Scheme selection for feeding back channel state information

Abstract

An apparatus may be configured to receive an instruction indicating one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency; and reporting CSI feedback to the network node using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme. Another apparatus may be configured to select one of an ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency; transmitting, to the UE, an instruction indicating a selected one of an ML CSI feedback scheme or a non-ML CSI feedback scheme; and receiving, from the UE, CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. non-provisional application serial number 18 / 190,052, filed on March 24, 2023, entitled “SCHEME SELECTION FOR FEEDINGBACK CHANNEL STATE INFORMATION,” the disclosure of which is expressly incorporated herein by reference in its entirety. Technical Field

[0002] The present disclosure relates generally to communication systems, and more particularly to selection of channel state information feedback schemes with or without the assistance of machine learning models. Background Art

[0003] Wireless communication systems are widely deployed to provide a variety of telecommunication services, such as telephony, video, data, messaging, and broadcasts. Typical wireless communication systems employ multiple-access technologies capable of supporting communication with multiple users by sharing available system resources. Examples of such multiple-access technologies include code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal frequency division multiple access (OFDMA), single-carrier frequency division multiple access (SC-FDMA), and time division synchronous code division multiple access (TD-SCDMA).

[0004] These multiple-access technologies have been adopted in various telecommunication standards to provide a common protocol that enables different wireless devices to communicate at the city, national, regional, and even global levels. One example telecommunication standard is 5G New Radio (NR). 5G NR is part of the ongoing mobile broadband evolution being promulgated by the 3rd Generation Partnership Project (3GPP) to address new requirements related to latency, reliability, security, scalability (e.g., for the Internet of Things (IoT)), and other requirements. 5G NR includes services associated with enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and ultra-reliable low-latency communications (URLLC). Some aspects of 5G NR may be based on the 4G Long Term Evolution (LTE) standard. Further improvements to 5G NR technology are needed. Furthermore, these improvements may also be applicable to other multiple-access technologies and telecommunication standards that employ them. Summary of the Invention

[0005] The following presents a simplified summary of one or more aspects in order to provide a basic understanding of such aspects. This summary is not an extensive overview of all contemplated aspects, and is intended to neither identify key or critical elements of all aspects nor delineate 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 the more detailed description that is presented later.

[0006] In an aspect of the disclosure, a method, a computer-readable medium, and an apparatus are provided. The apparatus can be a user equipment (UE) or a component thereof. The apparatus can be configured to receive, from a network node, an instruction indicating one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback; and report, to the network node, the CSI feedback using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0007] In another aspect of the disclosure, another method, another computer-readable medium, and another apparatus are provided. The other apparatus can be a network node or a component thereof. The other apparatus can be configured to select one of a ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to be used by a UE for CSI feedback; transmit, to the UE, an instruction indicating the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme; and receive, from the UE, the CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0008] To the accomplishment of the foregoing and related aspects, one or more aspects comprise the features as broadly indicated herein above and specifically described below. The following description and drawings are illustrative of the one or more aspects and are not to be construed as limiting thereof. Numerous options, alternatives, and variations will be apparent to the skilled artisan. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 is a diagram illustrating an example of a wireless communications system and an access network.

[0010] Figure 2 is a diagram illustrating an example disaggregated base station architecture.

[0011] Figure 3A is a diagram illustrating an example of a first frame in accordance with various aspects of the present disclosure.

[0012] Figure 3Bis a diagram illustrating an example of downlink channels within a subframe in accordance with various aspects of the present disclosure.

[0013] Figure 3C is a diagram illustrating an example of a second frame in accordance with various aspects of the present disclosure.

[0014] Figure 3D is a diagram illustrating an example of uplink channels within a subframe in accordance with various aspects of the present disclosure.

[0015] Figure 4 is a diagram illustrating an example of a base station and user equipment (UE) in an access network.

[0016] Figure 5 is a block diagram illustrating a machine learning (ML) assisted scheme for CSI feedback.

[0017] Figure 6 is a call flow diagram illustrating selecting a ML assisted scheme or a ML unassisted scheme for reporting CSI feedback by a UE to a network node.

[0018] Figure 7 is a flow diagram illustrating an example of a method of wireless communication at a UE.

[0019] Figure 8 is a flow diagram illustrating an example of a method of wireless communication at a network node.

[0020] Figure 9 is a diagram illustrating an example of a hardware implementation for an example apparatus.

[0021] Figure 10 is a diagram illustrating another example of a hardware implementation for another example apparatus. DETAILED DESCRIPTION

[0022] The detailed description set forth below, in connection with the appended drawings and embodiments described herinin, is intended as a description of various configurations and is not intended to represent the only configurations in which the concepts described herein can be practiced. The detailed description includes specific details for the purpose of providing a thorough understanding of various concepts. However, it will be apparent to those skilled in the art that the concepts described herein can be practiced without some or all of the specific details. In some instances, well known structures, components, and so on have not been shown or described in excruciating detail in order to avoid obscuring the concepts.

[0023] Several aspects of telecommunication systems will now be presented with reference to various apparatus and methods. These apparatus and methods will be described in the following detailed description and illustrated in the accompanying drawings by various blocks, components, circuits, processes, algorithms, etc. (collectively referred to as "elements"). These elements can be implemented with electronic hardware, computer software, or any combination thereof. Whether such elements are implemented as hardware or software depends on the particular application and design constraints imposed on the overall system.

[0024] By way of example, an element, or any portion of an element, or any combination of elements can be implemented as a "processing system" that includes 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 a chip (SoC), baseband processors, field programmable gate arrays (FPGAs), programmable logic devices (PLDs), state machines, gated logic, discrete hardware circuits, and other suitable hardware configured to perform the various functionality described throughout this disclosure. One or more processors in the processing system can execute software. Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software components, applications, software applications, software packages, routines, subroutines, objects, executables, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0025] Accordingly, in one or more example embodiments, the functions described can be implemented in hardware, software, or any combination thereof. If implemented in software, the functions can be stored on or encoded as one or more instructions or computer-executable code on a computer-readable medium. Computer-readable media includes computer storage media. Storage media can be any available media that can be accessed by a computer. By way of example, and not limitation, such computer-readable media can comprise RAM, ROM, EEPROM, optical disk storage, magnetic disk storage, other magnetic storage devices, combinations of the aforementioned types of computer-readable media, or any other medium that is capable of being used to store computer executable code in the form of instructions or data structures that can be accessed by a computer.

[0026] Figure 1is a diagram illustrating an example of a wireless communications system and an access network 100. The wireless communications system (also referred to as a wireless wide area network (WWAN)) includes base stations 102, user equipment (UE) 104, an Evolved Packet Core (EPC) 160, and another core network 190 (e.g., a 5G Core (5GC)). The base stations 102 can include macro cells, such as high power cellular base stations, and / or small cells, such as low power cellular base stations including femtocells, pico cells, and micro cells.

[0027] The base stations 102 configured for 4G Long Term Evolution (LTE) (collectively referred to as the Evolved Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (E-UTRAN)) can interface with the EPC 160 through first backhaul links 132 (e.g., S1 interface). The base stations 102 configured for 5G New Radio (NR) (collectively referred to as the Next Generation Radio Access Network (RAN) (NG-RAN)) can interface with the core network 190 through second backhaul links 134. In addition to other functions, the base stations 102 can perform one or more of: transfer of user data, radio channel

[0028] In some aspects, the base stations 102 can communicate directly or indirectly (e.g., through the EPC 160 or core network 190) with each other via third backhaul links 136 (e.g., X2 interface). The first backhaul links 132, the second backhaul links 134, and the third backhaul links 136 can be wired, wireless, or some combination thereof. At least some of the base stations 102 can be configured for integrated access and backhaul (IAB). Accordingly, such base stations can communicate wirelessly with other base stations (which can also be configured for IAB).

[0029] At least some of the base stations 102 configured for IAB can have a split architecture including a plurality of units, some or all of which can be collocated or distributed and can communicate with one another. For example, Figure 2 The following illustrates an example disaggregated base station 200 architecture including at least one of a central unit (CU) 210, a distributed unit (DU) 230, a radio unit (RU) 240, a remote radio head (RRH), a remote unit, and / or another similar unit configured to implement one or more layers of a radio protocol stack.

[0030] Base station 102 can communicate wirelessly with UEs 104. Examples of UEs 104 include cellular phones, smartphones, Session Initiation Protocol (SIP) phones, laptops, personal digital assistants (PDAs), satellite radios, global positioning systems, multimedia devices, video devices, digital audio players (e.g., MP3 players), cameras, game consoles, tablet devices, smart devices, wearable devices, vehicles, electric meters, gas pumps, large or small kitchen appliances, healthcare devices, implants, sensors / actuators, displays, or any other similarly functional device. Some of UEs 104 may be referred to as IoT devices (e.g., parking meters, gas pumps, toasters, vehicles, heart rate monitors, etc.).

[0031] UE 104 may also be referred to as a station, 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 handset, a user agent, a mobile client, a client, or some other suitable terminology.

[0032] Each base station 102 may provide communication coverage for a corresponding geographic coverage area 110 (which may also be referred to as a "cell"). Potentially, two or more geographic coverage areas 110 may at least partially overlap with each other, or one of the geographic coverage areas 110 may encompass another of the geographic coverage areas. For example, a small cell 102' may have a coverage area 110' that overlaps with the coverage area 110 of one or more macro base stations 102. A network that includes both small cells and macro cells may be referred to as a heterogeneous network. A heterogeneous network may also include a home evolved Node B (eNB) (HeNB), which may provide service to a restricted group known as a closed subscriber group (CSG).

[0033] The communication links 120 between the base stations 102 and the UEs 104 can include uplink (also referred to as reverse link) transmissions from a UE 104 to a base station 102 and / or downlink (also referred to as forward link) transmissions from a base station 102 to a UE 104. The communication links 120 can use multiple-input and multiple-output (MIMO) antenna technology, including spatial multiplexing, beamforming, and / or transmit diversity. The wireless links or radio links can be time-discrete, frequency-discrete or code-discrete with respect to data transmission. A link can be established from a base station 102 to a UE 104 or from a UE 104 to a base station 102. A link from a base station 102 to a UE 104 can be made up of one or more component carriers in one or more frequency bands. Each link can be established according to a schedule of a carrier aggregation (CA) that is configured for the UE 104. The schedule can indicate a primary component carrier and a set of one or more secondary component carriers. The primary component carrier can be on a primary cell (PCell) and the set of one or more secondary component carriers can be on a set of secondary cells (SCells). The base stations 102 and / or UEs 104 can use beamforming to transmit or receive data.

[0034] A CC can include a primary CC and one or more secondary CCs. The primary CC can be referred to as a primary cell (PCell) and each secondary CC can be referred to as a secondary cell (SCell). The PCell can also be referred to as a “serving cell” when the UE is known to both the base station at an access network level and at least one core network entity (e.g., an AMF and / or an MME) at a core network level, such as in a case when the UE is in a radio resource control (RRC) connected state. In some instances in which a UE is configured with carrier aggregation, the PCell and each of one or more SCells can be serving cells.

[0035] Certain UEs 104 can communicate with each other using device-to-device (D2D) communication link 158. The D2D communication link 158 can use the downlink / uplink WWAN spectrum. The D2D communication link 158 can 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 a physical sidelink control channel (PSCCH). D2D communication can be through a variety of wireless D2D communications systems, such as for example, WiMedia, Bluetooth, ZigBee, Wi-Fi based on the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard, LTE, or NR.

[0036] The wireless communications system can further include a Wi-Fi access point (AP) 150 in communication with Wi-Fi stations (STAs) 152 via communication links 154, e.g., in a 5 gigahertz (GHz) unlicensed spectrum. When communicating in an unlicensed spectrum, STAs 152 / AP 150 can perform a clear channel assessment (CCA) prior to communicating in order to determine whether the channel is available.

[0037] The small cells 102' can operate in a licensed and / or an unlicensed spectrum. When operating in an unlicensed spectrum, the small cells 102' can employ NR and use the same unlicensed spectrum as used by the Wi-Fi AP 150 (e.g., 5 GHz, etc.). The small cells 102' employing NR in an unlicensed spectrum can improve coverage and / or increase capacity of the access network.

[0038] The electromagnetic spectrum is often subdivided based on frequency / wavelength into various classes, bands, channels, etc. In 5G NR, two initial operating bands have been identified as frequency range designations FR1 (410 MHz - 7.125 GHz) and FR2 (24.25 GHz - 52.6 GHz). The frequencies between FR1 and FR2 are often referred to as mid-band frequencies. Despite a portion of FR1 being greater than 6 GHz, FR1 is often referred to (interchangeably) as a “Sub-6 GHz” band in various documents and articles. A similar nomenclature issue sometimes occurs with respect to FR2, which is often referred to (interchangeably) as a “millimeter wave” band in documents and articles, despite such being different from the extremely high frequency (EHF) band (30 GHz - 300 GHz) which is designated as a “millimeter wave” (or “mmWave” or simply “mmW”) band by the International Telecommunications Union (ITU). In some aspects, “mmW” or “near mmW” can additionally or alternatively refer to the 60 GHz frequency range, which can include multiple channels about 60 GHz. For example, a 60 GHz band can refer to a set of channels spanning from 57.24 GHz to 70.2 GHz.

