use multiple ML models with different compression rates for CSI decompression. Regarding CSI omission, the UE may select the ML model with the highest compression rate that can fit the overhead constraints of the PUCCH / PUSCH for CSI reporting. In some implementations, the gNB may apply the same CSI omission rule to determine the ML model for CSI decompression, where the ML model with the highest compression rate that can fit the overhead constraints of the PUCCH / PUSCH for CSI reporting should be selected.

[0146] In some other implementations, for CSI reporting, the UE may report an additional indicator to report the ML model index. The gNB and the UE may maintain a list of ML models with different compression rates. In one example, such an ML model may be configured via RRC signaling from the gNB to the UE, or via UE capability reporting, or may be predefined. In one implementation, such an indicator may be explicitly transmitted in the CSI report, for example, an indicator for explicitly reporting the ML model index. In another implementation, the gNB may configure or trigger X (X is an integer greater than 1) PUCCH / PUSCH resources for CSI reporting, where each resource is used for one ML model. By selecting the corresponding PUCCH / PUSCH resource, the UE may implicitly report the ML model index. In another implementation, such an indicator may be implicitly reported based on a scrambling sequence of a DMRS or PUCCH / PUSCH, where each scrambling sequence corresponds to an ML model. By receiving such an indicator, the gNB may determine the ML model to be used for CSI decompression.

[0147] In an embodiment, for N Rep,1 AI-based CSI reports and N Rep,2 For CSI omission of non-AI based CSI reports, the gNB and UE may assume AI based CSI reports with higher or lower priority values ​​compared to non-AI based CSI reports. In different implementations, the UE shall first determine to omit AI based or non-AI based CSI reports, and if further omissions are still required, the UE may apply CSI omission to the remaining non-AI or AI based CSI. For AI based CSI omission, the above embodiments may be applied. For non-AI based CSI omission, the CSI omission rules defined in Section 5.2.3 of 38.214 may be applied.

[0148] In another embodiment, for N Rep,1 AI-based CSI reports and N Rep,2 For CSI omission of non-AI based CSI reports, the gNB and UE shall first determine the priority of the CSI report based on the priority of CSI-reportConfig, and then the priority within CSI-reportConfig may be determined based on the priority group. Parallel processing of ML-based CSI

[0149] Fig.151500 is a signaling diagram illustrating the process of a CPU framework for AI-based CSI processing. In some embodiments, AI-based CSI can be implemented based on different hardware than non-AI-based CSI, such as a neural processing unit (NPU). The NPU can be dedicated or shared with other applications, such as non-wireless communication-based applications. In some other embodiments, AI-based CSI can be implemented based on the same hardware as the hardware for non-AI-based CSI. Therefore, it is necessary to introduce a second CPU framework for AI-based CSI processing. In this IDF, a Type 2 CPU is defined as a CPU for AI-based CSI processing that cannot be shared with non-AI-based CSI. Traditional CPUs for non-AI-based CSI processing are represented as Type 1 CPUs.

[0150] The UE may report at least one UE capability, i.e., whether it supports AI-based CSI compression and the maximum number of Type2 CPUs it supports. The gNB may provide configuration of the CSI framework for multiple CSI reports, wherein the gNB may provide the configuration for CSI reporting through RRC signaling, such as CSI-reportConfig. For semi-persistent and aperiodic CSI reporting, the gNB may send a MAC CE or DCI to trigger the corresponding report, wherein the gNB uses a MAC CE to activate semi-persistent CSI reporting and uses a DCI to trigger aperiodic CSI reporting. After receiving control signaling for triggering at least one AI-based CSI report, the UE may determine the CPU status and determine which CSI report(s) should be a valid CSI report, such as a CSI report with the latest measurement results. Then, if the number of Type1 or Type2 CPUs exceeds the maximum number of Type1 or Type2 CPUs, the UE may send updated CSI for high priority CSI reporting and outdated CSI for low priority CSI reporting. Fig.16 is a method diagram 1600 of UE behavior for a CPU framework for AI-based CSI processing. Fig.17 FIG1700 is a method diagram of gNB behavior for a CPU framework for AI-based CSI processing.

[0151] In an embodiment, during UE capability reporting, the UE may report at least one UE capability, i.e., whether it supports AI-based CSI processing, whether AI-based CSI processing can share the CPU for non-AI-based processing, and the maximum number of Type2 CPUs per component carrier (CC) or per band or per band combination or per UE. In addition to the UE capability on the maximum number of Type1 CPUs, the UE may also report the UE capability for Type2 CPUs, for example, simultaneousCSI-ReportsPerCC for the number of Type1 CPUs per CC, and simultaneousCSI-ReportsAllCC for the number of Type1 CPUs on all CCs. In one example, if the UE reports that it supports AI-based CSI processing using 0 Type2 CPUs, it can be assumed that AI-based CSI processing can share the CPU for non-AI-based processing. In addition, if the NPU for AI-based CSI processing can be shared with other applications, the UE may report the on / off duration of the Type2 CPU. When the Type2 CPU is off, it cannot be counted in the AI-based CSI processing. The on / off duration may be reported based on the maximum periodicity and the on duration of the Type2 CPU.

[0152] In an embodiment, the gNB may configure whether the AI-based CSI report should occupy the Type2 CPU. For the AI-based CSI report that does not occupy the Type2 CPU, it should occupy the Type1 CPU. In some implementations, the gNB may provide such CPU type indication for the CSI report through RRC signaling, such as RRC parameters in CSI-reportConfig. In some further implementations, the gNB may provide such CPU type indication for the CSI report through MAC CE. In one example, the gNB may provide the CPU type indication for the semi-persistent CSI report through MAC CE. In another example, the gNB may provide the CPU type indication based on a separate MAC CE, wherein the MAC CE may provide at least one element of the serving cell index, bandwidth part index, CSI-reportConfig ID, and CPU type indication for the indicated CSI-reportConfig. In some further implementations, the gNB may provide such CPU type indication for the CSI report through DCI. In one example, for aperiodic CSI report, the gNB may indicate the CPU type through the DCI used to trigger the CSI report. A new field may be added in the DCI to indicate the CPU type. Alternatively, the gNB may configure a CPU type corresponding to the CSI triggering state, and by indicating a different value of the CSI request field in the DCI, the gNB may indicate the corresponding CPU type.

[0153] In an embodiment, if the UE uses Type2 CPU to report AI-based CSI processing, in one implementation, the Type2 CPU is occupied based on rules similar to the Type1 CPU. For example, a periodic or semi-persistent CSI report (excluding the initial semi-persistent CSI report on the PUSCH after the PDCCH triggers the report) occupies the Type2 CPU from the first symbol of the earliest resource in each CSI-RS / CSI-IM / SSB resource used for channel or interference measurement, no later than the corresponding latest CSI-RS / CSI-IM / SSB timing of the corresponding CSI reference resource, until the last symbol of the configured PUSCH / PUCCH carrying the report. Aperiodic CSI reports occupy the Type2 CPU from the first symbol after the PDCCH triggers the CSI report until the last symbol of the scheduled PUSCH carrying the report. When the PDCCH reception includes two PDCCH candidates from two corresponding search space sets, as described in clause 10.1 of 3GPP TS 38.213, in order to determine the CPU occupancy duration, the PDCCH candidate that ends later in time is used. The initial semi-persistent CSI report on PUSCH after PDCCH triggering occupies Type2 CPUs from the first symbol after the PDCCH until the last symbol of the scheduled PUSCH carrying the report. When a PDCCH reception includes two PDCCH candidates from two corresponding search space sets, as described in clause 10.1 of 3GPP TS 38.213, for determining the CPU occupancy duration, the PDCCH candidate that ends later in time is used. In this implementation, the gNB and UE may consider Type2 CPUs as a subset of Type1 CPUs. Therefore, if Type2 CPUs are occupied, Type1 CPUs are also occupied. Then, the maximum number of Type2 CPUs reported in the UE capabilities should be less than or equal to the maximum number of Type1 CPUs reported in the UE capabilities.

[0154] Fig.18 A timing diagram 1800 showing Type2 CPU occupancy rules. Since the NPU will only be used after the UE acquires the channel, it can be assumed that the Type2 CPU is occupied after X symbols after the first / last symbol of the earliest / latest channel measurement resource (CMR) or interference measurement resource (IMR) and until Y symbols before the first / last symbol of the PUSCH / PUCCH for CSI reporting. The values ​​of X and Y can be predefined or reported by the UE capability. Fig.18 An example of Type 2 CPU occupancy rule is shown.

[0155] FIG. 19A to FIG. 19BExample timing diagrams 1900-1950 showing CPU occupancy rules for AI-based periodic / semi-persistent CSI reporting. FIG. 20A to FIG. 20B Example timing diagrams 2000-2050 showing CPU occupancy rules for AI-based aperiodic CSI reporting. If the UE reports that AI-based CSI processing requires a Type 2 CPU, the AI-based CSI processing may also require a Type 1 CPU because some other CSI processing processes (e.g., CSI-RS decoding, RI and CQI calculations) may not require neural network processing.

[0156] In one implementation, both Type 1 and Type 2 CPUs are occupied based on the same rules as Type 1 CPU occupation. In another implementation, the gNB and UE may determine that Type 1 CPU is occupied based on the legacy rules, while Type 2 CPU is occupied after X symbols after the first / last symbol of the earliest / latest channel measurement resource (CMR) or interference measurement resource (IMR) and until Y symbols before the first / last symbol of PUSCH / PUCCH for CSI reporting. The values ​​of X and Y may be predefined or reported by UE capabilities.

[0157] The gNB and UE may determine that the Type1 CPU is occupied based on the conventional rules (excluding the duration for which the Type2 CPU is occupied). The Type2 CPU is occupied after X symbols after the first / last symbol of the earliest / latest Channel Measurement Resource (CMR) or Interference Measurement Resource (IMR) and until Y symbols before the first / last symbol of the PUSCH / PUCCH for CSI reporting. The values ​​of X and Y may be predefined or reported by the UE capability.

[0158] If AI-based CSI processing requires only Type 2 CPU, the UE may process N with higher priority according to Section 5.2.5 of 38.214 after checking the Type 2 CPU occupancy status. CPU,2 AI-based CSI reports, of which N CPU,2 is the maximum number of Type2 CPUs reported by the UE during UE capability reporting. For other low priority CSI reports, the UE may report outdated CSI.

[0159] If AI-based CSI processing requires only Type 1 and Type 2 CPUs, the UE may process min{N CPU,1 -n CPU,1 ,N CPU,2} AI-based CSI reports, including NCPU,1 is the maximum number of Type 1 CPUs reported by the UE during UE capability reporting, and n CPU,1 is the number of CPUs used by other higher priority non-AI based CSI reports. For other lower priority CSI reports, the UE may report outdated CSI. In another embodiment, the gNB may avoid configuring or triggering CSI reports that require more Type 1 and Type 2 CPUs than the maximum number of Type 1 and Type 2 CPUs reported by the UE from the UE capability report.

[0160] Fig.21 A signaling diagram 2100 showing CSI reporting based on minimum processing delay. The minimum processing delay for AI-based CSI should depend on whether dedicated hardware can be used. Using dedicated hardware (e.g., NPU), neural network processing should be faster than traditional hardware-based approaches. In addition, the minimum processing delay also depends on the neural network architecture. If the gNB scheduling results in a scheduling offset that is smaller than the minimum processing delay, the UE may decide to report outdated CSI or ignore the DCI without transmitting HARQ-ACK or data on the PUSCH triggered by the DCI.

[0161] The UE may report the minimum processing delay capability for AI-based CSI reporting. The gNB may provide configuration of the CSI framework for multiple CSI reports, where the gNB may provide the configuration for CSI reporting through RRC signaling such as CSI-reportConfig. For semi-persistent and aperiodic CSI reporting, the gNB may send a MAC CE or DCI to trigger the corresponding report, where the MAC is used to activate semi-persistent CSI reporting and the DCI is used to trigger aperiodic CSI reporting. The UE may determine the scheduling offset for CSI reporting so that it may be determined whether to perform AI-based CSI reporting. If the scheduling offset satisfies the minimum processing delay for AI-based CSI reporting, the UE may report AI-based CSI.

[0162] The UE may report the minimum processing delays Z and Z' for AI-based CSI reporting, where Z and Z' may indicate when the CSI request field on the DCI triggers the AI-based CSI reporting on the PUSCH, such that if the first uplink symbol used to carry the corresponding AI-based CSI report including the timing advance effect starts no earlier than symbol Z, and if the first uplink symbol used to carry the nth AI-based CSI report including the timing advance effect starts no earlier than symbol Z', then the UE shall provide a valid AI-based CSI report for the nth triggered report.

[0163] The gNB and UE may determine the minimum processing delays Z and Z' based on whether the UE adopts Type2 CPU for AI-based CSI reporting. In one example, two groups of Z and Z' may be predefined, where for AI-based CSI reporting, the first group may be used for UEs that do not adopt Type2 CPU, and the second group may be used for UEs that adopt Type2 CPU.

[0164] Since the minimum processing delay depends on the ML model, the UE may report the supported Z and Z' for each ML model it supports. The UE may report more than 1 ML model it supports. The gNB may indicate the ML model to be used for AI-based CSI through higher layer signaling, such as RRC signaling in CSI-reportConfig. Then, the corresponding minimum processing delay should apply to AI-based CSI reporting.

[0165] Due to limitations of dedicated hardware, the UE may not be able to perform parallel processing for AI-based CSI reports. The minimum processing delay may then be determined based on the number of AI-based CSI reports used simultaneously for CSI reporting. The minimum processing delay Z and Z' for one AI-based CSI report may be predefined or reported by the UE via UE capabilities. For N AI-based CSI reports, the minimum processing delay may be uNZ and uNZ', where u may be predefined, configured by the gNB via higher layer signaling, such as RRC signaling in CSI-reportConfig, or reported by the UE capabilities in a range within (0, 1]. The UE may also report whether it supports parallel processing as a second UE capability.

[0166] For triggered AI based CSI reporting, if the scheduling offset does not meet the minimum Z or Z' requirement, the UE may ignore the DCI without transmitting HARQ-ACK or data on the PUSCH scheduled by the DCI. For triggered AI based CSI reporting, if the scheduling offset does not meet the minimum Z or Z' requirement, the UE may report outdated CSI for all triggered CSI reports. For triggered AI based CSI reporting, if the scheduling offset does not meet the minimum Z or Z' requirement, the UE may report outdated CSI for triggered CSI reports that do not meet the minimum Z or Z' requirement. The gNB may avoid configuring or triggering AI based CSI reporting where the scheduling offset is less than the minimum processing delay for AI based CSI reporting.

[0167] Fig. 222200 is a diagram illustrating an example of a hardware implementation of a UE device 2202. The device 2202 may be a UE 102, a component of a UE 102, or may implement UE functionality. In some aspects, the device 2202 may include a wireless baseband processor 2224 (also referred to as a modem) coupled to one or more transceivers 2222 (e.g., a wireless RF transceiver). The wireless baseband processor 2224 may include an on-chip memory 2224'. In some aspects, the device 2202 may further include one or more subscriber identity module (SIM) cards 2220 and an application processor 2206 coupled to a secure digital (SD) card 2208 and a screen 2210. The application processor 2206 may include an on-chip memory 2206'.

[0168] The device 2202 may further include a Bluetooth module 2212, a WLAN module 2214, an SPS module 2216 (e.g., a GNSS module), and a cellular module 2217 within one or more transceivers 2222. The Bluetooth module 2212, the WLAN module 2214, the SPS module 2216, and the cellular module 2217 may include an on-chip transceiver (TRX) (or in some cases, only a receiver (RX)). The Bluetooth module 2212, the WLAN module 2214, the SPS module 2216, and the cellular module 2217 may include their own dedicated antennas and / or utilize antenna 2280 for communication. The device 2202 may further include one or more sensor modules 2218 (e.g., an atmospheric pressure sensor / altimeter; a motion sensor such as an inertial management unit (IMU), a gyroscope, and / or an accelerometer; light detection and ranging (LIDAR), radio-aided detection and ranging (RADAR), sound navigation and ranging (SONAR), a magnetometer, audio, and / or other technologies for positioning), an additional module of memory 2226, a power supply 2230, and / or a camera 2232.

[0169] The wireless baseband processor 2224 communicates with another UE 102 and / or with a RU associated with the base station 104 via one or more antennas 2280 through the transceiver 2222. The wireless baseband processor 2224 and the application processor 2206 can each include a computer-readable medium / memory 2224', 2206', respectively. The additional modules of the memory 2226 can also be regarded as computer-readable media / memory. Each computer-readable medium / memory 2224', 2206', 2226 can be non-temporary. The wireless baseband processor 2224 and the application processor 2206 are each responsible for general processing, including executing software stored on a computer-readable medium / memory. When the software is executed by the wireless baseband processor 2224 / application processor 2206, it causes the wireless baseband processor 2224 / application processor 2206 to perform the various functions described. The computer-readable medium / memory can also be used to store data manipulated by the wireless baseband processor 2224 / application processor 2206 when executing the software. The wireless baseband processor 2224 / application processor 2206 may be a component of the UE 102. The device 2202 may be a processor chip (modem and / or application) and include only the wireless baseband processor 2224 and / or application processor 2206, and in another configuration, the device 2202 may be the entire UE 102 and include additional modules of the device 2202.

[0170] As described above, the UE-based CSI processing component 140 is configured to receive a CSI-RS from a network entity. The UE-based CSI processing component 140 is further configured to select a wideband precoder for compressing the CSI report based on the measurement of the CSI-RS. The selection of the wideband precoder is based on the entire bandwidth for receiving the CSI-RS. The UE-based CSI processing component 140 is further configured to compress the CSI report using one or more subband eigenvectors of one or more precoded estimated subband channels through the selected wideband precoder. The UE-based CSI processing component 140 is further configured to transmit a first PMI indicating the wideband precoder and a second PMI indicating the one or more subband eigenvectors to the network entity.

[0171] In a further example, the UE-based CSI processing component 140 is configured to receive one or more CSI-RS from a network entity to transmit one or more CSI reports to the network entity. At least one CSI report of the one or more CSI reports is based on an ML model. When a total payload size of the one or more CSI reports exceeds a threshold, the UE-based CSI processing component 140 is further configured to omit at least a portion of at least one CSI report based on the ML model. The UE-based CSI processing component 140 is further configured to transmit one or more CSI reports to the network entity, wherein at least the portion of at least one CSI report is omitted from the transmission.

[0172] In still further examples, the UE-based CSI processing component 140 is configured to receive a configuration for a CSI report processing procedure from a network entity. The configuration is for at least one of: (a) processing multiple CSI reports using an ML model, or (b) processing a first set of CSI reports using an ML model without processing a second set of CSI reports. The UE-based CSI processing component 140 is further configured to receive multiple CSI-RSs from the network entity. Measurements of the multiple CSI-RSs are based on UE capabilities for the CSI report processing procedure. The UE-based CSI processing component 140 is further configured to transmit a CSI report to the network entity based on the UE capabilities for the CSI report processing procedure.

[0173] The UE-based CSI processing component 140 may be within the wireless baseband processor 2224, the application processor 2206, or both the wireless baseband processor 2224 and the application processor 2206. The UE-based CSI processing component 140 may be one or more hardware components explicitly configured to perform the process / algorithm, implemented by one or more processors configured to perform the process / algorithm, stored in a computer-readable medium for implementation by one or more processors, or some combination thereof.