[0039] In accordance with the above, the terms “Sub-6 GHz,” “6 GHz,” “7 GHz,” and the like, as used herein with respect to ranges, can broadly represent frequencies less than 6 GHz, frequencies less than 7 GHz, frequencies within FR1, and / or frequencies that can include the mid-band frequencies, unless otherwise specifically stated. Additionally, the term “millimeter wave,” and other similar terminology, as used herein with respect to ranges, can broadly represent frequencies that can include the mid-band frequencies, frequencies within FR2, and / or frequencies within the EHF band, unless otherwise specifically stated.

[0040] The base stations 102 can be implemented as macro base stations or small cell base stations and can operate according to 3GPP LTE, 5G NR, or other wireless communication protocols. The base stations 102 can also be multi- standard radio (MSR) base stations that operate according to two or more wireless communication protocols. The base stations 102 can be collectively administered by a core network 104, which can be an evolved packet core (EPC) or 5G core (5GC). The core network 104 can also serve as a gateway for the base stations 102 to interface with other networks, like the Internet. Further, the core network 104 can provide mobility management functions, bearer management functions, and / or authentication functions.

[0041] The base stations 102 can transmit beacon signals 106 to assist UEs 104 in synchronizing and / or accessing the network. The base stations 102 can also transmit synchronization signals (e.g., primary and secondary synchronization signals (PSS and SSS)) and / or channel state information (CSI) reference signals (RSs) for use by UEs 104 in determining downlink channel state feedback. The base stations 102 can transmit the beacon signals 106, synchronization signals, and / or CSI-RSs in the form of signals 108, 110, and 112, respectively. The base stations 102 can also transmit physical broadcast channels (PBCH) and / or physical downlink shared channels (PDSCHs) in the form of signals 114 and 116, respectively.

[0042] In various aspects, one or more of the base stations 102 / 180 can include and / or be referred to as a gNB, a NodeB, an eNB, an access point, a transceiver base 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), or some other suitable terminology.

[0043] In some aspects, one or more of the base stations 102 / 180 can be connected to the EPC 160 and can provide a respective access point to the EPC 160 for one or more of the UEs 104. The EPC 160 can include a Mobility Management Entity (MME) 162, other MMEs 164, a Serving Gateway 166, an MBMS Gateway 168, a Broadcast Multicast Service Center (BM-SC) 170, and a Packet Data Network (PDN) Gateway 172. The MME 162 can be in communication with a Home Subscriber Server (HSS) 174. The MME 162 is the control node that processes the signaling between the UEs 104 and the EPC 160. Generally, the MME 162 provides bearer and connection management. All user Internet Protocol (IP) packets are transferred through the Serving Gateway 166, which is connected to the PDN Gateway 172. The PDN Gateway 172 provides UE IP address allocation as well as other functions. The PDN Gateway 172 and the BM-SC 170 are connected to the IP Services 176. The IP Services 176 can include the Internet, an intranet, an IP Multimedia Subsystem (IMS), a Packet-Switched (PS) Streaming Service, and / or other IP services. The BM-SC 170 can provide functions for MBMS user service provisioning and delivery. The BM-SC 170 can serve as an entry point for content provider MBMS transmissions, can be used to authorize and initiate MBMS Bearer Services, and can be used to schedule MBMS transmissions. The MBMS Gateway 168 can be used to

[0044] In some other aspects, one or more of the base stations 102 / 180 can be connected to the core network 190 and can provide a respective access point to the core network 190 for one or more of the UEs 104. The core network 190 can include an Access and Mobility Management Function (AMF) 192, other AMFs 193, a Session Management Function (SMF) 194, and a User Plane Function (UPF) 195. The AMF 192 can be in communication with a Unified Data Management (UDM) 196. The AMF 192 is the control node that processes the signaling between the UEs 104 and the core network 190. Generally, the AMF 192 provides Quality of Service (QoS) flow and session management. All user IP packets are transferred through the UPF 195. The UPF 195 provides UE IP address allocation as well as other functions. The UPF 195 is connected to the IP Services 197. The IP Services 197 can include the Internet, an intranet, an IMS, a PS streaming service, and / or other IP services.

[0045] In certain aspects, the UE 104 can be configured to receive, from the base station 102 / 180, an instruction indicating a selection 198 of one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback. The UE 104 can use the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme to report CSI feedback to the base station 102 / 180.

[0046] Correspondingly, the base station 102 / 180 can be configured for selecting 198 one of a ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use by the UE 104 for CSI feedback. The base station 102 / 180 can transmit, to the UE 104, an instruction indicating the selection 198 of one of the ML CSI feedback scheme or the non-ML CSI feedback scheme. The base station 102 / 180 can receive, from the UE 104, CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0047] Although the present disclosure can focus on 5G NR, the concepts and various aspects described herein can be applicable to other similar areas, such as LTE, LTE-Advanced (LTE-A), Code Division Multiple Access (CDMA), Global System for Mobile Communications (GSM), and / or other wireless / radio access technologies.

[0048] Figure 2 An illustration diagram illustrating an example disaggregated base station 200 architecture is shown. Deployments of communication systems, such as 5G NR systems, can utilize various components or constituent parts arranged in a variety of ways. In a 5G NR system or network, a network node, network entity, mobility element of a network, RAN node, core network node, network element, or network equipment, such as a base station, or one or more elements (or one or more components) performing base station functionality, can be implemented in an aggregated or disaggregated architecture. For example, a base station (or network node) can be implemented as an aggregated base station (also referred to as a self-standing base station or monolithic base station) or a disaggregated base station.

[0049] A disaggregated base station can be configured to utilize a radio protocol stack that is physically or logically distributed among two or more units, such as one or more CUs, one or more DUs, or one or more RUs. In some aspects, a CU can be implemented within a RAN node, and one or more DUs can be co-located with the CU or, alternatively, can be geographically or virtually distributed in one or more other RAN nodes. The DUs can be implemented to communicate with one or more RUs. Each of the CUs, DUs, and RUs can also be implemented as virtual units, such as a virtual central unit (VCU), virtual distributed unit (VDU), or virtual radio unit (VRU).

[0050] Base station type operations or network designs can take into account the disaggregated nature of base station functionality. For example, a disaggregated base station can be utilized in an IAB network, an open radio access network (O-RAN (such as network configurations initiated by the O-RAN Alliance)), or a virtualized radio access network (vRAN, also referred to as a cloud radio access network (C-RAN)). Disaggregation can include distributing functionality across two or more units at various physical locations, as well as virtually distributing functionality of at least one unit, which can enable flexibility in network design. The various units of a disaggregated base station or disaggregated RAN architecture can be configured for wired or wireless communication with at least one other unit.

[0051] The disaggregated base station 200 architecture can include one or more CUs 210 that can communicate directly with a core network 220 via a backhaul link, or indirectly through one or more disaggregated base station units, such as a near real-time (near-RT) RAN intelligence controller (RIC) 225 via an E2 link, or a non-real-time (non-RT) RIC 215 associated with a service management and orchestration (SMO) framework 205, or both. The CUs 210 can communicate with one or more DUs 230 via respective fronthaul links, such as Fl interfaces. The DUs 230 can communicate with one or more RUs 240 via respective front-haul links. The RUs 240 can communicate with respective UEs 104 via one or more radio frequency (RF) access links. In some implementations, a UE 104 can be simultaneously served by multiple RUs 240.

[0052] Each of the units (i.e., CU 210, DU 230, RU 240, and near-RT RIC 225, non-RT RIC 215, and SMO framework 205) can include or be coupled to one or more interfaces configured to receive or transmit signals, data, or information (collectively, signals) via wired or wireless transmission media. Each of the units, or an associated processor or controller providing instructions to the communication interfaces of these units, can be configured to communicate with one or more of the other units via the transmission media. For example, the units can include wired interfaces configured to receive or transmit signals to one or more of the other units over a wired transmission medium. Additionally, the units can include wireless interfaces, which can include receivers, transmitters, or transceivers (such as RF transceivers) configured to receive or transmit signals to one or more of the other units over a wireless transmission medium, or both.

[0053] In some aspects, the CU 210 can host one or more higher layer control functions. Such control functions can include RRC, packet data convergence protocol (PDCP), service data adaptation protocol (SDAP), etc. Each control function can utilize an interface configured to communicate signals with other control functions hosted by the CU 210. The CU 210 can be configured to handle user plane functionality (i.e., central unit-user plane (CU-UP)), control plane functionality (i.e., central unit-control plane (CU-CP)), or a combination thereof. In some implementations, the CU 210 can be logically split into one or more CU-UP units and one or more CU-CP units. When implemented in an O-RAN configuration, the CU-UP units can communicate bi-directionally with the CU-CP units via an interface, such as an El interface. As needed, the CU 210 can be implemented to communicate with the DU 230 for network control and signaling.

[0054] The DU 230 can correspond to a logical unit that includes one or more base station functions for controlling operation of one or more RUs 240. In some aspects, the DU 230 can host one or more of a radio link control (RLC) layer, a medium access control (MAC) layer, and one or more high physical (PHY) layers (such as modules for forward error correction (FEC) encoding and decoding, scrambling, modulation and demodulation, etc.) in accordance with, at least in part, a functional split, such as those defined by the 3rd Generation Partnership Project (3GPP). In some aspects, the DU 230 can further host one or more low PHY layers. Each layer (or module) can be implemented with an interface configured to communicate signals with other layers (and modules) hosted by the DU 230 or with control functions hosted by the CU 210.

[0055] Lower layer functionality can be implemented by one or more RUs 240. In some deployments, the RUs 240 controlled by the DU 230 can correspond to logical nodes that host RF processing functions or low PHY layer functions (such as performing fast Fourier transform (FFT), inverse FFT (iFFT), digital beamforming, or physical random access channel (PRACH) extraction and filtering, etc.), or both, based at least in part on a functional split, such as a lower layer functional split. In such an architecture, the RUs 240 can be implemented to handle over-the-air (OTA) communications with one or more UEs 104. In some implementations, real-time and non-real-time aspects of control plane and user plane communications with the RUs 240 can be controlled by the corresponding DU 230. In some scenarios, this configuration can enable the DUs 230 and CUs 210 to be implemented in a cloud-based RAN architecture, such as a vRAN architecture.

[0056] The SMO framework 205 can be configured to support RAN deployment and orchestration of non-virtualized network elements and virtualized network elements. For non- virtualized network elements, the SMO framework 205 can be configured to support deployment of dedicated physical resources for RAN coverage requirements, which can be managed via an operations and maintenance interface, such as an Ol interface. For virtualized network elements, the SMO framework 205 can be configured to interact with a cloud computing platform, such as Open Cloud (O-Cloud) 290, to perform network element lifecycle management, such as to instantiate virtualized network elements, via a cloud computing platform interface, such as an 02 interface. Such virtualized network elements can include, but are not limited to, CUs 210, DUs 230, RUs 240, and near-RT RICs 225. In some implementations, the SMO framework 205 can communicate with hardware aspects of a 4G RAN, such as Open eNB (O-eNB) 211, via an Ol interface. Additionally, in some implementations, the SMO framework 205 can communicate directly with one or more RUs 240 via an Ol interface. The SMO framework 205 can also include a non-RT RIC 215 configured to support functionality of the SMO framework 205.

[0057] The non-RT RIC 215 can be configured to include logical functions that enable near real-time control and optimization of RAN elements and resources, artificial intelligence / machine learning (AI / ML) workflows including model training and updates, or policy-based direction of applications / features in the near-RT RIC 225. The non-RT RIC 215 can be coupled to, or in communication with, the near-RT RIC 225, such as via an Al interface. The near-RT RIC 225 can be configured to include logical functions that enable near real-time control and optimization of RAN elements and resources via data collection and actions through an interface, such as via an E2 interface, that connects one or more CUs 210, one or more DUs 230, or both, and an O-eNB with the near-RT RIC 225.

[0058] In some implementations, to generate AI / ML models to be deployed in near-RT RIC 225, non-RT RIC 215 can receive parameters or external rich information from an external server. Such information can be utilized by near-RT RIC 225 and can be received at SMO framework 205 or non-RT RIC 215 from non-network data sources or from network functions. In some examples, non-RT RIC 215 or near-RT RIC 225 can be configured to tune RAN behavior or performance. For example, non-RT RIC 215 can monitor long-term trends and patterns of performance and employ AI / ML models to perform corrective actions through SMO framework 205, such as via reconfiguration of 01, or via creation of RAN management policies, such as Al policies.

[0059] Figure 3A is a diagram illustrating an example of a first subframe 300 within a 5G NR frame structure. Figure 3B is a diagram illustrating an example of downlink channels within a 5G NR subframe 330. Figure 3C is a diagram illustrating an example of a second subframe 350 within a 5G NR frame structure. Figure 3D is a diagram illustrating an example of uplink channels within a 5G NR subframe 380. The 5G NR frame structure can be frequency division duplex (FDD), where for a particular set of subcarriers (carrier system bandwidth), the subframes within that set of subcarriers are dedicated to downlink or uplink; or time division duplex (TDD), where for a particular set of subcarriers (carrier system bandwidth), the subframes within that set of subcarriers are dedicated to both downlink and uplink. In the examples provided by Figure 3A and Figure 3C In the examples provided, the 5G NR frame structure is assumed to be TDD, with subframe 4 configured with slot format 28 (mostly downlink) and subframe 3 configured with slot format 34 (mostly uplink), where D is downlink, U is uplink, and F is flexibly used between downlink / uplink. While subframes 3, 4 are shown with slot formats 34, 28, respectively, any particular subframe can be configured with any of the various available slot formats 0-61. Slot formats 0, 1 are all downlink, all uplink, respectively. Other slot formats 2-61 include a mix of downlink, uplink, and flexible symbols. UEs are configured with a slot format (dynamically, by downlink control information (DCI), or semi-statically / statically, by RRC signaling) by a received slot format indicator (SFI). Note that the following description also applies to 5G NR frame structures that are TDD.