[0174] As shown, the device 2202 may include various components configured for various functions. In one configuration, the device 2202 and in particular the wireless baseband processor 2224 and / or the application processor 2206 include: a component for receiving a CSI-RS from a network entity; a component for selecting a wideband precoder for compressing a CSI report based on a measurement of the CSI-RS, wherein the selection of the wideband precoder is based on the entire bandwidth for receiving the CSI-RS; a component for compressing the CSI report using one or more subband eigenvectors of one or more precoded subband channels through the selected wideband precoder; and a component for transmitting / sending a first PMI indicating a wideband precoder and a second PMI indicating one or more subband eigenvectors to the network entity. The apparatus 2202 further includes: means for receiving one or more channel state information-reference signals (CSI-RS) from a network entity to transmit one or more CSI reports to the network entity, at least one of the one or more CSI reports being based on an ML model; means for omitting at least a portion of at least one CSI report based on the ML model when a total payload size of the one or more CSI reports exceeds a threshold; and means for transmitting the one or more CSI reports to the network entity, wherein at least a portion of at least one CSI report is omitted from the transmission. The apparatus 2202 further includes: means for receiving a configuration for a CSI report processing procedure from the network entity, wherein the configuration is for at least one of: (a) processing a plurality of CSI reports using the ML model, or (b) processing a first set of CSI reports using the ML model without processing a second set of CSI reports; means for receiving a plurality of CSI-RSs from the network entity, wherein measurements of the plurality of CSI-RSs are based on UE capabilities for the CSI report processing procedure; and means for transmitting the CSI report to the network entity based on the UE capabilities for the CSI report processing procedure. The means may be the UE-based CSI processing component 140 of the apparatus 2202 configured to perform the functions recited by the means.

[0175] Fig.23 2300 is a diagram illustrating an example of a hardware implementation of one or more network entities 2304. The one or more network entities 2304 may be a base station, a component of a base station, or may implement base station functionality. The one or more network entities 2304 may include at least one of a CU 110, a DU 108, or a RU 106. For example, depending on the layer functionality handled by the network-based CSI processing component 150, the one or more network entities 2304 may include a CU 110; both a CU 110 and a DU 108; each of a CU 110, a DU 108, and a RU 106; a DU 108; both a DU 108 and a RU 106; or a RU 106.

[0176] CU 110 may include a CU processor 2312. CU processor 2312 may include on-chip memory 2312'. In some aspects, CU 110 may further include an additional memory module 2314 and a communication interface 2318. CU 110 communicates with DU 108 via a mid-range link (such as an F1 interface). DU 108 may include a DU processor 2332. DU processor 2332 may include on-chip memory 2332'. In some aspects, DU 108 may further include an additional memory module 2334 and a communication interface 2338. DU 108 communicates with RU 106 via a front-range link. RU 106 may include a RU processor 2342. RU processor 2342 may include on-chip memory 2342'. In some aspects, RU 106 may further include an additional memory module 2344, one or more transceivers 2346, an antenna 2380, and a communication interface 2348. RU 106 communicates wirelessly with UE 102 .

[0177] On-chip memory 2312', 2332', 2342' and additional memory modules 2314, 2334, 2344 can each be considered as a computer-readable medium / memory. Each computer-readable medium / memory can be non-temporary. Each of the processors 2312, 2332, 2342 is responsible for general processing, including executing software stored on the computer-readable medium / memory. The software, when executed by the corresponding processor, causes the processor to perform the various described functions. The computer-readable medium / memory can also be used to store data manipulated by the processor when executing the software.

[0178] As described above, the network-based CSI processing component 150 is configured to transmit a CSI-RS to a UE to select a wideband precoder for compressing a CSI report based on a measurement of the CSI-RS. The selection of the wideband precoder is based on the entire bandwidth used to transmit the CSI-RS. The network-based CSI processing component 150 is further configured to receive a CSI report from the UE, the CSI report being compressed by the selected wideband precoder using one or more subband eigenvectors of one or more precoded subband channels, a first PMI indicating the wideband precoder, and a second PMI indicating the one or more subband eigenvectors. The network-based CSI processing component 150 is further configured to decompress the CSI report using the first PMI and the second PMI as inputs to an ML model, which outputs a decompressed CSI report.

[0179] In a further example, the network-based CSI processing component 150 is configured to transmit one or more CSI-RSs to the UE to obtain one or more CSI reports. At least one CSI report of the one or more CSI reports is based on the ML model. The network-based CSI processing component 150 is further configured to receive the one or more CSI reports from the UE, wherein when a total payload size of the one or more CSI reports exceeds a threshold, at least a portion of the at least one CSI report based on the ML model is omitted from receiving the one or more CSI reports.

[0180] In still further examples, the network-based CSI processing component 150 is configured to transmit a configuration for a CSI report processing procedure to the UE. The configuration is for at least one of: (a) processing multiple CSI reports using the ML model, or (b) processing a first set of CSI reports using the ML model without processing a second set of CSI reports. The network-based CSI processing component 150 is further configured to transmit multiple CSI-RSs to the UE. The measurement of the multiple CSI-RSs is based on the UE capabilities for the CSI report processing procedure. The network-based CSI processing component 150 is further configured to transmit a CSI report to the UE based on the UE capabilities for the CSI report processing procedure.

[0181] The network-based CSI processing component 150 may be within one or more processors of one or more of the CU 110, DU 108, and RU 106. The network-based CSI processing component 150 may be one or more hardware components explicitly configured to perform the described process / algorithm, implemented by one or more processors configured to perform the described process / algorithm, stored within a computer-readable medium for implementation by one or more processors, or some combination thereof.

[0182] The one or more network entities 2304 may include various components configured for various functions. In one configuration, the one or more network entities 2304 include: a component for transmitting a CSI-RS to a UE to select a wideband precoder for compressing a CSI report based on a measurement of the CSI-RS, the selection of the wideband precoder based on the entire bandwidth used to transmit the CSI-RS; a component for receiving a CSI report from the UE, the CSI report being compressed by the selected wideband precoder using one or more subband eigenvectors of one or more precoded subband channels, a first PMI indicating the wideband precoder, and a second PMI indicating the one or more subband eigenvectors; and a component for decompressing the CSI report using the first PMI and the second PMI as inputs to an ML model, the ML model outputting a decompressed CSI report. The one or more network entities 2304 further include: means for transmitting one or more CSI-RS to a UE to obtain one or more CSI reports, at least one of the one or more CSI reports being based on an ML model; and means for receiving one or more CSI reports from the UE, wherein at least a portion of at least one CSI report based on the ML model is omitted from receiving the one or more CSI reports when a total payload size of the one or more CSI reports exceeds a threshold. The one or more network entities 2304 further include: means for transmitting a configuration for a CSI report processing procedure to the UE, wherein the configuration is for at least one of the following: (a) processing multiple CSI reports using an ML model, or (b) processing a first set of CSI reports using an ML model without processing a second set of CSI reports; means for transmitting multiple CSI-RS to the UE, wherein measurements of the multiple CSI-RS are based on UE capabilities for the CSI report processing procedure; and means for transmitting the CSI report to the UE based on the UE capabilities for the CSI report processing procedure. The means may be a network-based CSI processing component 150 of the one or more network entities 2304 configured to perform the functions recited by the means.

[0183] The specific order or hierarchy of the boxes in the process and flow chart disclosed herein is an illustration of an example method. Therefore, the specific order or hierarchy of the boxes in the process and flow chart can be rearranged. Some boxes can also be merged or deleted. The attached method claims present the elements of each box in the example order, and are not limited to the specific order or hierarchy presented in the claims, process and flow chart.

[0184] The detailed description set forth herein describes various configurations in conjunction with the accompanying drawings, but does not represent the only configuration in which the concepts described herein can be practiced. The detailed description includes specific details for providing a comprehensive explanation of the various concepts. However, these concepts can be practiced without using these specific details. In some cases, well-known structures and components are shown in block diagram form in order to avoid blurring such concepts.

[0185] Various aspects of wireless communication systems (such as telecommunication systems) are presented with reference to various apparatus and methods. These apparatus and methods are described in the detailed description that follows and are illustrated in the accompanying drawings by various blocks, components, circuits, processes, call flows, systems, algorithms, etc. (collectively, "elements"). These elements may be implemented using electronic hardware, computer software, or a combination thereof. Whether such elements are implemented as hardware or software depends on the specific application and design constraints imposed on the overall system.

[0186] Elements, or any part of elements or any combination of elements can be implemented as a "processing system" including one or more processors. Examples of processors include microprocessors, microcontrollers, graphics processing units (GPUs), central processing units (CPUs), application processors, digital signal processors (DSPs), reduced instruction set computing (RISC) processors, systems on chip (SoCs), baseband processors, field programmable gate arrays (FPGAs), programmable logic devices (PLDs), state machines, gating logic, discrete hardware circuits, and other similar hardware configured to perform various functions described throughout this disclosure. One or more processors in a processing system can execute software, which can be referred to as software, firmware, middleware, microcode, hardware description language, or other. Software should be broadly interpreted as meaning instructions, instruction sets, codes, code segments, program codes, programs, subroutines, software components, applications, software applications, software packages, routines, subroutines, objects, executables, execution threads, processes, functions, or any combination thereof.

[0187] If the functions described herein are implemented in software, the functions may be stored on a computer-readable medium (such as a non-transitory computer-readable storage medium) or encoded as one or more instructions or codes on the computer-readable medium. Computer-readable media include computer storage media and may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable ROM (EEPROM), optical disk storage devices, magnetic disk storage devices, other magnetic storage devices, combinations of these types of computer-readable media, or any other medium that can be used to store computer-executable code in the form of computer-accessible instructions or data structures. The storage medium can be any available medium that is accessible to a computer.

[0188] The aspects, implementations, and / or use cases described herein may be implemented across many different platform types, devices, systems, shapes, sizes, and packaging arrangements. For example, the aspects, implementations, and / or use cases may be generated via integrated chip implementations and other non-module component-based devices such as end-user devices, vehicles, communication devices, computing devices, industrial equipment, retail / procurement devices, medical devices, artificial intelligence (AI)-enabled devices, machine learning (ML)-enabled devices, and the like. The aspects, implementations, and / or use cases may range from chip-level or modular components to non-modular or non-chip-level implementations, and further to aggregated, distributed, or original equipment manufacturer (OEM) devices or systems incorporating one or more of the techniques described herein.

[0189] Devices incorporating the various aspects and features described herein may also include additional components and features for implementing and practicing the various aspects and features claimed and described. For example, the transmission and reception of wireless signals necessarily include many components for analog and digital purposes, such as hardware components, antennas, RF chains, power amplifiers, modulators, buffers, processors, interleavers, adders / summers, etc. The techniques described herein may be practiced in a wide variety of devices, chip-level components, systems, distributed arrangements, aggregated or disaggregated components, end-user devices, etc., in various configurations.

[0190] The description herein is provided to enable those skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects. Therefore, the claims are not limited to the various aspects described herein, but should be interpreted in view of the full scope of the disclosure consistent with the language of the claims.

[0191] Unless explicitly stated, references to singular elements do not mean "one and only one", but rather "one or more". Terms such as "if", "when ..." and "at ..." do not imply an immediate temporal relationship or reaction. That is, these phrases (such as "when ...") do not imply an immediate action in response to the occurrence of an action or during the occurrence of an action, but simply mean that if a certain condition is met, an action will occur, but no specific or immediate time constraints are required for the occurrence of the action. Unless explicitly stated otherwise, the term "some" refers to one or more. Combinations such as "at least one of A, B, or C" or "one or more of A, B, or C" include any combination of A, B, and / or C, such as A and B, A and C, B and C, or A and B and C, and may include multiple A, multiple B, and / or multiple C, or may include only A, only B, or only C. A set should be interpreted as a set of elements in which the number of elements is one or more.

[0192] The structural equivalents and functional equivalents of the elements of various aspects described in the entire disclosure known or later learned by those of ordinary skill in the art are expressly incorporated herein by reference and are covered by the claims. The words "module", "mechanism", "element", "device", etc. may not be substitutes for the word "component". Therefore, unless the claim element is clearly stated using the phrase "component for...", any claim element shall not be interpreted as a means plus function. As used herein, the phrase "based on" should not be interpreted as a reference to a closed information set, one or more conditions, one or more factors, etc. In other words, unless clearly stated differently, the phrase "based on A" (where "A" can be information, conditions, factors, etc.) should be interpreted as "at least based on A".

[0193] The following examples are merely illustrative and may be combined with other examples or teachings described herein without limitation.

[0194] Example 1 is a method for performing wireless communication at a UE, comprising: receiving a CSI-RS from a base station; selecting a wideband precoder W1 based on the CSI-RS; compressing a subband feature vector based on the CSI-RS and the wideband precoder W1; and sending a first PMI indicating the wideband precoder W1 and a second PMI indicating the compressed subband feature vector to the base station.

[0195] Example 2 may be combined with Example 1, and further includes receiving an RRC message from the base station, the RRC message configuring a wideband precoder codebook and the number of beams of the wideband precoder W1.

[0196] Example 3 may be combined with any one of Examples 1 to 2, and includes the RRC message being an RRC reconfiguration message.

[0197] Example 4 can be combined with any one of Examples 1 to 3, and further comprises, before compression, determining the subband feature vector as The first v columns of the eigenvector of , where N indicates the number of CSI-RS resource elements of subband S; is the channel estimated based on the CSI-RS at resource element k.

[0198] Example 4 can be combined with any one of Examples 1 to 3, and further includes compressing the sub-band feature vector based on an AI / ML model.

[0199] Example 5 can be combined with any one of Examples 1 to 4, and further includes receiving an RRC message for configuring the AI / ML model from the base station.

[0200] Example 6 may be combined with any one of Examples 1 to 5, and include that the AI / ML model is a predefined AI / ML model or is predetermined by the UE.

[0201] Example 7 can be combined with any one of Examples 1 to 6, and includes sending the first PMI and the second PMI further comprising: generating a PUCCH transmission using a short PUCCH format, and including the first PMI and the second PMI in a single part of the short PUCCH format; and transmitting the PUCCH transmission to the base station.

[0202] Example 8 can be combined with any one of Examples 1 to 7, and includes sending the first PMI and the second PMI further comprising: generating a PUCCH transmission using a long PUCCH format, and including the first PMI and the second PMI in CSI part 2 of the long PUCCH format; and transmitting the PUCCH transmission to the base station.

[0203] Example 9 can be combined with any one of Examples 1 to 8, and includes sending the first PMI and the second PMI further including: generating a PUCCH transmission using a long PUCCH format, and including the first PMI in CSI part 1 of the short PUCCH format and including the second PMI in CSI part 2; and transmitting the PUCCH transmission to the base station.

[0204] Example 10 can be combined with any one of Examples 1 to 9, and includes sending the first PMI and the second PMI further comprising: generating a PUSCH transmission, and including the first PMI and the second PMI in a CSI part 2 of the PUSCH transmission; and transmitting the PUSCH transmission to the base station.

[0205] Example 11 can be combined with any one of Examples 1 to 10, and includes sending the first PMI and the second PMI further including: generating a PUSCH transmission, and including the first PMI in CSI part 1 of the PUSCH transmission and including the second PMI in CSI part 2; and transmitting the PUSCH transmission to the base station.

[0206] Example 12 is a method for performing wireless communication at a base station, comprising: transmitting a CSI-RS to a UE; transmitting a first RRC message to the UE, the first RRC message configuring a wideband precoder codebook and the number of beams of the wideband precoder W1; receiving a first precoder matrix indicator (PMI) indicating the wideband precoder W1 and a second PMI indicating a compressed subband feature vector from the UE; decompressing the compressed subband feature vector to obtain an uncompressed subband feature vector; and transmitting a PDSCH signal using a precoder based on the uncompressed subband feature vector.

[0207] Example 13 can be combined with Example 12, and includes deriving the precoder based on the uncompressed subband eigenvector and the wideband PMI.

[0208] Example 14 can be combined with any one of Examples 12 to 13, and further includes decompressing the compressed sub-band feature vector based on the ML model to obtain an uncompressed sub-band feature vector based on the ML model.

[0209] Example 15 can be combined with any one of Examples 12 to 14, and further includes transmitting a second RRC message configuring the AI / ML model to the UE.

[0210] Example 16 may be combined with any one of Examples 12 to 15, and include that the AI / ML model is a predefined AI / ML model or is predetermined by the UE.

[0211] Example 17 may be combined with any one of Examples 12 to 16, and including receiving the first PMI and the second PMI further including: receiving a PUCCH transmission including the first PMI and the second PMI according to a short PUCCH format.

[0212] Example 18 may be combined with any one of Examples 12 to 17, and including receiving the first PMI and the second PMI further including: receiving a PUCCH transmission including the first PMI and the second PMI in CSI part 2 according to a long PUCCH format.

[0213] Example 19 may be combined with any one of Examples 12 to 18, and including receiving the first PMI and the second PMI further comprising: receiving a PUCCH transmission including the first PMI in CSI part 1 and the second PMI in CSI part 2 according to a long PUCCH format.

[0214] Example 20 may be combined with any one of Examples 12 to 19, and including receiving the first PMI and the second PMI further including: receiving a physical uplink shared channel (PUSCH) transmission including the first PMI and the second PMI in CSI part 2.

[0215] Example 21 may be combined with any one of Examples 12 to 20, and including receiving the first PMI and the second PMI further including: receiving a PUSCH transmission including the first PMI in CSI part 1 and the second PMI in CSI part 2.

[0216] Example 22 is a method for performing wireless communications at a UE, comprising: receiving one or more CSI-RSs from a base station; obtaining uplink resources to be used for CSI reporting; obtaining at least one ML-processed CSI report based on at least one first CSI-RS among the one or more CSI-RSs; obtaining at least one non-AI-based CSI report based on at least one second CSI-RS among the one or more CSI-RSs; identifying that the uplink resources of CSI part 2 are insufficient to carry at least one AI-based CSI report and at least one non-AI-based CSI report; determining the priority of the at least one AI-based CSI report and at least one non-AI-based CSI report; omitting one or some of the at least one AI-based CSI report and / or the at least one non-AI-based CSI report from CSI part 2 based on the priority; and transmitting the remaining CSI reports on the uplink resources to the base station in CSI part 2.

[0217] Example 23 may be combined with Example 22, and include that the first at least one first CSI-RS and the at least one second CSI-RS may include the same CSI-RS and / or different CSI-RS.

[0218] Example 24 can be combined with any one of Examples 22 to 23, and includes that the uplink resource is a resource for PUSCH transmission.

[0219] Example 25 can be combined with any one of Examples 22 to 24, and includes that the uplink resource is a resource for PUCCH transmission.

[0220] Example 26 can be combined with any one of Examples 22 to 24, and further includes receiving at least one first CSI report configuration for configuring an AI-based CSI report from the base station; and receiving at least one second CSI report configuration for configuring a non-AI-based CSI report from the base station; and includes the UE generating the at least one AI-based CSI report based on the at least one first CSI report configuration and at least one received first CSI-RS, and generating the at least one non-AI-based CSI report based on the at least one second CSI report configuration and at least one received second CSI-RS.

[0221] Example 27 can be combined with any one of Examples 22 to 26, and includes determining the priority of the at least one AI-based CSI report and the at least one non-AI-based CSI report further including: determining the priority of each CSI report in the at least one AI-based CSI report and the at least one non-AI-based CSI report based on time domain reporting behavior, CSI-reportConfigId, serving cell ID, number of reports and / or whether the CSI report is AI-based or non-AI-based CSI report.

[0222] Example 28 can be combined with any one of Examples 22 to 27, and includes that the determining and omitting steps further include: determining that the at least one non-AI based CSI report has a lower priority than the at least one AI based CSI report; and omitting one, some or all of the at least one non-AI based CSI report based on the low priority.

[0223] Example 29 may be combined with any one of Examples 22 to 28, and include that the determining and omitting steps further include: determining that the at least one AI-based CSI report has a lower priority than the at least one non-AI-based CSI report; and omitting one, some or all of the at least one AI-based CSI report based on the low priority.

[0224] Example 30 is a method for performing wireless communication at a UE, comprising: receiving a CSI-RS from a base station; obtaining uplink resources to be used for CSI reporting; determining a first PMI, a second PMI, and a third PMI for AI-based CSI reporting based on the CSI-RS, which respectively indicate a wideband precoder, a compressed feature vector of a channel with a selected wideband precoder in an even subband, and a compressed feature vector of a channel with a selected wideband precoder in an odd subband; identifying that the uplink resources of CSI part 2 are insufficient to carry the first PMI, the second PMI, and the third PMI; determining the priorities of the first PMI, the second PMI, and the third PMI; based on the priority CSI part 2, omitting one or some of the first PMI, the second PMI, and the third PMI from CSI part 2; and after the omission, transmitting another one of the first PMI, the second PMI, and the third PMI on the uplink resources of CSI part 2 to the base station.