[0060] Other wireless communication technologies can have different frame structures and / or different channels. For example, a 10 millisecond (ms) frame can be divided into 10 equally sized subframes (1 ms). Each subframe can include one or more slots. A subframe can also include mini-slots, which can include 7, 4, or 2 symbols. Each slot can include 7 or 14 symbols, depending on the slot configuration. For slot configuration 0, each slot can include 14 symbols, and for slot configuration 1, each slot can include 7 symbols. Symbols on the downlink can be cyclic prefix (CP) orthogonal frequency division multiplexing (OFDM) (CP-OFDM) symbols. Symbols on the uplink can be CP-OFDM symbols (for high throughput scenarios) or discrete Fourier transform (DFT) spread OFDM (DFT-s-OFDM) symbols (also known as single carrier frequency division multiple access (SC-FDMA) symbols) (for power limited scenarios; limited to single stream transmission). The number of slots within a subframe is based on the slot configuration and the numerology. For slot configuration 0, different numerologies 0 to 4 allow for 1, 2, 4, 8, and 16 slots per subframe, respectively. For slot configuration 1, different numerologies 0 to 2 allow for 2, 4, and 8 slots per subframe, respectively. Thus, for slot configuration 0 and numerology, there are 14 symbols per slot and 2 µ slots per subframe. The subcarrier spacing and symbol length / duration are a function of the numerology. The subcarrier spacing can equal kHz, where is the numerology 0 to 4. Thus, the subcarrier spacing for numerology = 0 is 15 kHz, and the subcarrier spacing for numerology = 4 is 240 kHz. The symbol length / duration is inversely related to the subcarrier spacing. Figures 3A-3D An example is provided of slot configuration 0 with 14 symbols per slot and numerology = 2 with 4 slots per subframe. The slot duration is 0.25 ms, the subcarrier spacing is 60 kHz, and the symbol duration is approximately 16.67 microseconds (ps). Within a set of frames, there can be one or more different bandwidth parts (BWPs) that are frequency division multiplexed (see Figure 3B ). Each BWP can have one particular numerology.

[0061] A resource grid can be used to represent the frame structure. Each time slot includes a resource block (RB) (also referred to as a physical RB (PRB)) that extends 12 consecutive subcarriers. The resource grid is divided into multiple resource elements (REs). The number of bits carried by each RE depends on the modulation scheme.

[0062] As Figure 3AAs illustrated, some of the REs carry at least one pilot signal for the UE, such as a reference signal (RS). Broadly speaking, the RS may be used for beam training and management, tracking and positioning, channel estimation, and / or other such purposes. In some configurations, the RS may include at least one demodulation RS (DM-RS) for channel estimation at the UE (indicated as R for a particular configuration). x , where 100x is the port number, but other DM-RS configurations are possible) and / or at least one channel state information (CSI) RS (CSI-RS). In some other configurations, the RS may additionally or alternatively include at least one beam measurement (or management) RS (BRS), at least one beam refinement RS (BRRS) and / or at least one phase tracking RS (PT-RS).

[0063] Figure 3B Examples of various downlink channels within a subframe of a frame are illustrated. The physical downlink control channel (PDCCH) carries DCI within one or more control channel elements (CCEs), each CCE consisting of nine RE groups (REGs), and each REG consisting of four consecutive REs in an OFDM symbol. The PDCCH within one BWP may be referred to as a control resource set (CORESET). Additional BWPs may be located at higher and / or lower frequencies across the channel bandwidth. The primary synchronization signal (PSS) may be within symbol 2 of a particular subframe of a frame. UEs (such as Figure 1 The UE 104 may use the PSS to determine subframe / symbol timing and physical layer identification. The Secondary Synchronization Signal (SSS) may be within symbol 4 of a particular subframe of a frame. Figure 1 The UE 104) can use the SSS to determine the physical layer cell identity group number and radio frame timing. Based on the physical layer identity and the physical layer cell identity group number, the UE can determine the physical cell identifier (PCI). Based on the PCI, the UE can determine the position of the aforementioned DM-RS. The physical broadcast channel (PBCH) carrying the master information block (MIB) can be logically grouped with the PSS and SSS to form a synchronization signal (SS) / PBCH block (also referred to as an SS block (SSB)). The MIB provides the number of RBs in the system bandwidth and the system frame number (SFN). The physical downlink shared channel (PDSCH) carries user data, broadcast system information not sent through the PBCH (such as the system information block (SIB)), and paging messages.

[0064] like Figure 3CSome of the REs carry DM-RS (indicated as R for one particular configuration, but other DM-RS configurations are possible) for channel estimation at the base station. The UE can transmit DM-RS for the physical uplink control channel (PUCCH) and DM-RS for the physical uplink shared channel (PUSCH). The PUSCH DM-RS can be transmitted in the first one or two symbols of the PUSCH. The PUCCH DM-RS can be transmitted in different configurations depending on whether short or long PUCCHs are transmitted and on the particular PUCCH format used. The UE can transmit a sounding reference signal (SRS). The SRS can be transmitted in the last symbol of a subframe. The SRS can have a comb- type structure, and a UE can transmit an SRS on one of the combs. The SRS can be used by a base station for channel quality estimation to enable frequency- dependent scheduling on the uplink.

[0065] Figure 3D Examples of various uplink channels are illustrated. The PUCCH can be located as indicated in one configuration. The PUCCH carries uplink control information (UCI), which can include a scheduling request (SR), a channel quality indicator (CQI), a precoding matrix indicator (PMI), a rank indicator (RI), and hybrid automatic repeat request (HARQ) acknowledgment (ACK) / negative acknowledgment (NACK) feedback. The PUSCH carries data, and can additionally be used to carry a buffer status report (BSR), a power headroom report (PHR), and / or UCI.

[0066] Figure 4is a block diagram of a base station 410 in communication with a UE 450 in an access network 400. In the downlink, IP packets from the EPC 160 can be provided to a controller / processor 475. The controller / processor 475 implements layer 2 (L2) and layer 3 (L3) functionality. L3 includes RRC layer functionality, and L2 includes SDAP layer, PDCP layer, RLC layer, and MAC layer functionality. The controller / processor 475 provides RRC layer functionality associated with broadcasting of system information (e.g., MIB, SIBs), RRC connection control (e.g., RRC connection paging, RRC connection establishment, RRC connection modification, and RRC connection release), inter radio access technology (RAT) mobility, and measurement configuration associated with reporting measurements to the EPC 160; PDCP layer functionality associated with header compression / decompression, security (ciphering, deciphering, integrity protection, integrity verification), and handover support functions; RLC layer functionality associated with the transfer of upper layer packet data units (PDUs), error correction through ARQ, concatenation, segmentation, and reassembly of upper layer SDUs, re-segmentation of RLC data PDUs, and reordering of RLC data PDUs; and MAC layer functionality associated with mapping between logical channels and transport channels, multiplexing of MAC SDUs onto transport blocks (TBs), demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction through HARQ, priority handling, and logical channel prioritization.

[0067] The transmit (TX) processor 416 and the receive (RX) processor 470 implement layer 1 (LI) functionality associated with various signal processing functions. LI, which includes a PHY layer, can include error detection on the transport channels, forward error correction (FEC) coding / decoding of the transport channels, interleaving, rate matching, mapping to physical channels, modulation / demodulation of physical channels, and MIMO antenna processing. The TX processor 416 handles mapping to signal constellations based on various modulation schemes (e.g., binary phase-shift keying (BPSK), quadrature phase-shift keying (QPSK), M-phase-shift keying (M-PSK), M-quadrature amplitude modulation (M-QAM)). The coded and modulated symbols can then be split into parallel streams. Each stream can then be mapped to an OFDM subcarrier, multiplexed with a reference signal (e.g., pilot) in the time and / or frequency domain, and then combined together using an inverse fast Fourier transform (IFFT) to produce a physical channel carrying a time domain OFDM symbol stream. The OFDM stream is spatially precoded to produce multiple spatial streams if multiple spatial streams are

[0068] At the UE 450, each receiver 454RX receives a signal through at least one respective antenna 452. Each receiver 454RX recovers information modulated onto an RF carrier and provides the information to the receive (RX) processor 456. The TX processor 468 and the RX processor 456 implement Ll functionality associated with various signal processing functions. The RX processor 456 can perform spatial processing on the information to recover any spatial streams destined for the UE 450. If multiple spatial streams are destined for the UE 450, they can be combined by the RX processor 456 into a single OFDM symbol stream. The RX processor 456 then converts the OFDM symbol stream from the time domain to the frequency domain using a fast Fourier transform (FFT). The frequency domain signal comprises a separate OFDM symbol stream for each subcarrier of the OFDM signal. The symbols on each subcarrier, and the reference signal, are recovered and demodulated by determining the most likely signal constellation points transmitted by the base station 410. These soft decisions can be based on channel estimates computed by the channel estimator 458. The soft decisions are then decoded and deinterleaved to recover the data and control signals that were originally transmitted by the base station 410 on the physical channel. The data and control signals are then provided to the controller / processor 459, which implements L3 and L2 functionality.

[0069] The controller / processor 459 can be associated with a memory 460 that stores program codes and data. The memory 460 can be referred to as a computer-readable medium. In the uplink, the controller / processor 459 provides demultiplexing between transport and logical channels, packet reassembly, deciphering, header decompression, and control signal processing to recover IP packets from the EPC 160. The controller / processor 459 is also responsible for error detection using an ACK and / or NACK protocol to support HARQ operations.

[0070] Similar to the functionality described in connection with the downlink transmission by the base station 410, the controller / processor 459 provides RRC layer functionality associated with system information (e.g., MIB, SIBs) acquisition, RRC connections, and measurement reporting; PDCP layer functionality associated with header compression / decompression, and security (ciphering, deciphering, integrity protection, integrity verification); RLC layer functionality associated with the transfer of upper layer PDUs, error correction through ARQ, concatenation, segmentation, and reassembly of RLC SDUs, re-segmentation of RLC data PDUs, and reordering of RLC data PDUs; and MAC layer functionality associated with mapping between logical channels and transport channels, multiplexing of MAC SDUs onto TBs, demultiplexing of MAC SDUs from TBs, scheduling information reporting, error correction through HARQ, priority handling, and logical channel prioritization.

[0071] The TX processor 468 can use a channel estimator 458 to derive the channel response based on the reference signals or feedback transmitted by the base stations 410. The TX processor 468 then uses this channel response to select the appropriate coding and modulation schemes for the data rate selected by the controller / processor 459. The spatial stream generated by the TX processor 468 can be provided to different antenna 452 via separate transmitters 454TX. Each transmitter 454TX can modulate a respective spatial stream of the spatial streams onto a RF carrier.

[0072] The uplink transmission is processed at the base station 410 in a manner similar to that described in connection with the receiver function at the UE 450. Each receiver 418RX receives a signal through its respective antenna 420. Each receiver 418RX recovers information modulated onto an RF carrier and provides the information to a RX processor 470.

[0073] The controller / processor 475 can be associated with a memory 476 that stores program codes and data. The memory 476 can be referred to as a computer-readable medium. In the uplink, the controller / processor 475 provides demultiplexing between transport and logical channels, packet reassembly, deciphering, header decompression, control signal processing to recover IP packets from UE 450. IP packets from the controller / processor 475 can be provided to the EPC 160. The controller / processor 475 is also responsible for error detection using an ACK and / or NACK protocol to support HARQ operations.

[0074] In some aspects, at least one of the TX processor 468, the RX processor 456, and the controller / processor 459 can be configured to perform aspects relating to selecting a ML- assisted or non-ML-assisted scheme for CSI feedback from a UE (e.g., the UE 104 and / or the UE 450) to a network node (e.g., the base station 102 / 180 and / or the base station 410), as exemplified and described for element 198 of FIG. 13. Figure 1

[0075] In some other aspects, at least one of the TX processor 416, the RX processor 470, and the controller / processor 475 can be configured to perform aspects relating to selecting a ML-assisted or non-ML-assisted scheme for CSI feedback from a UE (e.g., the UE 104 and / or the UE 450) to a network node (e.g., the base station 102 / 180 and / or the base station 410), as exemplified and described for element 198 of FIG. 13. Figure 1

[0076] Figure 5 is a block diagram 500 illustrating a ML-assisted scheme for CSI feedback. An encoder 522 can be implemented at a UE, such as the UE 104 of FIG. 13 and / or the UE 450 of FIG. 14, while a decoder 524 can be implemented at a network node, such as the base station 102 / 180 of FIG. 13 and / or the base station 410 of FIG. 14. Figure 1 Figure 4 Figure 1 Figure 4

[0077] The encoder 522 and the decoder 524 can be matched and can include respective ones of ML models 526, 528. When a UE calculates raw CSI 530, the UE can encode the raw CSI 530 using the ML model 526. Such encoding can be used to predict one or more elements of a channel matrix and / or other CSI elements.