[0225] Example 31 can be combined with Example 30 and include that the uplink resources are resources for PUSCH transmission.

[0226] Example 32 can be combined with any one of Examples 30 to 31, and includes that the uplink resource is a resource for PUCCH transmission.

[0227] Example 33 may be combined with any one of Examples 30 to 32, and further include receiving a CSI report configuration for configuring an AI-based CSI report from the base station; and including the UE receiving the CSI-RS based on the CSI report configuration.

[0228] Example 34 can be combined with any one of Examples 30 to 33, and includes determining the priority of the PMI and the omission step further includes: determining that the second PMI and the third PMI have a lower priority than the first PMI; and omitting the second PMI and / or the third PMI based on the low priority.

[0229] Example 35 can be combined with any one of Examples 30 to 34, and includes determining the priority of the PMI and the omitting step further includes: determining that the first PMI has a lower priority than the second PMI and the third PMI; and omitting the first PMI based on the low priority.

[0230] Example 36 may be combined with any one of Examples 30 to 35, and include that the second PMI has a higher priority than the third PMI.

[0231] Example 37 may be combined with any one of Examples 30 to 36, and include the second PMI having a lower priority than the third PMI.

[0232] Example 38 can be combined with any one of Examples 30 to 37, and includes that the uplink resources are resources for PUSCH transmission.

[0233] Example 39 can be combined with any one of Examples 30 to 38, and includes that the uplink resource is a resource for PUCCH transmission.

[0234] [Start] Example 40 is a method for wireless communication at a UE, comprising: receiving a channel state information reference signal (CSI-RS) from a base station; obtaining uplink resources to be used for channel state information (CSI) reporting; determining a first PMI and a second PMI for artificial intelligence (AI)-based CSI reporting based on the CSI-RS, which respectively indicate a wideband precoder and a compressed feature vector for channels with a selected wideband precoder in all subbands; identifying that the uplink resources of CSI part 2 are insufficient to carry the first PMI, the second PMI; determining a priority of the first PMI and the second PMI; omitting one of the first PMI and the second PMI in CSI part 2 based on the priority; and after the omission, transmitting the other of the first PMI and the second PMI on the uplink resources of CSI part 2 to the base station.

[0235] Example 41 may be combined with Example 40 and include determining the priority of the PMI and the omitting step further includes: determining that the first PMI has a higher priority than the second PMI; and omitting part or all of the second PMI based on the low priority.

[0236] Example 42 can be combined with any one of Examples 40 to 41, and includes determining the priority of the PMI and the omitting step further includes: determining that the second PMI has a higher priority than the first PMI; and omitting the first PMI based on the low priority.

[0237] Example 43 may be combined with any one of Examples 40 to 42, and include that for the second PMI, some bits may have a higher priority than other bits of the PMI, and lower priority bits may be omitted.

[0238] Example 44 may be combined with any of Examples 40 to 43, and include that the higher priority bit and the lower priority bit may be determined based on a single AL / ML model.

[0239] Example 45 may be combined with any one of Examples 40 to 44, and include that the higher priority bit and the lower priority bit may be determined based on different AL / ML models.

[0240] Example 46 can be combined with any one of Examples 40 to 45, and include: the higher priority bit can be predefined or configured through RRC signaling.

[0241] Example 47 can be combined with any one of Examples 40 to 45, and includes that the uplink resources are resources for PUSCH transmission.

[0242] Example 48 can be combined with any one of Examples 40 to 46, and includes that the uplink resources are resources for PUCCH transmission.

[0243] Example 49 is a method for performing wireless communication at a UE, comprising: receiving one or more CSI-RSs from a base station; obtaining uplink resources to be used for CSI reporting; determining at least one first feedback element for AI-based CSI reporting based on at least one first CSI-RS among the one or more CSI-RSs; determining at least one second feedback element for non-AI-based CSI reporting based on at least one second CSI-RS among the one or more CSI-RSs; identifying that the uplink resources of CSI part 2 are insufficient to carry the at least one first feedback element and the at least one second feedback element; determining the priority of the at least one first feedback element and the at least one second feedback element; based on the priority, omitting one or some of the at least one first feedback element and / or the at least one second feedback element in CSI part 2; and transmitting the remaining feedback elements on the uplink resources of CSI part 2 after the omission.

[0244] Example 50 may be combined with Example 49 and include that the first at least one first CSI-RS and the at least one second CSI-RS may include the same CSI-RS and / or different CSI-RS.

[0245] Example 51 can be combined with any one of Examples 49 to 50, and includes that the uplink resources are resources for PUSCH transmission.

[0246] Example 52 can be combined with any one of Examples 49 to 51, and includes that the uplink resources are resources for PUCCH transmission.

[0247] Example 53 can be combined with any one of Examples 49 to 52, and further includes: receiving at least one first CSI report configuration for configuring an AI-based CSI report from the base station; and receiving at least one second CSI report configuration for configuring a non-AI-based CSI report from the base station; and including: the UE obtains the at least one first feedback element and / or receives the at least one first CSI-RS based on the at least one first CSI report configuration, and obtains the at least one second feedback element and / or receives the at least one second CSI-RS based on the at least one second CSI report configuration.

[0248] Example 54 can be combined with any one of Examples 49 to 53, and includes that the determining and omitting steps further include: determining that the at least one second feedback element has a lower priority than the at least one first feedback element; and omitting one, some or all of the at least one non-AI-based CSI reports based on the low priority.

[0249] Example 55 can be combined with any one of Examples 49 to 54, and include that the determining and omitting steps further include: determining that the at least one first feedback element has a lower priority than the at least one second feedback element; and omitting one, some or all of the at least one AI-based CSI reports based on the low priority.

[0250] Example 56 is a method for performing wireless communication at a UE, comprising: receiving a CSI-RS from a base station; obtaining uplink resources to be used for CSI reporting; determining a set of PMIs for AI-based CSI reporting based on at least one AI / ML model and the CSI-RS, which indicates compressed feature vectors of channels with a selected wideband precoder in all subbands; identifying that the uplink resources of CSI part 2 are insufficient to carry some of the set of PMIs; and transmitting to the base station a PMI in the set of PMIs having a maximum bit width on the uplink resources of CSI part 2 that can fit the maximum payload size of the uplink resource.

[0251] Example 57 can be combined with Example 45 and include that the AI / ML model can be predefined or configured through RRC signaling.

[0252] Example 58 can be combined with any one of Examples 56 to 57, and includes that the uplink resources are resources for PUSCH transmission.

[0253] Example 59 can be combined with any one of Examples 56 to 58, and includes that the uplink resources are resources for PUCCH transmission.

[0254] Example 60 is a method for performing wireless communication at a UE, comprising: receiving a CSI-RS from a base station; obtaining uplink resources to be used for CSI reporting: determining a first PMI for AI-based CSI reporting based on the CSI-RS, which indicates a compressed feature vector of channels in all subbands; determining a second PMI for non-artificial intelligence (non-AI)-based CSI reporting based on the CSI-RS and a codebook; identifying that the uplink resources of CSI part 2 are insufficient to carry the first PMI; and transmitting a second PMI on the uplink resources of CSI part 2 to the base station, wherein the CSI of the first PMI is omitted.

[0255] Example 61 can be combined with Example 60, and include that the codebook can be predefined or configured through RRC signaling.

[0256] Example 62 can be combined with any one of Examples 60 to 61, and includes that the uplink resources are resources for PUSCH transmission.

[0257] Example 63 can be combined with any one of Examples 60 to 62, and includes that the uplink resources are resources for PUCCH transmission.

[0258] Example 64 is a method for performing wireless communications at a base station, comprising: transmitting one or more CSI-RSs; transmitting control signaling to a UE to grant uplink resources to be used for CSI reporting; transmitting control signaling to the UE to configure at least one AI-based CSI report based on at least one first CSI-RS among the one or more CSI-RSs; transmitting control signaling to the UE to configure at least one non-AI-based CSI report based on at least one second CSI-RS among the one or more CSI-RSs; receiving a transmission including one or more CSI reports from the UE on the uplink resources; identifying that the uplink resources of CSI part 2 are insufficient to carry the at least one AI-based CSI report and the at least one non-AI-based CSI report; determining to omit the AI-based CSI report and / or the non-AI-based CSI report in CSI part 2 from the transmission or multiple CSI reports; and determining or identifying the AI-based CSI report and / or the non-AI-based CSI report to be used for MIMO communication with the encoder and considered for omission based on the one or more CSI reports.

[0259] Example 65 can be combined with Example 64 and further includes transmitting a PDSCH transmission to the UE using the precoder.

[0260] Example 66 can be combined with any one of Examples 64 to 65, and includes determining that the precoder further includes: determining or identifying the CSI-RS associated with the one or more CSI reports based on omitting information of the AI-based CSI report and / or the non-AI-based CSI report; determining or identifying the precoder based on the association and the one or more CSI reports.

[0261] Example 67 may be combined with any one of Examples 64 to 66, and include that the first at least one first CSI-RS and the at least one second CSI-RS may include the same CSI-RS and / or different CSI-RS.

[0262] Example 68 can be combined with any one of Examples 64 to 67, and includes that the uplink resources are resources for PUSCH transmission.

[0263] Example 69 can be combined with any one of Examples 64 to 68, and includes that the uplink resources are resources for PUCCH transmission.

[0264] Example 70 may be combined with any one of Examples 64 to 69, and include determining to omit an AI-based CSI report and / or a non-AI-based CSI report from the transmission, or the one or more CSI reports are based on a CSI omission rule.

[0265] Example 71 can be combined with any one of Examples 64 to 70, and includes that the CSI omission rule includes that the AI-based CSI report and the non-AI-based CSI report have a lower priority, and the base station determines to omit the AI-based CSI report and the non-AI-based CSI report based on the low priority.

[0266] Example 72 can be combined with any one of Examples 64 to 71, and includes that the CSI omission rule includes that the AI-based CSI report has a lower priority than the non-AI-based CSI report, and the UE determines to omit the AI-based CSI report based on the low priority.

[0267] Example 73 can be combined with any one of Examples 64 to 72, and includes that the CSI omission rule includes that the non-AI based CSI report has a lower priority than the AI ​​based CSI report, and the base station determines to omit the AI ​​based CSI report based on the low priority.

[0268] Example 74 is a method for performing wireless communication at a base station, comprising: transmitting CSI-RS; transmitting control signaling to a UE to grant uplink resources to be used for CSI reporting; receiving a CSI report from the UE on the uplink resources; identifying that the uplink resources of CSI part 2 are insufficient to carry a first PMI, a second PMI, and a third PMI, which respectively indicate a wideband precoder, a compressed characteristic vector of a channel with a selected wideband precoder in an even subband, and a compressed characteristic vector of a channel with a selected wideband precoder in an odd subband; determining to omit one or some of the first PMI, the second PMI, and the third PMI in the CSI report; and determining or identifying a precoder for MIMO communication based on the remaining PMIs in the CSI report.

[0269] Example 75 can be combined with Example 74 and further include transmitting a PDSCH transmission to the UE using the precoder.

[0270] Example 76 can be combined with any one of Examples 74 to 75, and includes: the base station determines to omit one or some of the first PMI, the second PMI and the third PMI in the CSI report based on a CSI omission rule.

[0271] Example 77 can be combined with any one of Examples 74 to 76, and includes that the CSI omission rule includes: the second PMI and / or the third PMI has a lower priority than the first PMI, and the base station determines to omit the second PMI and / or the third PMI based on the low priority.

[0272] Example 78 can be combined with any one of Examples 74 to 77, and includes that the CSI omission rule includes: the second PMI has a lower priority than the first PMI and the third PMI, and the base station determines to omit the second PMI based on the low priority.

[0273] Example 79 can be combined with any one of Examples 74 to 78, and includes that the CSI omission rule includes: the third PMI has a lower priority than the first PMI and the second PMI, and the base station determines to omit the third PMI based on the low priority.

[0274] Example 80 can be combined with any one of Examples 74 to 79, and includes that the CSI omission rule includes: the first PMI has a lower priority than the second PMI and the third PMI, and the base station determines to omit the first PMI based on the low priority.

[0275] Example 81 can be combined with any one of Examples 74 to 80, and includes that the uplink resources are resources for PUSCH transmission.

[0276] Example 82 can be combined with any one of Examples 74 to 81, and includes that the uplink resources are resources for PUCCH transmission.

[0277] Example 83 is a method for performing wireless communication at a base station, comprising: transmitting a CSI-RS; transmitting control signaling to a UE to grant uplink resources to be used for CSI reporting; receiving a CSI report from the UE on the uplink resources; identifying that the uplink resources of CSI part 2 are insufficient to carry a first PMI and a second PMI, which respectively indicate a wideband precoder and a compressed feature vector of a channel with a selected wideband precoder in all subbands; determining to omit one or both of the first PMI and the second PMI or a first part (e.g., some bits) of the second PMI in the CSI report; and determining or identifying a precoder for MIMO communication based on the remaining PMI in the CSI report or the remaining part of the second PMI.

[0278] Example 84 may be combined with Example 83 and include if the base station determines to omit both the first PMI and the second PMI, the base station determining or identifying the precoder for MIMO communication based on CSI part 1 in the CSI report.

[0279] Example 85 can be combined with any one of Examples 83 to 84, and includes that the CSI part 1 includes CQI and / or RI.

[0280] Example 86 can be combined with any one of Examples 83 to 85, and further includes transmitting a PDSCH transmission to the UE using the precoder.

[0281] Example 87 can be combined with any one of Examples 83 to 86, and includes the base station determining to omit one or both of the first PMI and the second PMI or a portion of the second PMI in the CSI report based on a CSI omission rule.

[0282] Example 88 can be combined with any one of Examples 83 to 87, and includes that the CSI omission rule includes: the second PMI has a lower priority than the first PMI, and the base station determines to omit the second PMI based on the low priority.

[0283] Example 89 can be combined with any one of Examples 83 to 88, and includes that the CSI omission rule includes: the first PMI has a lower priority than the second PMI, and the base station determines to omit the first PMI based on the low priority.

[0284] Example 90 can be combined with any one of Examples 83 to 89, and include that the CSI omission rule includes: the first part of the second PMI has a lower priority than the remaining part (e.g., other bits) of the second PMI, and the base station determines to omit the first part based on the low priority.

[0285] Example 91 can be combined with any one of Examples 83 to 90, and includes determining the first portion and the remaining portion based on a single AL / ML model.

[0286] Example 92 can be combined with any one of Examples 83 to 91, and includes determining the first portion and the remaining portion based on different AL / ML models.

[0287] Example 93 can be combined with any one of Examples 83 to 92, and includes the first part and the remaining part being predefined or configured by the base station via RRC signaling.

[0288] Example 94 can be combined with any one of Examples 83 to 93, and includes that the uplink resources are resources for PUSCH transmission.

[0289] Example 95 can be combined with any one of Examples 83 to 94, and includes that the uplink resource is a resource for PUCCH transmission.

[0290] Example 96 is a method for performing wireless communications at a base station, comprising: transmitting a CSI-RS; transmitting control signaling to a UE to grant uplink resources to be used for CSI reporting; determining a bit width of a set of PMIs for an AI-based CSI report based on at least one AI / ML model and the CSI-RS, which indicates compressed feature vectors of channels in all subbands; identifying a PMI with a maximum bit width that can fit on the uplink resource to the base station for a maximum payload size of the uplink resource for CSI part 2; and determining or identifying a precoder for MIMO communication based on the decoded PMI in the AI-based CSI report.

[0291] Example 97 can be combined with Example 96 and include that the AI / ML model can be predefined or configured through RRC signaling.

[0292] Example 98 can be combined with any one of Examples 96 to 97, and includes that the uplink resources are resources for PUSCH transmission.

[0293] Example 99 can be combined with any one of Examples 96 to 98, and includes that the uplink resources are resources for PUCCH transmission.

[0294] Example 100 is a method for performing wireless communications at a base station, comprising: transmitting a CSI-RS; transmitting control signaling to a UE to grant uplink resources to be used for CSI reporting; receiving a CSI report from the UE on the uplink resources; identifying that the uplink resources of CSI part 2 are insufficient to carry a first PMI and a second PMI, wherein the first PMI is an AI-based CSI report based on the CSI-RS and indicates a compressed feature vector of channels in all subbands, and the second PMI is a non-artificial intelligence (non-AI)-based CSI report based on the CSI-RS; and determining or identifying a precoder for MIMO communication based on the second PMI and a codebook on the uplink resource.

[0295] Example 101 may be combined with Example 100, and including that the codebook may be predefined or configured through RRC signaling.

[0296] Example 102 may be combined with any one of Examples 100 to 101 and include that the uplink resources are resources for PUSCH transmission.

[0297] Example 103 can be combined with any one of Examples 100 to 102, and includes that the uplink resource is a resource for PUCCH transmission.

[0298] Example 104 is a method for wireless communication at a UE, comprising: communicating with a base station; and performing CSI measurement and / or AI-based CSI reporting based on at least one first CSI processing capability for AI-based CSI reporting.

[0299] Example 105 can be combined with Example 104 and include the UE receiving at least one reference signal from the base station and performing the CSI measurement based on the at least one reference signal.

[0300] Example 106 can be combined with any one of Examples 104 to 105, and includes that performing the AI-based CSI report further includes: obtaining an AI-based CSI report from the CSI measurement and the AI ​​or ML model; and transmitting the AI-based CSI report to the base station, wherein the AI-based CSI report includes periodic CSI, semi-persistent CSI, non-periodic CSI, any delay category, codebook type and / or beam report.

[0301] Example 107 can be combined with any one of Examples 104 to 106, and include that the at least one reference signal includes a CSI-RS and / or an SSB.

[0302] Example 108 can be combined with any one of Examples 104 to 107, and includes the UE transmitting the at least one first CSI processing capability to the base station.

[0303] Example 109 may be combined with any one of Examples 104 to 108, and include that the at least one first CSI processing capability indicates that the UE supports AI-based CSI processing.

[0304] Example 110 can be combined with any of Examples 104 to 109, and includes the UE communicating with the base station on one or more frequency bands, and the at least one first CSI processing capability includes a capability to indicate a (maximum) number of AI-based CSI reports, for which the UE is capable of measuring and processing reference signals simultaneously for each frequency band or each component carrier (CC) in the frequency band.

[0305] Example 111 can be combined with any one of Examples 104 to 110, and includes the UE communicating with the base station on one or more frequency bands, and the at least one first CSI processing capability includes a capability to indicate a (maximum) number of AI-based CSI reports, for which the UE is capable of measuring and processing reference signals simultaneously for all frequency bands or all CCs in all frequency bands.

[0306] Example 112 can be combined with any one of Examples 104 to 111, and includes the UE having multiple CSI processing units (CPUs), some or all of which are used by the UE to perform the CSI measurement and / or AI-based CSI reporting, and determining the CPU occupancy of each CPU when performing the CSI measurement and / or AI-based CSI reporting.

[0307] Example 113 can be combined with any one of Examples 104 to 112, and includes the UE reducing the priority of at least one CSI measurement and / or at least one AI-based CSI report based on the at least one first CSI capability and CPU occupancy, and transmitting outdated CSI of the at least one AI-based CSI report.

[0308] Example 114 may be combined with any one of Examples 104 to 113, and include that the at least one first CSI processing capability indicates that the UE supports AI-based CSI processing.

[0309] Example 115 may be combined with any one of Examples 104 to 114, and include that the at least one first CSI processing capability includes a capability to indicate a minimum processing delay for AI-based CSI reporting.

[0310] Example 116 may be combined with any one of Examples 104 to 115, and include the at least one first CSI processing capability including a capability indicating a minimum processing delay of AI-based CSI reporting for FR1.

[0311] Example 117 may be combined with any one of Examples 104 to 116, and include the at least one first CSI processing capability including a capability indicating a minimum processing delay of AI-based CSI reporting for FR2.