[0078] ​​​​​​The output of the encoder 522 can include encoded CSI 532, which can be transmitted by the UE to the network node. Upon receiving the encoded CSI 532, the network node can provide the encoded CSI 532 to a decoder 524. The ML model 528 of the decoder 524 can help recover the CSI from the encoded CSI 532, and potentially, predict or supplement one or more CSI components, e.g., based on a decoding process. The output of the decoder 524 can be reconstructed CSI 534, which can be used by the network node to configure communications with the UE according to an estimated channel between the UE and the network node.

[0079] Figure 6 is a call flow diagram 600 illustrating selecting an ML-aided or ML-unaided scheme for CSI feedback reported by a UE 604 to a network node 602. The network node 602 can be at least one of a base station 102 / 180 of Figure 1 and / or a base station 410 of Figure 4 . The UE 604 can be at least one of a UE 104 of Figure 1 and / or a UE 450 of Figure 4 . In some aspects, for Figure 5 The encoder 522 and the decoder 524 shown and described are illustrative of being included in the UE 604 and the network node 602, respectively.

[0080] The network node 602 and the UE 604 can be configured to use one of an ML-aided or non-ML-aided scheme for CSI feedback. As shown and described for Figure 5 , the ML-aided scheme involves ML models implemented at the encoder and the decoder of the UE 604 and the network node 602, respectively. For successful recovery of the encoded CSI and to provide a sufficiently accurate channel estimate, the ML model of the encoder and the ML model of the decoder can match each other. The CSI can be used for signaling from the network node 602 to the UE 604, such as data on PDSCH.

[0081] In the event of a mismatch between the encoder and the decoder, system performance can degrade and spectral efficiency can be impaired. In some such instances, the non-ML-aided scheme (or legacy scheme) can achieve better performance, at least until the ML models can be adjusted such that the encoder and the decoder match again. Accordingly, the network node 602 and the UE 604 can select between the ML-aided scheme and the non-ML-aided scheme for CSI feedback, depending on which scheme is able to achieve better spectral efficiency. The selection between the ML-aided scheme and the non-ML-aided scheme can be based on a UE decision or based on a network decision.

[0082] When operating on a cell provided by the network node 602, the UE 604 can receive various signaling from the network node 602 across a number of physical channels. For example, the network node 602 can transmit control information to the UE 604 on a PDCCH and data to the UE on a PDSCH. For a UE-based decision 610 to select a CSI feedback scheme, the UE 604 can indicate a selection of a CSI feedback scheme to the network node 602. As part of this process, the UE 604 can determine various spectral efficiencies.

[0083] In some aspects, in addition to data, the signaling 622 transmitted by the network node 602 to the UE 604 on the PDSCH 614 can also include reference signals, such as CSI-RS and / or DM-RS. The UE 604 can receive such signaling 622 and can measure one or more quantities or characteristics based on the signaling. For example, the UE 604 can measure one or more of a reference signal received power (RSRP), a reference signal received quality (RSRQ), a signal-to-noise ratio (SNR), a signal-to-interference-plus-noise ratio (SINR), and / or the like. In some aspects, the UE 604 can calculate, compute, derive, or otherwise determine at least one of a rank (or RI), a modulation and coding scheme (MCS), a block error rate (BLER), a PMI, a CQI, a received signal strength indicator (RSSI), and / or another value indicative of a quality and / or characteristic of a channel between the UE 604 and the network node 602.

[0084] In some aspects, the UE 604 can be configured to obtain a first spectral efficiency (SPEF) measurement based on at least one of a rank, an MCS, or a BLER associated with the signaling 622 received on the PDSCH 614 from the network node 602. For example, the signaling 622 can include, among other things, a set of CSI-RS and / or a set of data signals on the PDSCH 614. Based on one or more of a rank, an MCS, and / or a BLER obtained using the signaling 622 or associated therewith, the UE 604 can infer, derive, compute, or otherwise determine a spectral efficiency with which data is received on the PDSCH 614. The UE 604 can express the spectral efficiency as a measure in bits per second per Hertz (bit / s / Hz), and thus, a measure in bit / s / Hz can convey a net data rate in bits per second (bps) divided by a bandwidth in Hz.

[0085] Signaling received by the UE 604 on the PDSCH 614 can be transmitted by the network node 602 based on or in association with CSI feedback transmitted by the UE 604 to the network node 602. In some aspects, the UE 604 can calculate, compute, or otherwise determine CSI, e.g., based on measured quantities or characteristics associated with the signaling on the PDSCH 614. In some instances, the UE 604 and the network node 602 can initially employ an ML-aided scheme for CSI feedback. Thus, the UE 604 can encode the CSI (e.g., a channel matrix) using an encoder that includes an ML model that can estimate (e.g., predict or forecast) channel conditions between the UE 604 and the network node 602. Thus, the first SPEF measurement described above can indicate a first spectral efficiency achieved using the ML-aided scheme for CSI feedback.

[0086] In some aspects, the UE 604 can be further configured to obtain a second measurement based on signaling 624 received on a measurement resource 616, such as a channel measurement resource (CMR) or an interference measurement resource (IMR). The measurement resource 616 can be a resource scheduled by the network node 602 to be dedicated for measurements, such as channel measurements and / or interference measurements, by the UE 604. The UE 604 can receive the signaling 624 on the measurement resource 616 and can measure one or more quantities or characteristics based on the signaling. For example, the UE 604 can measure one or more of RSRP, RSRQ, SNR, SINR, and / or the like. In some other aspects, the UE 604 can calculate, compute, derive, or otherwise determine at least one of a rank (or RI), an MCS, a BLER, a PMI, a CQI, an RSSI, and / or another value indicative of a channel quality and / or characteristic between the UE 604 and the network node 602.

[0087] In some aspects, the UE 604 can use a non-ML-assisted scheme for CSI feedback (e.g., a reference CSI feedback scheme or a legacy CSI feedback scheme) to calculate, compute, or otherwise determine CSI based on the signaling 624 received on the measurement resources 616. The non-ML-assisted scheme can be any suitable CSI feedback scheme that is not characterized by an encoder / decoder with an ML model. In some aspects, multiple (non-ML-assisted or legacy) CSI feedback schemes can be available as the reference CSI feedback scheme. In some aspects, the UE 604 can be configured to select one of the CSI feedback schemes based at least on a size of a payload configured for CSI reporting encoded using an ML model or other AI-assisted mechanism. For example, the UE 604 can select one of the CSI feedback schemes having a payload size that is approximately equal to or within a threshold amount of a payload size of the CSI encoded using an ML-assisted encoder. For example, a standard promulgated by a standards body (such as the 3GPP) can specify a set of reference CSI feedback schemes or a set of legacy CSI feedback schemes having different CSI feedback payload sizes, and the UE 604 can select the CSI feedback scheme having a payload size corresponding (e.g., equal to or within a range including) a payload size of the ML-assisted CSI feedback. In some other aspects, the UE 604 can receive an RRC configuration message (or other RRC signaling message) from the network node 602 that (explicitly or implicitly) indicates which of the CSI feedback schemes should be implemented as the reference.

[0088] In some aspects, the UE 604 can be configured to obtain a second SPEF measurement based on measurements on the measurement resources 616. For example, the signaling 622 can include a set of CSI-RSs and / or a set of data signals on the PDSCH 614, among other examples. Based on one or more of a rank, MCS, and / or BLER obtained using the signaling 622 or in association therewith, the UE 604 can infer, derive, compute, or otherwise determine a spectral efficiency with which data is received using the reference CSI feedback scheme or the legacy CSI feedback scheme. For example, the UE 604 can receive another set of signals on at least one of the CMRs and / or IMRs. The UE 604 can measure an energy, signal strength, and / or another characteristic of the other set of signals. The UE 604 can infer, derive, compute, or otherwise determine another spectral efficiency with which the other set of signals is received on the CMRs and / or IMRs based on the measured energy, signal strength, and / or other characteristic of the other set of signals.

[0089] The other set of signals received by the UE 604 on the CMR and / or IMR can be transmitted by the network node 602 based on or in association with other CSI feedback provided by the UE 604 to the network node 602, but such other CSI feedback can be derived and encoded using a reference CSI feedback scheme that is not assisted by ML or another AI-assisted mechanism. Thus, the second measurement can be indicative of a second spectral efficiency that can be achieved using a scheme for CSI feedback that is not assisted by ML and AI.

[0090] The UE 604 can be configured to compare the first SPEF measurement and the second SPEF measurement. Based on the comparison, the UE 604 can be configured to determine which of the two measurements is higher or greater, which can be indicative of a more efficient, less error-prone, and / or otherwise “better” use of spectral resources. The UE 604 can select between the ML-assisted scheme and the non-ML assisted scheme for CSI feedback depending on which scheme is associated with the higher or greater measurement (and thus the better spectral efficiency). In some aspects, the UE 604 can compare the first SPEF measurement and the second SPEF measurement, in which case the selection between the ML-assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision by the UE 604.

[0091] In some aspects, the UE 604 can be configured to transmit a request to the network node 602 to use a non-ML CSI feedback scheme for CSI feedback when the first SPEF measurement is less than the second SPEF measurement. In some aspects, the request can be included in a MAC control element (CE). In some other aspects, the request can be included in UE assistance information (UAI).

[0092] The first SPEF measurement can be indicative of a first spectral efficiency achieved using the ML-assisted scheme for CSI feedback. The first SPEF measurement being less than the second SPEF measurement, for example by an offset or threshold amount, can be indicative of an encoder-decoder mismatch for the ML model currently being used for CSI feedback. Such a mismatch can cause spectral efficiency to be impaired, and potentially cause the ML-assisted CSI feedback scheme to become inferior to a non-ML assisted CSI feedback scheme or a legacy CSI feedback scheme. Thus, in order to transmit data efficiently and accurately over the channel, the network and the UE 604 can fall back to a non-ML assisted CSI feedback scheme, and / or the ML model being used to encode the CSI feedback can be retrained, retuned, and / or otherwise reconfigured to more accurately capture the channel state properties that the CSI is intended to indicate.

[0093] However, a second SPEF measurement less than a first SPEF measurement can indicate that the ML model provides more accurate and / or responsive predictions of channel conditions. Accordingly, UE 604 and network node 602 can continue to use the ML model to encode and decode CSI feedback from UE 604 to network node 602.

[0094] In some implementations, the decision to switch CSI feedback schemes can be a network-based decision 612. According to some aspects, UE 604 can be configured to receive, from network node 602, a request 630 for information indicating a second spectral efficiency (i.e., a spectral efficiency associated with a legacy CSI feedback scheme). For example, request 630 can be received in a MAC CE or RRC signaling message.

[0095] UE 604 can be configured to obtain one or more measurements associated with the second spectral efficiency. For example, as described for UE-based decision 610, UE 604 can be configured to obtain measurements based on signaling 624 received on measurement resources 616, such as CMRs or IMRs. UE 604 can measure one or more quantities or characteristics based on signaling received from network node 602 (e.g., signaling 624). For example, UE 604 can measure one or more of RSRP, RSRQ, SNR, SINR, and the like. In some other aspects, UE 604 can calculate, compute, derive, or otherwise determine at least one of rank (or RI), MCS, BLER, PMI, CQI, RSSI, and / or another value indicative of a quality and / or characteristic of a channel between UE 604 and network node 602.

[0096] Using the obtained measurements associated with the non-ML assisted CSI feedback scheme, UE 604 can determine a spectral efficiency associated with the non-ML assisted CSI feedback scheme. In some aspects, UE 604 can estimate the second spectral efficiency using CSI calculated with the non-ML assisted CSI feedback scheme. UE 604 can calculate the second spectral efficiency as a measurement in bits / s / Hz. Accordingly, when the CSI feedback is not assisted by an ML or another AI mechanism, the measurement in bits / s / Hz can convey a net data rate in bps divided by a bandwidth in Hz.

[0097] The UE 604 can be configured to transmit information indicating the second spectral efficiency to the network node 602. The information 632 indicating the second spectral efficiency can include a measurement in units of bits / s / Hz. In some aspects, the UE 604 can transmit such information in response to receiving a request 630 from the network node 602 for the information 632 indicating the second spectral efficiency. For example, the UE 604 can indicate the information 632 (indicating the second spectral efficiency) in a MAC CE. In some other aspects, the UE 604 can transmit the information 632 indicating the second spectral efficiency to the network node 602 in the absence of a request for the information indicating the second spectral efficiency; that is, the UE 604 can be configured to autonomously determine to transmit the information 632 indicating the second spectral efficiency to the network node 602. The UE 604 can do so, for example, periodically or in the event that the spectral efficiency associated with the ML-assisted CSI feedback scheme falls below a threshold. The UE 604 can include the information 632 indicating the second spectral efficiency in a UAI.