[0312] Example 118 can be combined with any of Examples 104 to 117, and include that the at least one first CSI processing capability includes the capability to indicate a minimum processing delay for AI-based CSI reports for one or more frequency bands (each of them), one or more frequency band combinations (each of them), each frequency band in a frequency band combination, or one or more frequency ranges (each of them).

[0313] Example 119 may be combined with any one of Examples 104 to 118, and include the at least one first CSI processing comprising a capability to indicate whether the AI-based CSI processing shares a CPU with a non-AI-based processing.

[0314] Example 120 can be combined with any of Examples 104 to 119 and include the at least one first CSI processing including the ability to indicate a number of CPUs (e.g., Type 2 CPUs) shared by AI-based CSI reporting and non-AI-based reporting.

[0315] Example 121 can be combined with any of Examples 104 to 120, and include the at least one first CSI processing including the capability to indicate the number of CPUs (e.g., Type 2 CPUs) shared by AI-based CSI reporting and non-AI-based reporting for one or more frequency bands (each of), one or more frequency band combinations (each of), each frequency band in a frequency band combination, or one or more frequency ranges (each of).

[0316] Example 122 can be combined with any one of Examples 104 to 121, and includes the UE transmitting the on / off duration of the Type 2 CPU to the base station.

[0317] Example 123 can be combined with any one of Examples 104 to 122, and includes the UE transmitting a periodic or semi-persistent AI-based CSI report to the base station (excluding the initial semi-persistent AI-based CSI report on the PUSCH after the PDCCH triggers a report), and the AI-based CSI report occupies the Type 2 CPU from the first symbol of the earliest resource of each CSI-RS / CSI-IM / SSB resource used for channel or interference measurement, no later than the corresponding latest CSI-RS / CSI-IM / SSB timing of the corresponding CSI reference resource, until the last symbol of the configured PUSCH / PUCCH carrying the report.

[0318] Example 124 can be combined with any one of Examples 104 to 123, and includes the UE transmitting a non-periodic AI-based CSI report to the base station, and the AI-based CSI report occupies the Type 2 CPU from the first symbol after the PDCCH triggers the CSI report until the last symbol of the scheduled PUSCH carrying the report.

[0319] Example 125 can be combined with any one of Examples 104 to 124, and includes the UE transmitting an initial semi-persistent AI-based CSI report to the base station on the PUSCH after the PDCCH is triggered, and the initial semi-persistent AI-based CSI report occupies the Type 2 CPU from the first symbol after the PDCCH to the last symbol of the scheduled PUSCH carrying the report.

[0320] Example 126 can be combined with any one of Examples 104 to 125, and includes the UE determining a Type 2 CPU occupied by Y symbols after X symbols after the first / last symbol of the earliest / latest CMR or IMR and before the first / last symbol of the PUSCH / PUCCH for AI-based CSI reporting.

[0321] Example 127 can be combined with any one of Examples 104 to 126, and include X and Y being predefined or reported by UE capabilities.

[0322] Example 128 can be combined with any of Examples 104 to 127, and includes each of the AI-based CSI reports employing both Type 1 and Type 2 CPUs.

[0323] Example 129 can be combined with any one of Examples 104 to 128, and includes the UE transmitting a periodic or semi-persistent CSI report to the base station (excluding an initial semi-persistent CSI report on the PUSCH after a PDCCH triggers a report), wherein the CSI report occupies Type 1 and Type 2 CPUs from the first symbol of the earliest resource of each CSI-RS / CSI-IM / SSB resource used for channel or interference measurement, no later than the corresponding latest CSI-RS / CSI-IM / SSB timing of the corresponding CSI reference resource, until the last symbol of the configured PUSCH / PUCCH carrying the report.

[0324] Example 130 can be combined with any one of Examples 104 to 129, and includes the UE transmitting a non-periodic AI-based CSI report to the base station, wherein the AI-based CSI report occupies Type 1 and Type 2 CPUs from the first symbol after the PDCCH triggers the CSI report until the last symbol of the scheduled PUSCH carrying the report.

[0325] Example 131 can be combined with any one of Examples 104 to 130, and includes the UE transmitting an initial semi-persistent AI-based CSI report to the base station on the PUSCH after the PDCCH is triggered, and the AI-based CSI report occupies Type 1 and Type 2 CPUs from the first symbol after the PDCCH to the last symbol of the scheduled PUSCH carrying the report.

[0326] Example 132 may be combined with any one of Examples 104 to 131, and includes that the Type 1 CPU is occupied, but excludes the duration that the Type 2 CPU is occupied.

[0327] Example 133 can be combined with any one of Examples 104 to 132, and includes the UE reporting an N with a higher priority.CPU,2 The effective CSI reported by AI-based CSI, where N CPU,2 It is the maximum number of Type2 CPUs.

[0328] Example 134 can be combined with any one of Examples 104 to 133, and includes the UE reporting to the base station a min{N CPU,1 -n CPU,1 ,N CPU,2} effective CSIs reported by AI-based CSIs, where N CPU,1 is the maximum number of Type1 CPUs, n CPU,1 is the number of CPUs used by other higher priority non-AI based CSI reports, and N CPU,2 It is the maximum number of Type2 CPUs.

[0329] Example 135 can be combined with any one of Examples 104 to 134, and includes the UE reporting outdated CSI of other triggered AI-based CSI reports to the base station.

[0330] Example 136 can be combined with any of Examples 104 to 135, and include that the capability includes parameters Z and Z' of the AI-based CSI report, wherein Z indicates the duration between the DCI and the first uplink symbol used to carry the corresponding AI-based CSI report including the timing advance effect, and Z' indicates the duration between the DCI and the first uplink symbol used to carry the nth AI-based CSI report including the timing advance effect.

[0331] Example 137 can be combined with any one of Examples 104 to 136, and includes the UE determining parameters Z and Z' based on whether the UE supports Type 2 CPU for AI-based CSI reporting.

[0332] Example 138 can be combined with any one of Examples 104 to 137, and includes the UE including parameters Z and Z' for each supported neural network type.

[0333] Example 139 may be combined with any one of Examples 104 to 138, and include parameters Z and Z' of an AI-based CSI report being predefined.

[0334] Example 140 can be combined with any one of Examples 104 to 139, and includes the capability indication parameters uNZ and uNZ', and the UE reports N AI-based CSI reports for AI-based CSI reporting to the base station based on the defined parameters uNZ and uNZ', where u is predefined, configured by the base station in higher layer signaling to the UE, or reported by the UE in a capability sent to the base station.

[0335] Example 141 can be combined with any one of Examples 104 to 140, and includes the UE receiving DCI from the base station, and if the scheduling offset in the DCI does not satisfy the minimum processing delay indicated in the parameter Z or Z', the UE ignores the DCI without transmitting HARQ-ACK or data on the PUSCH scheduled by the DCI.

[0336] Example 142 can be combined with any one of Examples 104 to 141, and includes the UE receiving DCI from the base station, and if the scheduling offset does not satisfy the minimum processing delay indicated in the parameter Z or Z', the UE reports outdated CSI of all triggered CSI reports to the base station.

[0337] Example 143 can be combined with any one of Examples 104 to 142, and includes if the scheduling offset does not satisfy the minimum processing delay indicated in the parameter Z or Z', the UE reports to the base station the outdated CSI of the triggered CSI report that does not satisfy the minimum processing delay indicated in the parameter Z or Z'.

[0338] Example 144 can be combined with any one of Examples 104 to 143, and further includes performing CSI measurement and / or artificial intelligence based non-AI CSI reporting based on at least one second CSI processing capability for non-AI based CSI reporting.

[0339] Example 145 can be combined with any one of Examples 104 to 144, and includes the UE receiving at least one reference signal from the base station and performing the CSI measurement based on the at least one reference signal.

[0340] Example 146 can be combined with any one of Examples 104 to 145, and further includes obtaining a non-AI based CSI report from the CSI measurement; and transmitting the non-AI based CSI report to the base station, wherein the non-AI based CSI report includes periodic CSI, semi-persistent CSI, non-periodic CSI, any delay category, codebook type and / or beam report.

[0341] Example 147 can be combined with any of Examples 104 to 146, and include that the at least one reference signal includes a CSI-RS and / or an SSB.

[0342] Example 148 can be combined with any one of Examples 104 to 147, and includes the UE transmitting the at least one second CSI processing capability to the base station.

[0343] Example 149 may be combined with any one of Examples 104 to 148, and include that the at least one second CSI processing capability indicates that the UE supports non-AI based CSI processing.

[0344] Example 150 can be combined with any one of Examples 104 to 149, and includes the UE communicating with the base station on one or more frequency bands, and the at least one second CSI processing capability includes the ability to indicate a (maximum) number of non-AI based CSI reports, for which the UE is capable of simultaneously measuring and processing reference signals for each of the frequency bands.

[0345] Example 151 can be combined with any one of Examples 104 to 150, and includes the UE communicating with the base station on one or more frequency bands, and the at least one second CSI processing capability includes the ability to indicate a (maximum) number of non-AI based CSI reports, for which the UE is capable of measuring and processing reference signals for all frequency bands simultaneously.

[0346] Example 152 can be combined with any one of Examples 104 to 151, and includes the UE determining CPU occupancy when performing the CSI measurement and / or non-AI based CSI reporting.

[0347] Example 153 may be combined with any one of Examples 104 to 152, and include that the at least one second CSI processing capability includes a capability to indicate a minimum processing delay for non-AI based CSI reporting.

[0348] Example 154 can be combined with any one of Examples 104 to 153, and includes the UE using a portion of the CPU to perform the CSI measurement and / or AI-based CSI reporting, and determining the CPU occupancy of each portion of the CPU when performing the CSI measurement and / or non-AI-based CSI reporting, and including AI-based CSI reporting and non-AI-based CSI reporting sharing the CPU.

[0349] Example 155 can be combined with any one of Examples 104 to 154, and include: the UE has multiple CPUs for non-AI based CSI reporting, some or all of which are used by the UE to perform the CSI measurement and / or non-AI based CSI reporting, and the CPU occupancy of each of the CPUs is determined when performing the CSI measurement and / or non-AI based CSI reporting, and includes the UE having a first group of CPUs dedicated to AI-based CSI reporting and a second group of CPUs dedicated to non-AI based CSI reporting.