[0098] Accordingly, the network node 602 can be configured to obtain a first measurement indicating a spectral efficiency associated with a non-ML-assisted CSI feedback scheme, as provided by the UE 604. Additionally, the network node 602 can be configured to obtain a first estimate associated with the first spectral efficiency using HARQ feedback for signaling transmitted to the UE 604 on a PDSCH. For example, the network node 602 can count, over a period of time, at least one of a number of ACK messages and / or a number of NACK messages received from the UE 604 in response to signaling transmitted by the network node 602 on a PDSCH. The network node 602 can estimate a number of bits transmitted on the PDSCH that were successfully received by the UE 604 based on a number of TBs that were acknowledged as successfully received and / or a number of TBs that were not acknowledged as successfully received, in proportion to a total number of bits transmitted on the PDSCH. Over a discrete period of time on a discrete bandwidth, the network node 602 can estimate a spectral efficiency associated with the ML CSI feedback scheme in units of bits / s / Hz.

[0099] Further, the network node 602 can be configured to obtain a second estimate associated with the first spectral efficiency using at least one of an output of a decoder comprising an ML model or a CQI report based on the ML CSI feedback. For example, the network node 602 can obtain a message (e.g., a CSI report) sent by the UE 604, and the network node 602 can provide the message to a decoder with an ML model. The output of the decoder can include a reconstructed message (e.g., a reconstructed CSI) intended to recover the original message sent by the UE 604. The network node 602 can determine a number of errors in the reconstructed message, and / or the network node 602 can determine that a number of elements of the CSI matrix is inaccurate and / or does not adequately represent the channel H.

[0100] The network node 602 can select one of the ML CSI feedback scheme associated with the first spectral efficiency or the non-ML CSI feedback scheme associated with the second spectral efficiency. In some instances of the network-based decision 612, the network node 602 can compare the first measurement, the first estimate, and the second estimate. The network node 602 can select the non-ML CSI feedback scheme when the second spectral efficiency is greater than the first spectral efficiency, but can select the ML CSI feedback scheme when the second spectral efficiency is less than the first spectral efficiency.

[0101] In some aspects, in the case that the first measurement is greater than both the first estimate and the second estimate (which can be similar or approximately equal), the network node 602 can determine that the ML model is making the reconstruction of the CSI report inaccurate. For example, the ML model mismatch of the encoder and decoder can be degrading network performance represented via spectral efficiency. In such aspects, the network node 602 can select the ML-unassisted CSI feedback scheme, and thus, the CSI reporting between the UE 604 and the network node 602 can fall back to ML-unassisted CSI reporting.

[0102] In some other aspects, the first measurement can be similar to or approximately equal to the second estimate, and the first measurement and the second estimate can be greater than the first estimate. In such other aspects, the network node 602 can determine that the ML model of the decoder can provide some level of accuracy for the CSI used to recover the message, but that the channel estimate can be degraded. The network node 602 can determine to reconfigure the ML model, such as by adjusting weights, parameters, features, and / or other characteristics of the ML model. In such other aspects, the network node 602 can select the ML-assisted CSI feedback scheme or the ML-unassisted CSI feedback scheme.

[0103] In yet other aspects, the network node 602 can determine that the first estimate and the second estimate can be similar or approximately equal, both of which are greater than the first measurement. In such further aspects, the network node 602 can determine that the ML model improves spectral efficiency relative to the non-ML CSI feedback scheme, and thus, the network node 602 can select the ML-assisted CSI feedback scheme.

[0104] The network node 602 can transmit, to the UE 604, an instruction 640 indicating one of the ML CSI feedback scheme associated with the first spectral efficiency or the non-ML CSI feedback scheme associated with the second spectral efficiency to use for CSI feedback. In some aspects, the instruction 640 can be transmitted in at least one of a MAC CE and / or a RRC signaling message. In some aspects, the network node 602 can further reconfigure the ML model implemented at the encoder of the UE 604, such as by transmitting an instruction to the UE 604 to add, remove, and / or adjust weights and / or other parameters of the ML model, so that the channel between the UE 604 and the network node 602 can be more accurately estimated and spectral resources can be more efficiently utilized.

[0105] After receiving the instruction 640, the UE 604 can be configured to report CSI feedback 642 to the network node 602 using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme. For example, when the ML CSI feedback scheme is selected, the UE 604 can be configured to encode the CSI based on an ML model, such as an ML model that estimates or assists in estimating one or more quantities or characteristics of the channel between the UE 604 and the network node 602. Illustratively, one or more elements of the channel matrix can be encoded, estimated, or otherwise affected using the ML model or another AI mechanism. The UE 604 can transmit such encoded CSI feedback 642 to the network node 602.

[0106] However, when the non-ML CSI feedback scheme is selected, the UE 604 can be configured to calculate, compute, measure, derive, or otherwise determine the CSI without assistance of an ML model or other AI-assisted mechanism. For example, the UE 604 can use the signals received from the network node 602 to calculate, compute, derive, or otherwise determine the CSI such that the UE 604 obtains a channel matrix in which all elements are not produced by an output of a neural network or other AI-assisted mechanism. In some instances, the non-ML CSI feedback scheme can be implemented as a legacy (e.g., known) CSI feedback scheme. The UE 604 can transmit the CSI feedback 642 to the network node 602.

[0107] Figure 7is a flowchart 700 of a method of wireless communication. The method can be performed or executed by a UE (e.g., the UE 104, the UE 450, the UE 604), another wireless communication device (e.g., the apparatus 902), or one or more components thereof. One or more of the illustrated blocks can be omitted, transposed, and / or contemporaneously executed in various different aspects.

[0108] At 702, the UE can be configured to obtain a first measurement based on at least one of a rank, an MCS, or a BLER associated with signaling received on a PDSCH from a network node. For example, the UE can receive a set of signals (e.g., a set of CSI-RSs and / or a set of data signals) from the network node on a set of resources configured as a PDSCH. The UE can measure an energy, a signal strength, and / or another characteristic of the set of signals. The UE can use the measured energy, signal strength, and / or another characteristic to calculate, compute, infer, or otherwise determine one or more of a rank, an MCS, and / or a BLER associated with the set of signals. Based on or in association with one or more of the rank, the MCS, and / or the BLER, the UE can infer, derive, calculate, or otherwise determine a spectral efficiency at which data is received on the PDSCH.

[0109] The set of signals received on the PDSCH by the UE can be transmitted by the network node based on or in association with CSI feedback transmitted to the network node by the UE. The UE can encode such CSI feedback using an ML model or other AI-assisted mechanism included in or otherwise associated with an encoder of the UE. Thus, the first measurement can be indicative of a first spectral efficiency achieved using an ML-assisted scheme for the CSI feedback.

[0110] At 704, the UE can be configured to obtain a second measurement based on signaling received on at least one of a CMR or an IMR associated with a reference CSI feedback scheme. For example, the UE can receive another set of signals on at least one of the CMR and / or the IMR. The UE can measure an energy, a signal strength, and / or another characteristic of the other set of signals. The UE can infer, derive, calculate, or otherwise determine another spectral efficiency at which the other set of signals is received on the CMR and / or the IMR based on the measured energy, signal strength, and / or other characteristic of the other set of signals.

[0111] The other set of signals received by the UE on the CMR and / or IMR can be transmitted by the network node based on or in association with other CSI feedback provided by the UE to the network node, but such other CSI feedback can be derived and encoded using a reference CSI feedback scheme that is not assisted by ML or another AI assisted mechanism. Thus, the second measurement can be indicative of a second spectral efficiency that can be achieved using a scheme for CSI feedback that is not assisted by ML and AI. For example, the reference CSI feedback scheme can be a legacy CSI feedback scheme.

[0112] In some aspects, a plurality of (non-ML assisted or legacy) CSI feedback schemes can be available as the reference CSI feedback scheme. In some aspects, the UE can be configured to select one of the CSI feedback schemes based at least on a size of a payload configured for CSI reporting encoded using the ML model or other AI assisted mechanism. For example, the UE can select one of the CSI feedback schemes having a payload size that is approximately equal to or within a threshold amount of a payload size of the CSI encoded using the ML assisted encoder. In some other aspects, the UE can receive an RRC configuration message (or other RRC signaling message) from the network node that (explicitly or implicitly) indicates which of the CSI feedback schemes should be implemented as the reference.

[0113] At 706, the UE can be configured to receive, from the network node, a request for information indicative of the second spectral efficiency. In some aspects, the UE can receive the request for information indicative of the second spectral efficiency, in which case the selection between the ML assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision by the network node.

[0114] At 708, the UE can be configured to transmit, to the network node, information indicative of the second spectral efficiency. In some aspects, the UE can transmit such information in response to receiving the request for information indicative of the second spectral efficiency from the network node (as described in connection with 706). In some other aspects, the UE can transmit the information indicative of the second spectral efficiency to the network node in the absence of the request for information indicative of the second spectral efficiency; that is, the UE can be configured to autonomously determine to transmit the information indicative of the second spectral efficiency to the network node. For example, the UE can compare the first measurement to a threshold, and if the first measurement fails to satisfy (e.g., is less than) the threshold, the UE can autonomously determine to transmit the second measurement to the network node. In some aspects, the UE can transmit the information indicative of the second spectral efficiency, in which case the selection between the ML assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision by the network node.

[0115] At 710, the UE may be configured to compare a first measurement associated with a first spectral efficiency and a second measurement associated with a second spectral efficiency. Based on the comparison, the UE may be configured to determine which of the two measurements is higher or greater, which may indicate a more efficient, less error-prone, and / or other "better" use of spectrum resources. The UE may select between an ML-assisted scheme and a non-ML-assisted scheme for CSI feedback based on which scheme is associated with the higher or greater measurement (and, therefore, associated with better spectral efficiency). In some aspects, the UE may compare the first measurement associated with the first spectral efficiency and the second measurement associated with the second spectral efficiency, in which case the selection between the ML-assisted scheme and the non-ML-assisted scheme for CSI feedback is made based on a decision of the UE.

[0116] At 712, the UE may be configured to send a request to the network node to use a non-ML CSI feedback scheme for CSI feedback when the first measurement is less than the second measurement. In some aspects, the request may be included in a MAC CE. In some other aspects, the request may be included in a UAI. In some aspects, the UE may send a request to use a non-ML CSI feedback scheme for CSI feedback, in which case the selection between the ML-assisted scheme and the non-ML-assisted scheme for CSI feedback is based on the UE's decision.

[0117] The first measurement may indicate a first spectral efficiency achieved using an ML-assisted scheme for CSI feedback. A first measurement that is less than the second measurement, for example by an offset or threshold amount, may indicate an encoder-decoder mismatch for the ML model currently being used for CSI feedback. Such a mismatch may cause spectral efficiency to suffer and potentially cause the ML-assisted CSI feedback scheme to become inferior to a non-ML-assisted CSI feedback scheme or a legacy CSI feedback scheme. Therefore, in order to efficiently and accurately send data over the channel, the network and the UE may fall back to a non-ML-assisted CSI feedback scheme and / or the ML model being used to encode the CSI feedback may be retrained, retuned, and / or otherwise reconfigured to more accurately capture the channel state attributes that the CSI is intended to indicate.

[0118] However, a second measurement that is less than the first measurement may indicate that the ML model provides a more accurate and / or responsive snapshot of the channel state. Accordingly, the UE and the network node may continue to use the ML model to encode and decode CSI feedback from the UE to the network node.

[0119] At 714, the UE can be configured to receive, from the network node, an instruction indicating one of the ML CSI feedback scheme associated with the first spectral efficiency or the non-ML CSI feedback scheme associated with the second spectral efficiency to use for CSI feedback. In some aspects, the instruction can be received in at least one of a MAC CE and / or a RRC signaling message. In some aspects, the network node can further reconfigure the ML model implemented at the encoder of the UE, such as by sending an instruction to the UE to add, remove, and / or adjust weights and / or other parameters of the ML model, so that the channel between the UE and the network node can be more accurately estimated and spectral resources can be more efficiently utilized.

[0120] At 716, the UE can be configured to report, to the network node, CSI feedback using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme. For example, when the ML CSI feedback scheme is selected, the UE can be configured to encode the CSI based on a ML model, such as a ML model that estimates or assists in estimating one or more quantities or characteristics of the channel between the UE and the network node. Illustratively, one or more elements of the channel matrix can be encoded, estimated, or otherwise affected using a ML model or another AI mechanism. The UE can send such encoded CSI to the network node.

[0121] However, when the non-ML CSI feedback scheme is selected, the UE can be configured to calculate, compute, measure, derive, or otherwise determine the CSI without assistance of a ML model or other AI assisted mechanism. For example, the UE can calculate, compute, derive, or otherwise determine the CSI using signals received from the network node, such that the UE obtains a channel matrix in which all elements are not produced by an output of a neural network or other AI assisted mechanism. In some instances, the non-ML CSI feedback scheme can be implemented as a legacy (e.g., known) CSI feedback scheme. The UE can send the CSI to the network node.

[0122] Figure 8 FIG. 8 is a flow diagram of a method of wireless communication. The method can be performed by or at a base station (e.g., base station 102 / 180, 410), a network node (e.g., network node 602), another wireless communication device (e.g., device 1002), or one or more components thereof. One or more of the illustrated blocks can be omitted, transposed, and / or contemporaneously executed in various different arrangements, according to various different aspects.

[0123] At 802, the base station can be configured to receive, from the UE, a request to use a non-ML CSI feedback scheme instead of an ML CSI feedback scheme for CSI feedback. In some aspects, the request can indicate that a first spectral efficiency associated with the ML CSI feedback scheme is less than a second spectral efficiency associated with the non-ML CSI feedback scheme. In some aspects, the request is included in one of a MAC CE or a UAI. In some aspects, the request is associated with a reference CSI feedback scheme that is based on at least one of a payload size associated with the ML CSI feedback scheme or an RRC configuration message transmitted to the UE. In some aspects, the base station can receive the request for information indicating the second spectral efficiency, in which case the selection between the ML assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision of the UE.

[0124] At 804, the base station can be configured to transmit, to the UE, a request for at least one measurement indicating a second spectral efficiency associated with the non-ML CSI feedback scheme. In some aspects, the base station can transmit the request for the at least one measurement indicating the second spectral efficiency, in which case the selection between the ML assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision of the base station.

[0125] At 806, the base station can be configured to receive, from the UE, the at least one measurement indicating the second spectral efficiency. The at least one measurement indicating the second spectral efficiency can be included in a message associated with at least one of a periodic or a UAI. In some aspects, the base station can receive the at least one measurement indicating the second spectral efficiency, in which case the selection between the ML assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision of the base station.

[0126] At 808, the base station can be configured to obtain a first estimate associated with the first spectral efficiency using HARQ feedback for signaling transmitted to the UE on a PDSCH. For example, the base station can count, over a time period, at least one of a number of ACK messages and / or a number of NACK messages received from the UE in response to signaling transmitted by the base station on a PDSCH. The base station can estimate a number of bits transmitted on the PDSCH that are successfully received by the UE based on a number of TBs that are acknowledged as successfully received and / or a number of TBs that are not acknowledged as successfully received, in proportion to a total number of bits transmitted on the PDSCH. The base station can estimate a spectral efficiency associated with the ML CSI feedback scheme in units of bits / s / Hz over a discrete time period on a discrete bandwidth. In some aspects, the base station can obtain the first estimate associated with the first spectral efficiency, in which case the selection between the ML assisted scheme and the non-ML assisted scheme for CSI feedback is made based on a decision of the base station.

[0127] At 810, the base station can be configured to obtain a second estimate associated with the first spectral efficiency using at least one of an output of a decoder including an ML model or a CQI report based on ML CSI feedback. For example, the base station can obtain a message sent by the UE, and the base station can provide the message to a decoder with an ML model. The output of the decoder can include a reconstructed message intended to recover the original message sent by the UE. The base station can determine a number of errors in the reconstructed message, and / or the base station can determine that a number of elements of the CSI matrix is inaccurate and / or cannot adequately represent the channel H. In some aspects, the base station can obtain a second estimate associated with the first spectral efficiency, in which case the selection between the ML-aided and non-ML aided schemes for CSI feedback is made based on a decision of the base station.

[0128] At 812, the base station can select one of the ML CSI feedback scheme associated with the first spectral efficiency or the non-ML CSI feedback scheme associated with the second spectral efficiency. For example, the base station can compare the first measurement, the first estimate, and the second estimate. The base station can select the non-ML CSI feedback scheme when the second spectral efficiency is greater than the first spectral efficiency, but can select the ML CSI feedback scheme when the second spectral efficiency is less than the first spectral efficiency.

[0129] In some aspects, in the case that the first measurement is greater than both the first estimate and the second estimate (which can be similar or approximately equal), the base station can determine that the ML model is making the reconstruction of the CSI report inaccurate. For example, the ML model mismatch of the encoder and decoder can be degrading network performance represented via spectral efficiency. Potentially, the CSI report between the UE and the base station can fall back to an ML un-aided CSI report.

[0130] In some other aspects, the first measurement can be similar to or approximately equal to the second estimate, and the first measurement and the second estimate can be greater than the first estimate. In such other aspects, the base station can determine that the ML model of the decoder can provide some level of accuracy for the CSI used to recover the message, but the channel estimate can be degraded. The base station can determine to reconfigure the ML model, such as by adjusting weights, parameters, features, and / or other characteristics of the ML model.

[0131] In still other aspects, the base station can determine that the first estimate and the second estimate can be similar or approximately equal, both of which are greater than the first measurement. In such further aspects, the base station can determine that the ML model improves spectral efficiency relative to the non-ML CSI feedback scheme.

[0132] At 814, the base station can be configured to transmit an instruction to the UE indicating a selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme. In some aspects, the instruction can be transmitted in at least one of a MAC CE and / or a RRC signaling message.

[0133] At 816, the base station can be configured to reconfigure the ML model associated with the ML CSI feedback scheme. For example, the base station can calculate, compute, or otherwise determine one or more weights, biases, and / or other parameters of the ML model such that the channel between the UE and the network node can be more accurately estimated and the spectrum resources can be more efficiently utilized. In some aspects, the base station can identify one or more nodes associated with the ML model that are causing inaccuracies in the CSI reports. The base station can adjust the weights and / or biases of the one or more identified nodes.

[0134] At 818, the base station can be configured to receive, from the UE, CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme. For example, when the ML CSI feedback scheme is selected, the base station can decode the CSI based on the ML model, such as an ML model that decodes or estimates a quantity or a characteristic of a channel matrix representing the channel between the UE and the network node. Illustratively, one or more elements of the channel matrix can be decoded, estimated, or otherwise affected using the ML model or another AI mechanism.

[0135] However, when the non-ML CSI feedback scheme is selected, the base station can rely on the CSI feedback from the UE without assistance of the ML model or the other AI- assisted mechanism. For example, the base station can be configured to reconstruct a channel matrix in which all elements are not produced by an output of a neural network or other AI- assisted mechanism. In some instances, the non-ML CSI feedback scheme can be implemented as a legacy (e.g., known) CSI feedback scheme.

[0136] Figure 9 Diagram 900 is an example of a diagram 900 illustrating an example of a hardware implementation for the apparatus 902. The apparatus 902 can be a UE or similar device, or the apparatus 902 can be a component of a UE or similar device. The apparatus 902 can include a cellular baseband processor 904 (also referred to as a modem) and / or a cellular RF transceiver 922, which can be coupled with each other and / or integrated to the same package, component, circuit, chip, and / or other circuitry.

[0137] In some aspects, the device 902 may receive or include one or more subscriber identity module (SIM) cards 920, which may include one or more integrated circuits, chips, or similar circuits and may be removable or embedded. The one or more SIM cards 920 may carry identification and / or authentication information, such as an International Mobile Subscriber Identity (IMSI) and / or IMSI-related keys. In addition, the device 902 may include one or more of an application processor 906 coupled to a secure digital (SD) card 908 and a screen 910, a Bluetooth module 912, a wireless local area network (WLAN) module 914, a global positioning system (GPS) module 916, and / or a power supply 918.

[0138] The cellular baseband processor 904 communicates with the UE 104 and / or base station 102 / 180 via the cellular RF transceiver 922. The cellular baseband processor 904 may include computer-readable media / memory. The computer-readable media / memory may be non-transitory. The cellular baseband processor 904 is responsible for general processing, including executing software stored on the computer-readable media / memory. When executed by the cellular baseband processor 904, the software enables the cellular baseband processor 904 to perform the various functions described above. The computer-readable media / memory may also be used to store data manipulated by the cellular baseband processor 904 when executing the software. The cellular baseband processor 904 further includes a receiving component 930, a communication manager 932, and a transmitting component 934. The communication manager 932 includes one or more illustrated components. The components within the communication manager 932 may be stored in the computer-readable media / memory and / or configured as hardware within the cellular baseband processor 904.

[0139] exist Figure 4 In the context of , the cellular baseband processor 904 may be a component of the UE 450 and may include the memory 460 and / or at least one of the TX processor 468, the RX processor 456, and / or the controller / processor 459. In one configuration, the device 902 may be a modem chip and / or may be implemented as the baseband processor 904, while in another configuration, the device 902 may be the entire UE (e.g., Figure 4 UE 450) and may include some or all of the above components, circuits, chips, and / or other circuits illustrated in the context of device 902. In one configuration, the cellular RF transceiver 922 may be implemented as at least one of the transmitter 454TX and / or the receiver 454RX.

[0140] The reception component 930 can be configured to receive signaling on a wireless channel, such as signaling from a base station 102 / 180 or UE 104. The transmission component 934 can be configured to transmit signaling on a wireless channel, such as signaling to a base station 102 / 180 or UE 104. The communications manager 932 can coordinate or manage some or all of the wireless communications by the apparatus 902, including wireless communications across the reception component 930 and the transmission component 934.

[0141] The reception component 930 can provide some or all of the data and / or control information included in received signaling to the communications manager 932, and the communications manager 932 can generate and provide some or all of the data and / or control information to be included in transmitted signaling to the transmission component 934. The communications manager 932 can include various illustrated components, including one or more components configured to process received data and / or control information and / or one or more components configured to generate data and / or control information for transmission.

[0142] The communications manager 932 can include an obtaining component 940 that can be configured to obtain a first measurement based on at least one of a rank, an MCS, or a BLER associated with signaling received on a PDSCH from a base station 102 / 180, e.g., as described in connection with 702 of FIG. 7. For example, the reception component 930 can receive a set of signals (e.g., a set of CSI-RSs and / or a set of data signals) from the base station 102 / 180 on a set of resources configured as a PDSCH. The obtaining component 940 can measure an energy, a signal strength, and / or another characteristic of the set of signals. The obtaining component 940 can use the measured energy, signal strength, and / or another characteristic to calculate, compute, infer, or otherwise determine one or more of a rank, an MCS, and / or a BLER associated with the set of signals. Based on or in association with one or more of the rank, the MCS, and / or the BLER, the obtaining component 940 can calculate, estimate, infer, compute, or otherwise determine a spectral efficiency at which data is received on the PDSCH. Figure 7

[0143] The set of signals received on the PDSCH by the reception component 930 can be transmitted by the base station 102 / 180 based on or in association with CSI feedback transmitted to the base station 102 / 180 by the transmission component 934. The reporting component 944 described below can use an ML model or other AI-assisted mechanism included in or otherwise associated with an encoder of the apparatus 902 to encode such CSI feedback. Thus, the first measurement can indicate a first spectral efficiency achieved using an ML- assisted scheme for the CSI feedback.

[0144] ​The obtaining component 940 can be further configured to obtain the second measurement based on signaling received on at least one of the CMRs or IMRs associated with the reference CSI feedback scheme, e.g., as described in connection with 704. For example, the receiving component 930 can receive another set of signals on at least one of the CMRs and / or IMRs. The obtaining component 940 can measure an energy, a signal strength, and / or another characteristic of the other set of signals. The obtaining component 940 can calculate, estimate, derive, compute, or otherwise determine another spectral efficiency at which the other set of signals was received on the CMRs and / or IMRs based on the measured energy, signal strength, and / or other characteristic of the other set of signals. Figure 7

[0145] The other set of signals received by the receiving component 930 on the CMRs and / or IMRs can be transmitted by the base station 102 / 180 based on or associated with other CSI feedback provided by the apparatus 902 to the base station 102 / 180, but such other CSI feedback can be derived and encoded using a reference CSI feedback scheme that is not assisted by ML or another AI-assisted mechanism. Thus, the second measurement can be indicative of a second spectral efficiency that can be achieved using a scheme for CSI feedback that is not assisted by ML and AI. For example, the reference CSI feedback scheme can be a legacy CSI feedback scheme.

[0146] In some aspects, multiple (non-ML-assisted or legacy) CSI feedback schemes can be used as the reference CSI feedback scheme. In some aspects, the reporting component 944 (below) can be configured to select one of the CSI feedback schemes based at least on a size of a payload configured for a CSI report encoded using the ML model or other AI-assisted mechanism. For example, the comparing component 942 (below) can select one of the CSI feedback schemes having a payload size that is approximately equal to or within a threshold amount of a payload size of a CSI encoded using the ML-assisted encoder. In some other aspects, the receiving component 930 can receive a RRC configuration message (or another RRC signaling message) from the base station 102 / 180 that (explicitly or implicitly) indicates which of the CSI feedback schemes should be implemented as the reference.

[0147] The receiving component 930 can be configured to receive a request from the base station 102 / 180 for information indicative of the second spectral efficiency, e.g., as described in connection with 706. In some aspects, the receiving component 930 can receive the request for information indicative of the second spectral efficiency, in which case the selection between the ML-assisted scheme and the non-ML-assisted scheme for CSI feedback is made based on a decision of the base station 102 / 180. Figure 7

[0148] ​​The transmitting component 934 can be configured to transmit information indicating the second spectral efficiency to the base station 102 / 180, for example, in conjunction with Figure 7 708 . In some aspects, transmitting component 934 may transmit information indicating the second spectral efficiency in response to receiving a request from base station 102 / 180 for such information (as described in conjunction with 706 ). In some other aspects, transmitting component 934 may transmit information indicating the second spectral efficiency to base station 102 / 180 in the absence of a request for information indicating the second spectral efficiency; that is, apparatus 902 may be configured to autonomously determine to transmit information indicating the second spectral efficiency to base station 102 / 180. For example, comparing component 942 , described below, may compare the first measurement to a threshold, and if the first measurement fails to meet (e.g., is less than) the threshold, apparatus 902 (e.g., reporting component 944 , described below) may autonomously determine to transmit the second measurement to base station 102 / 180. In some aspects, transmitting component 934 may transmit information indicating the second spectral efficiency, in which case the selection between an ML-assisted scheme and a non-ML-assisted scheme for CSI feedback is based on a decision made by base station 102 / 180.