[0350] Example 156 can be combined with any one of Examples 104 to 155, and includes the UE reducing the priority of at least one CSI measurement and / or at least one non-AI based CSI report based on the at least one first CSI capability, the at least one second CSI capability and / or CPU occupancy, and transmitting the outdated CSI of the at least one non-AI based CSI report, and includes the UE determining that the AI-based CSI report has a higher priority than the non-AI based CSI report, and includes when the UE does not have sufficient CPU to perform CSI measurements and / or reports of the AI-based CSI report and the non-AI based CSI report, the UE transmits to the base station the outdated CSI of the non-AI based CSI report to which the UE is not allocated a CPU.

[0351] Example 157 can be combined with any one of Examples 104 to 156, and includes the UE reducing the priority of at least one CSI measurement and / or at least one AI-based CSI report based on the at least one first CSI capability, the at least one second CSI capability and / or CPU occupancy, and transmitting outdated CSI of the at least one AI-based CSI report, and includes the UE determining that the AI-based CSI report has a lower priority than the non-AI-based CSI report, and includes when the UE does not have sufficient CPU to perform CSI measurements and / or reports of the AI-based CSI report and the non-AI-based CSI report, the UE transmits to the base station outdated CSI of the AI-based CSI report to which the UE is not allocated a CPU.

[0352] Example 158 is a method for wireless communication at a base station, comprising: communicating with a UE; and receiving at least one first CSI processing capability for AI-based CSI reporting from the UE, a core network, or another base station; and based on the at least one first CSI processing capability, configuring the UE to perform CSI measurement and / or AI-based CSI reporting.

[0353] Example 159 can be combined with Example 158 and includes the base station transmitting at least one reference signal to the UE and receiving the AI-based CSI report associated with the at least one reference signal from the UE, wherein the AI-based CSI report includes periodic CSI, semi-persistent CSI, non-periodic CSI, any delay category, codebook type and / or beam report.

[0354] Example 160 can be combined with any of Examples 158 to 159 and include that the at least one reference signal includes a CSI-RS and / or an SSB.

[0355] Example 161 can be combined with any one of Examples 158 to 160, and includes the base station communicating with the UE on one or more frequency bands, and the at least one first CSI processing capability includes the capability to indicate a (maximum) number of AI-based CSI reports, for which the UE is capable of simultaneously measuring and processing reference signals for each of the frequency bands.

[0356] Example 162 can be combined with any one of Examples 158 to 161, and includes the base station communicating with the UE on one or more frequency bands, and the at least one first CSI processing capability includes a capability to indicate a (maximum) number of AI-based CSI reports, for which the UE is capable of measuring and processing reference signals for all of the frequency bands simultaneously.

[0357] Example 163 can be combined with any one of Examples 158 to 162, and includes the base station configuring the UE to perform CSI measurement and / or non-AI based CSI reporting.

[0358] Example 164 can be combined with any one of Examples 158 to 163, and includes the base station receiving at least one second CSI processing capability from the UE, the CN, or another base station, and configuring the UE to perform the CSI measurement and / or non-AI-based CSI reporting based on the at least one second CSI processing capability.

[0359] Example 165 can be combined with any one of Examples 158 to 164, and include that performing the non-AI based CSI report further includes: receiving a non-AI based CSI report from the UE, wherein the non-AI based CSI report includes periodic CSI, semi-persistent CSI, non-periodic CSI, any delay category, codebook type and / or beam report.

[0360] Example 166 may be combined with any one of Examples 158 to 165, and include the at least one second CSI processing capability indicating that the UE supports non-AI based CSI processing.

[0361] Example 167 can be combined with any one of Examples 158 to 166, and includes the base station communicating with the UE on one or more frequency bands, and the at least one second CSI processing capability includes the ability to indicate a (maximum) number of non-AI based CSI reports, for which the UE is able to simultaneously measure and process reference signals for each of the frequency bands.

[0362] Example 168 can be combined with any one of Examples 158 to 167, and includes the base station communicating with the UE on one or more frequency bands, and the at least one second CSI processing capability includes the ability to indicate a (maximum) number of non-AI based CSI reports, for which the UE is capable of measuring and processing reference signals for all of the frequency bands simultaneously.

[0363] Example 169 may be combined with any one of Examples 158 to 168, and include that the at least one second CSI processing capability includes a capability to indicate a minimum processing delay for non-AI based CSI reporting.

[0364] Example 170 may be combined with any one of Examples 158 to 169, and include that the at least one second CSI processing capability does not include a capability to indicate a minimum processing delay for non-AI based CSI reporting.

[0365] Example 171 is a device for wireless communication implementing a method as described in any of Examples 1 to 170.

[0366] Example 172 is an apparatus for wireless communication including components for implementing a method as described in any of Examples 1 to 170.

[0367] Example 173 is a non-transitory computer-readable medium storing computer executable code, which, when executed by at least one processor, causes the at least one processor to implement a method as described in any one of Examples 1 to 170.

Claims

1. A method for wireless communication at a user equipment (UE), comprising: receiving a channel state information-reference signal (CSI-RS) from a network entity; selecting a wideband precoder for compressing a channel state information (CSI) report based on the measurement of the CSI-RS, wherein the selection of the wideband precoder is based on an entire bandwidth for receiving the CSI-RS; compressing the CSI report using one or more subband eigenvectors of one or more precoded subband channels by the selected wideband precoder; as well as A first precoding matrix indicator (PMI) indicating the wideband precoder and a second PMI indicating the one or more subband eigenvectors are sent to the network entity.

2. The method of claim 1, wherein the CSI report is compressed using a machine learning (ML) model.

3. The method of any one of claims 1 to 2, further comprising receiving from the network entity a radio resource control (RRC) configuration for at least one of: the ML model for compressing the CSI report, or The codebook and number of beams used for the wideband precoder.

4. The method of any one of claims 1 to 2, wherein the ML model is at least one of a predetermined ML model or based on a predefined protocol.

5. The method according to any one of claims 1 to 4, wherein sending the first PMI and the second PMI further comprises: transmitting to the network entity a short physical uplink control channel (PUCCH) format that includes the first PMI and the second PMI in the same CSI part of the short PUCCH format, or A long PUCCH format is transmitted to the network entity, which includes at least the second PMI in a CSI part 2 of the long PUCCH format.

6. The method of any one of claims 1 to 4, wherein sending the first PMI and the second PMI further comprises: A physical uplink shared channel (PUSCH) transmission is transmitted to the network entity, which includes at least the second PMI in a CSI part 2 of the PUSCH transmission.

7. A method of wireless communication at a user equipment (UE), comprising: receiving one or more channel state information-reference signals (CSI-RS) from a network entity to transmit one or more CSI reports to the network entity, at least one CSI report of the one or more CSI reports being based on a machine learning (ML) model; when a total payload size of the one or more CSI reports exceeds a threshold, omitting at least a portion of the at least one CSI report based on the ML model; as well as The one or more CSI reports are transmitted to the network entity, wherein the at least a portion of the at least one CSI report is omitted from the transmission.

8. The method of claim 7, wherein omitting the at least a portion of the at least one CSI report is based on the at least a portion having at least one of: a first priority lower than one or more non-omitted CSI reports of the one or more CSI reports, or A second priority lower than the one or more non-omitted portions of the one or more CSI reports.

9. The method of any one of claims 7 to 8, wherein omitting the at least a portion of the at least one CSI report further comprises: omitting a first portion of the at least one CSI report based on the ML model; as well as A second portion of at least one other CSI report not based on the ML model is omitted, the at least one other CSI report included in the transmission of the one or more CSI reports to the network entity.

10. The method of any one of claims 7 to 8, wherein omitting the at least a portion of the at least one CSI report further comprises: discarding the at least one CSI report based on the ML model; as well as The discarded at least one CSI report is replaced with at least one substitute CSI report that is not based on the ML model, the at least one substitute CSI report being included in the transmission of the one or more CSI reports to the network entity.

11. The method of any one of claims 1 to 8 or 10, further comprising omitting a second portion of at least one other CSI report that is not based on the ML model, the at least one other CSI report being included in the transmission of the one or more CSI reports to the network entity.

12. The method of any one of claims 1 to 11, wherein the transmission of the one or more CSI reports to the network entity includes an indication of the at least one portion omitted from the at least one CSI report based on the ML model.

13. A method for wireless communication at a user equipment (UE), comprising: receiving a configuration for a channel state information (CSI) report processing procedure from a network entity, wherein the configuration is for at least one of: (a) processing a plurality of CSI reports using a machine learning (ML) model or (b) processing a first set of CSI reports using the ML model without processing a second set of CSI reports; receiving a plurality of channel state information-reference signals (CSI-RS) from the network entity, wherein measurement of the plurality of CSI-RS is based on UE capabilities for the CSI reporting process; as well as A CSI report is transmitted to the network entity based on the UE capabilities for the CSI reporting process.

14. The method of claim 13, further comprising transmitting a UE capability report to the network entity, the UE capability report indicating the UE capabilities for the CSI reporting process.

15. The method of any one of claims 13 to 14, wherein the UE capability used for the CSI report processing procedure corresponds to at least one of a first UE capability for occupying a type 2 CSI processing unit (CPU) or a second UE capability for minimum CSI report processing delay.

16. The method of any one of claims 13 to 15, wherein transmitting the CSI report to the network entity based on the UE capabilities for the CSI reporting process further comprises: At least one of: the plurality of CSI reports associated with the ML model, or a first CSI report associated with the ML model and a second CSI report not associated with the ML model is transmitted to the network entity.

17. The method of any one of claims 13 to 16, further comprising receiving a trigger indication to measure the multiple CSI-RS for at least one of: the multiple CSI reports associated with the ML model, or the first CSI report associated with the ML model and the second CSI report not associated with the ML model.

18. The method of any one of claims 13 to 17, wherein the measuring of the plurality of CSI-RSs occurs based on a priority of each of the at least one of: the plurality of CSI reports, or the first CSI report and the second CSI report.

19. An apparatus for wireless communication, comprising a memory and at least one processor, the at least one processor being coupled to the memory and configured to implement the method of any one of claims 1 to 18.

Images