[0149] The communication manager 932 may further include a comparison component 942 that may be configured to compare a first measure associated with a first spectral efficiency and a second measure associated with a second spectral efficiency, e.g., as described in conjunction with Figure 7 Based on the comparison, the comparison component 942 can be configured to determine which of the two measurements is higher or greater, which can indicate a more efficient, less erroneous, and / or other "better" use of the spectrum resources. The comparison component 942 can select between an ML-assisted scheme and a non-ML-assisted scheme for CSI feedback based on which scheme is associated with the higher or greater measurement (and, therefore, associated with better spectrum efficiency). In some aspects, the comparison component 942 can compare a first measurement associated with a first spectrum efficiency and a second measurement associated with a second spectrum efficiency, in which case the selection between the ML-assisted scheme and the non-ML-assisted scheme for CSI feedback is based on a decision made by the device 902.

[0150] The transmitting component 934 may be further configured to transmit a request to the base station 102 / 180 to use a non-ML CSI feedback scheme for CSI feedback when the first measurement is less than the second measurement, e.g., as in conjunction with Figure 7The request can be included in a MAC CE in some aspects. In some other aspects, the request can be included in a UAI. In some aspects, transmitting component 934 can transmit a request to use a non-ML CSI feedback scheme for CSI feedback, in which case the selection between the ML-aided scheme and the non-ML-aided scheme for CSI feedback is made based on a decision of apparatus 902.

[0151] A first measurement can indicate a first spectral efficiency achieved using an ML-aided scheme for CSI feedback. The first measurement being less than the second measurement, e.g., by an offset or threshold amount, can indicate an encoder-decoder mismatch for the ML model currently being used for CSI feedback. Such a mismatch can cause spectral efficiency to be impaired and potentially cause the ML-aided CSI feedback scheme to become inferior to a non-ML-aided CSI feedback scheme or a legacy CSI feedback scheme. Thus, to transmit data efficiently and accurately over a channel, base station 102 / 180 and apparatus 902 can fall back to a non-ML-aided CSI feedback scheme and / or the ML model being used to encode CSI feedback can be retrained, retuned, and / or otherwise reconfigured to more accurately capture the channel state properties that CSI is intended to indicate.

[0152] However, the second measurement being less than the first measurement can indicate that the ML model provides a more accurate and / or more responsive snapshot of the channel state. Thus, apparatus 902 and base station 102 / 180 can continue to use the ML model to encode and decode CSI feedback from apparatus 902 to base station 102 / 180.

[0153] Receiving component 930 can be further configured to receive, from base station 102 / 180, an instruction indicating one of an ML CSI feedback scheme associated with the first spectral efficiency or a non-ML CSI feedback scheme associated with the second spectral efficiency to use for CSI feedback, e.g., as described in connection with Figure 7 714. In some aspects, the instruction can be received in at least one of a MAC CE and / or a RRC signaling message. In some aspects, base station 102 / 180 can further reconfigure the ML model implemented at an encoder of apparatus 902, such as by transmitting an instruction to apparatus 902 to add, remove, and / or adjust weights and / or other parameters of the ML model, so that the channel between apparatus 902 and base station 102 / 180 can be more accurately estimated and spectral resources can be more efficiently utilized.

[0154] Communication manager 932 can further include reporting component 944, which can be configured to report CSI feedback to base station 102 / 180 using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme, e.g., as described in connection with Figure 7The reporting component 944 can be configured to encode the CSI based on an ML model, such as an ML model that estimates or assists in estimating one or more quantities or characteristics of the channel between the estimating or assisting estimating apparatus 902 and the base station 102 / 180, for example, when the ML CSI feedback scheme is selected. Illustratively, one or more elements of the channel matrix can be encoded, estimated, or otherwise affected using an ML model or another AI mechanism. The reporting component 944 can cause the transmitting component 934 to transmit such encoded CSI to the base station 102 / 180.

[0155] However, when the non-ML CSI feedback scheme is selected, the reporting component 944 can be configured to calculate, compute, measure, derive, or otherwise determine the CSI without assistance of an ML model or other AI-assisted mechanism. For example, the reporting component 944 can use the signals received from the base station 102 / 180 to calculate, compute, derive, or otherwise determine the CSI such that the reporting component 944 obtains a channel matrix in which all elements are not produced by an output of a neural network or other AI-assisted mechanism. In some instances, the non-ML CSI feedback scheme can be implemented as a legacy (e.g., known) CSI feedback scheme. The reporting component 944 can cause the transmitting component 934 to transmit the CSI to the base station 102 / 180.

[0156] The apparatus 902 can include additional components that perform some or all of the blocks, operations, signaling, etc. in the aforementioned call flow diagrams and / or flowcharts of Figure 6 and Figure 7 Thus, some or all of the blocks, operations, signaling, etc. in the aforementioned call flow diagrams and / or flowcharts of Figure 6 and Figure 7 may be performed by one or more components, respectively, and the apparatus 902 can include one or more such components. These components can be one or more hardware components specifically configured to carry out the stated processes / algorithm, implemented by a processor specifically configured to carry out the stated processes / algorithm, stored within a computer-readable medium so as to be implemented by a processor, or some combination thereof.

[0157] In one configuration, the apparatus 902, and in particular the cellular baseband processor 904, includes means for receiving, from a network node, an instruction indicating one of an ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback; and means for reporting CSI feedback to the network node using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0158] In one configuration, the apparatus 902 (and in particular the cellular baseband processor 904) can further include means for comparing a first measurement associated with a first spectral efficiency and a second measurement associated with a second spectral efficiency, and the instructions to indicate one of the ML CSI feedback scheme or the non-ML CSI feedback scheme are based on comparing the first measurement and the second measurement.

[0159] In one configuration, the apparatus 902 (and in particular the cellular baseband processor 904) can further include means for obtaining the first measurement based on at least one of a rank, a MCS, or a BLER associated with signaling received on a PDSCH from the network node; and means for obtaining the second measurement based on signaling received on at least one of a CMR or an IMR associated with the reference CSI feedback scheme.

[0160] In one configuration, the reference CSI feedback scheme is based on at least one of a payload size associated with the ML CSI feedback scheme or a RRC configuration message received from the network node.

[0161] In one configuration, the apparatus 902 (and in particular the cellular baseband processor 904) can further include means for sending a request to the network node to use the non-ML CSI feedback scheme for CSI feedback when the first measurement is less than the second measurement, and the instructions are received in response to the request.

[0162] In one configuration, the request is included in one of a MAC CE or a UAI.

[0163] In one configuration, the apparatus 902 (and in particular the cellular baseband processor 904) can further include means for sending information indicating the second spectral efficiency to the network node, and the instructions are associated with the information indicating the second spectral efficiency.

[0164] In one configuration, the apparatus 902 (and in particular the cellular baseband processor 904) can further include means for receiving a request from the network node for the information indicating the second spectral efficiency, and the information indicating the second spectral efficiency is sent in response to the request.

[0165] In one configuration, the information indicating the second spectral efficiency is sent to the network node when there is no request for the information indicating the second spectral efficiency.

[0166] In one configuration, the instructions are included in one of a MAC CE or a RRC signaling message.

[0167] The aforementioned components can be one or more of the aforementioned components of the apparatus 902 configured to perform the functions recited by the aforementioned components. As described supra, the apparatus 902 can include the TX processor 468, the RX processor 456, and the controller / processor 459. As such, in one configuration, the aforementioned components can be the TX processor 468, the RX processor 456, and the controller / processor 459 configured to perform the functions recited by the aforementioned components.

[0168] Figure 10 is a diagram 1000 illustrating an example of a hardware implementation for an apparatus 1002. The apparatus 1002 can be a base station or a similar device or system, or the apparatus 1002 can be a component of a base station or a similar device or system. The apparatus 1002 can include a baseband unit 1004. The baseband unit 1004 can communicate through a cellular RF transceiver. For example, the baseband unit 1004 can communicate with a UE 104 (such as for downlink and / or uplink communications) and / or with a base station 102 / 180 (such as for IAB) through the cellular RF transceiver.

[0169] The baseband unit 1004 can include a computer-readable medium / memory that can be non-transitory. The baseband unit 1004 is responsible for general processing, including the execution of software stored on the computer-readable medium / memory. The software, when executed by the baseband unit 1004, causes the baseband unit 1004 to perform the various functions described supra. The computer-readable medium / memory can also be used for storing data that is manipulated by the baseband unit 1004 when executing software. The baseband unit 1004 further includes a reception component 1030, a communication manager 1032, and a transmission component 1034. The communication manager 1032 includes the one or more illustrated components. The components of the communication manager 1032 can be stored in the computer-readable medium / memory and / or configured as hardware within the baseband unit 1004. The baseband unit 1004 can be a component of the base station 410 and can include the memory 476 and / or at least one of the TX processor 416, the RX processor 470, and the controller / processor 475.

[0170] The reception component 1030 can be configured to receive signaling over a wireless channel, such as signaling from a UE 104 or a base station 102 / 180. The transmission component 1034 can be configured to transmit signaling over a wireless channel, such as signaling to a UE 104 or a base station 102 / 180. The communication manager 1032 can coordinate or manage some or all of the wireless communications by the apparatus 1002, including wireless communications across the reception component 1030 and the transmission component 1034.

[0171] The reception component 1030 can provide some or all of the data and / or control information included in received signaling to a communication manager 1032, and the communication manager 1032 can generate and provide some or all of the data and / or control information to be included in transmitted signaling to a transmission component 1034. The communication manager 1032 can include various illustrated components, including one or more components configured to process received data and / or control information and / or one or more components configured to generate data and / or control information for transmission. In some aspects, the generation of data and / or control information can include packetizing or otherwise reformatting data and / or control information received from a core network (such as core network 190 or EPC 160) for transmission.

[0172] The reception component 1030 can be configured to receive, from the UE 104, a request to use a non-ML CSI feedback scheme instead of an ML CSI feedback scheme for CSI feedback, e.g., as described in connection with 802 of FIG. 8. In some aspects, the request can indicate that a first spectral efficiency associated with the ML CSI feedback scheme is less than a second spectral efficiency associated with the non-ML CSI feedback scheme. In some aspects, the request is included in one of a MAC CE or a UAI. In some aspects, the request is associated with a reference CSI feedback scheme that is based on at least one of a payload size associated with the ML CSI feedback scheme or an RRC configuration message transmitted to the UE 104. In some aspects, the reception component 1030 can receive a request for information indicating the second spectral efficiency, in which case the selection between the ML-aided and non-ML-aided schemes for CSI feedback is made based on a decision of the UE 104. Figure 8

[0173] The transmission component 1034 can be configured to transmit, to the UE 104, a request for at least one measurement indicating a second spectral efficiency associated with a non-ML CSI feedback scheme, e.g., as described in connection with 804 of FIG. 8. In some aspects, the transmission component 1034 can transmit the request for the at least one measurement indicating the second spectral efficiency, in which case the selection between the ML-aided and non-ML-aided schemes for CSI feedback is made based on a decision of the apparatus 1002. Figure 8

[0174] The reception component 1030 can be configured to receive, from the UE 104, the at least one measurement indicating the second spectral efficiency, e.g., as described in connection with 806 of FIG. 8. In some aspects, the reception component 1030 can receive the at least one measurement indicating the second spectral efficiency, in which case the selection between the ML-aided and non-ML-aided schemes for CSI feedback is made based on a decision of the UE 104. Figure 8 ​​The at least one measurement indicative of the second spectral efficiency can be included in a message associated with at least one of the periodicity or the UAI. In some aspects, the reception component 1030 can receive the at least one measurement indicative of the second spectral efficiency, in which case the selection between the ML- assisted and non-ML-assisted schemes for CSI feedback is made based on a decision of the apparatus 1002.

[0175] The communication manager 1032 includes an obtaining component 1040 that can be configured to obtain a first estimate associated with the first spectral efficiency using HARQ feedback for signaling transmitted on a PDSCH to the UE 104, e.g., as described in connection with Figure 8 808 of FIG. 8. For example, the obtaining component 1040 can count, over a time period, at least one of a number of ACK messages and / or a number of NACK messages received from the UE 104 in response to signaling transmitted on the PDSCH by the apparatus 802. The obtaining component 1040 can estimate a number of bits transmitted on the PDSCH that are successfully received by the UE 104 based on a number of TBs that are acknowledged as successfully received and / or a number of TBs that are not acknowledged as successfully received, in proportion to a total number of bits transmitted on the PDSCH. The obtaining component 1040 can estimate a spectral efficiency associated with the ML CSI feedback scheme in units of bits / s / Hz over a discrete time period on a discrete bandwidth. In some aspects, the obtaining component 1040 can obtain the first estimate associated with the first spectral efficiency, in which case the selection between the ML-assisted and non-ML-assisted schemes for CSI feedback is made based on a decision of the apparatus 1002.

[0176] The obtaining component 1040 can be further configured to obtain a second estimate associated with the first spectral efficiency using at least one of an output of a decoder that includes an ML model or a CQI report based on the ML CSI feedback, e.g., as described in connection with Figure 8 810 of FIG. 8. For example, the obtaining component 1040 can obtain a message transmitted by the UE 104, and the obtaining component 1040 can provide the message to a decoder that has an ML model. The output of the decoder can include a reconstructed message that aims to recover the original message transmitted by the UE 104. The obtaining component 1040 can determine a number of errors in the reconstructed message, and / or the obtaining component 1040 can determine that a number of elements of the CSI matrix are inaccurate and / or do not adequately represent the channel H. In some aspects, the obtaining component 1040 can obtain the second estimate associated with the first spectral efficiency, in which case the selection between the ML-assisted and non-ML-assisted schemes for CSI feedback is made based on a decision of the apparatus 1002.

[0177] The communication manager 1032 can further include a selection component 1042 that can be configured to select one of an ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency, e.g., as described in conjunction with Figure 8 For example, the selection component 1042 can compare the first measurement, the first estimate, and the second estimate. The selection component 1042 can select a non-ML CSI feedback scheme when the second spectral efficiency is greater than the first spectral efficiency, but can select an ML CSI feedback scheme when the second spectral efficiency is less than the first spectral efficiency.

[0178] In some aspects, if the first measurement is greater than both the first estimate and the second estimate (which may be similar or approximately equal), selection component 1042 may determine that the ML model is causing inaccurate reconstruction of the CSI report. For example, a mismatch in the ML model between the encoder and decoder may be degrading network performance as indicated by spectral efficiency. Potentially, CSI reporting between UE 104 and device 1002 may fall back to ML-unassisted CSI reporting.

[0179] In some other aspects, the first measurement can be similar to or approximately equal to the second estimate, and the first measurement and the second estimate can be greater than the first estimate. In such other aspects, the selection component 1042 can determine that the decoder's ML model can provide a certain degree of accuracy for recovering the CSI of the message, but the channel estimate may be degraded. The selection component 1042 can determine to reconfigure the ML model, such as by adjusting weights, parameters, features, and / or other characteristics of the ML model.

[0180] In still other aspects, selecting component 1042 can determine that the first estimate and the second estimate can be similar or approximately equal, both greater than the first measurement.In such further aspects, selecting component 1042 can determine that the ML model improves spectral efficiency relative to non-ML CSI feedback schemes.

[0181] The transmitting component 1034 can be configured to transmit an instruction to the UE 104 indicating a selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme, e.g., as combined with Figure 8 In some aspects, the instruction may be sent in at least one of a MAC CE and / or an RRC signaling message.

[0182] The communication manager 1032 may further include a reconfiguration component 1044 that may be configured to reconfigure the ML model associated with the ML CSI feedback scheme, such as in conjunction with Figure 8The reconfiguration component 1044 can be configured to reconfigure the ML model based on the received feedback. For example, the reconfiguration component 1044 can calculate, compute, or otherwise determine one or more weights, biases, and / or other parameters of the ML model such that the channel between the UE 104 and the network node can be more accurately estimated and spectral resources can be more efficiently utilized, as described in connection with 816. In some aspects, the reconfiguration component 1044 can identify one or more nodes associated with the ML model that caused the inaccuracy of the CSI report. The reconfiguration component 1044 can adjust the weights and / or biases of the one or more identified nodes.

[0183] The reception component 1030 can be configured to receive, from the UE 104, CSI feedback associated with a selected one of a ML CSI feedback scheme or a non-ML CSI feedback scheme, e.g., as described in connection with Figure 8 The apparatus 1002 can include means for receiving, from the UE 104, CSI feedback associated with a selected one of a ML CSI feedback scheme or a non-ML CSI feedback scheme, e.g., as described in connection with

[0184] However, when the non-ML CSI feedback scheme is selected, the apparatus 1002 can rely on the CSI feedback from the UE 104 without assistance of a ML model or another AI- assisted mechanism. For example, the apparatus 1002 can include a component configured to reconstruct a channel matrix in which all elements are not produced by an output of a neural network or other AI-assisted mechanism. In some instances, the non-ML CSI feedback scheme can be implemented as a legacy (e.g., known) CSI feedback scheme.

[0185] The apparatus 1002 can include additional components that perform some or all of the blocks, operations, signaling, etc. of the aforementioned call flow diagrams and / or flowcharts of Figure 6 and Figure 8 The apparatus 1002 can include additional components that perform some or all of the blocks, operations, signaling, etc. of the aforementioned call flow diagrams and / or flowcharts of Figure 6 and Figure 8 Some or all of the blocks, operations, signaling, etc. in the aforementioned call flow diagrams and / or flowcharts of

[0186] In one configuration, the apparatus 1002, and in particular the baseband unit 1004, includes means for selecting one of a ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to be used by a UE for CSI feedback; means for transmitting an instruction to the UE indicating the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme; and means for receiving, from the UE, CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0187] In one configuration, the apparatus 1002, and in particular the baseband unit 1004, can further include means for receiving, from the UE, a request to use the non-ML CSI feedback scheme for CSI feedback, and the request indicates that the first spectral efficiency is less than the second spectral efficiency, and the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the request.

[0188] In one configuration, the request is included in one of a MAC CE or a UAI.

[0189] In one configuration, the request is associated with a reference CSI feedback scheme that is based on at least one of a payload size associated with the ML CSI feedback scheme or a RRC configuration message transmitted to the UE.

[0190] In one configuration, the apparatus 1002, and in particular the baseband unit 1004, can further include means for receiving, from the UE, at least one measurement indicating the second spectral efficiency; means for obtaining a first estimate associated with the first spectral efficiency using HARQ feedback for signaling transmitted to the UE on a PDSCH; and means for obtaining a second estimate associated with the first spectral efficiency using at least one of an output of a decoder that includes a ML model or a CQI report based on ML CSI feedback, and the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the at least one measurement indicating the second spectral efficiency, the first estimate, and the second estimate.

[0191] In one configuration, the non-ML CSI feedback scheme is selected when the at least one measurement indicating the second spectral efficiency is greater than the first estimate and the second estimate, and the ML CSI feedback scheme is selected when the at least one measurement indicating the second spectral efficiency is less than the first estimate and the second estimate.

[0192] In one configuration, the apparatus 1002, and in particular the baseband unit 1004, can further include means for transmitting, to the UE, a request for the at least one measurement indicating the second spectral efficiency, and the at least one measurement indicating the second spectral efficiency is received in response to the request for the at least one measurement.

[0193] In one configuration, the at least one measurement indicating the second spectral efficiency is included in a message associated with at least one of a periodic or a UAI.

[0194] In one configuration, the instructions are included in one of a MAC CE or an RRC signaling message.

[0195] In one configuration, the apparatus 1002, and in particular the baseband unit 1004, can further include means for reconfiguring a ML model associated with the ML CSI feedback scheme.

[0196] The aforementioned means can be one or more of the aforementioned components of the apparatus 1002 configured to perform the functions recited by the aforementioned means. As described supra, the apparatus 1002 can include the TX Processor 416, the RX Processor 470, and the controller / processor 475. As such, in one configuration, the aforementioned means can be the TX Processor 416, the RX Processor 470, and the controller / processor 475 configured to perform the functions recited by the aforementioned means.

[0197] The specific order or hierarchy of various blocks or operations within each of the aforementioned processes, flow charts, and other diagrams herein can be performed in any order, omission, and / or concurrently, depending on the design of the method, without departing from the scope of the present disclosure. Also, certain blocks or operations can be combined or omitted, without departing from the scope of the present disclosure. The accompanying method claims present elements of the various blocks or operations in a sample order, and are not meant to be limited to the specific order or hierarchy presented.

[0198] The following example clauses are illustrative only, and can be combined with aspects of other embodiments or teachings described herein, without limitation.

[0199] 1. A method of wireless communication at a UE, the method comprising:

[0200] receiving, from a network node, instructions indicating one of a ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback; and

[0201] reporting the CSI feedback to the network node using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0202] 2. The method of clause 1, the method further comprising:

[0203] comparing a first measurement associated with the first spectral efficiency and a second measurement associated with the second spectral efficiency, and the instructions to indicate the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme are based on comparing the first measurement and the second measurement.

[0204] 3. The method of clause 2, the method further comprising:

[0205] obtaining the first measurement based on at least one of a rank, MCS, or BLER associated with signaling received from the network node on a PDSCH; and

[0206] obtaining the second measurement based on signaling received on at least one of a CMR or an IMR associated with a reference CSI feedback scheme.

[0207] 4. The method of clause 3, and the reference CSI feedback scheme is based on at least one of a payload size associated with the ML CSI feedback scheme or an RRC configuration message received from the network node.

[0208] 5. The method of clause 4, the method further comprising:

[0209] when the first measurement is less than the second measurement, sending a request to the network node to use the non-ML CSI feedback scheme for the CSI feedback, and the instructions are received in response to the request.

[0210] 6. The method of clause 5, and the request is included in one of a MAC CE or a UAI.

[0211] 7. The method of clause 1, the method further comprising:

[0212] sending information to the network node indicating the second spectral efficiency, and the instructions are associated with the information indicating the second spectral efficiency.

[0213] 8. The method of clause 7, the method further comprising:

[0214] receiving, from the network node, a request for the information indicating the second spectral efficiency, and the information indicating the second spectral efficiency is transmitted in response to the request.

[0215] 9. The method of any of clauses 7 or 8, and transmitting, to the network node, the information indicating the second spectral efficiency when there is no request for the information indicating the second spectral efficiency.

[0216] 10. The method of any of clauses 1-9, and the instructions are included in one of a MAC CE or an RRC signaling message.

[0217] 11. A method of wireless communication at a network node, the method comprising:

[0218] selecting one of a ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to be used by a UE for CSI feedback;

[0219] transmitting, to the UE, instructions indicating the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme; and

[0220] receiving, from the UE, the CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0221] 12. The method of clause 11, the method further comprising:

[0222] receiving, from the UE, a request to use the non-ML CSI feedback scheme for the CSI feedback, and the request indicates that the first spectral efficiency is less than the second spectral efficiency, and the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the request.

[0223] 13. The method of clause 12, and the request is included in one of a MAC CE or a UAI.

[0224] 14. The method of any of clauses 12 or 13, and the request is associated with a reference CSI feedback scheme that is based on at least one of a payload size associated with the ML CSI feedback scheme or an RRC configuration message transmitted to the UE.

[0225] 15. The method of any of clauses 12-14, the method further comprising:

[0226] receiving, from the UE, at least one measurement indicative of the second spectral efficiency;

[0227] obtaining a first estimate associated with the first spectral efficiency using HARQ feedback for signaling transmitted to the UE on a PDSCH;

[0228] obtaining a second estimate associated with the first spectral efficiency using at least one of an output of a decoder comprising an ML model or a CQI report based on ML CSI feedback, and

[0229] the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the at least one measurement indicative of the second spectral efficiency, the first estimate, and the second estimate.

[0230] 16. The method of clause 15, and when the at least one measurement indicative of the second spectral efficiency is greater than the first estimate and the second estimate, the non-ML CSI feedback scheme is selected, and when the at least one measurement indicative of the second spectral efficiency is less than the first estimate and the second estimate, the ML CSI feedback scheme is selected.

[0231] 17. The method of any of clauses 15 or 16, the method further comprising:

[0232] transmitting, to the UE, a request for the at least one measurement indicative of the second spectral efficiency, and the at least one measurement indicative of the second spectral efficiency is received in response to the request for the at least one measurement.

[0233] 18. The method of any of clauses 15-17, and the at least one measurement indicative of the second spectral efficiency is included in a message associated with at least one of periodic or UAI.

[0234] 19. The method of any of clauses 11-18, and the instructions are included in one of a MAC CE or an RRC signaling message.

[0235] 20. The method of any of clauses 11-19, the method further comprising:

[0236] reconfiguring an ML model associated with the ML CSI feedback scheme.

[0237] 21. An apparatus for wireless communication at a UE, the apparatus comprising:

[0238] a memory; and

[0239] at least one processor coupled to the memory and configured to:

[0240] receive, from a network node, an instruction indicating one of an ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback; and

[0241] report the CSI feedback to the network node using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0242] 22. The apparatus of clause 21, and the at least one processor is further configured to:

[0243] compare a first measurement associated with the first spectral efficiency and a second measurement associated with the second spectral efficiency, and the instruction indicating the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is based on comparing the first measurement and the second measurement.

[0244] 23. The apparatus of clause 22, and the at least one processor is further configured to:

[0245] obtain the first measurement based on at least one of a rank, a MCS, or a BLER associated with signaling received on a PDSCH from the network node; and

[0246] obtain the second measurement based on signaling received on at least one of a CMR or an IMR associated with a reference CSI feedback scheme.

[0247] 24. The apparatus of clause 23, and the reference CSI feedback scheme is based on at least one of a payload size associated with the ML CSI feedback scheme or an RRC configuration message received from the network node.

[0248] 25. The apparatus of clause 24, and the at least one processor is further configured to:

[0249] when the first measurement is less than the second measurement, send a request to the network node to use the non-ML CSI feedback scheme for the CSI feedback, and the instruction is received in response to the request.

[0250] 26. The apparatus of clause 25, and the request is included in one of a MAC CE or a UAI.

[0251] 27. The apparatus of clause 21, and the at least one processor is further configured to:

[0252] transmit, to the network node, information indicating the second spectral efficiency, and the instructions are associated with the information indicating the second spectral efficiency.

[0253] 28. The apparatus of clause 27, and the at least one processor is further configured to:

[0254] receive, from the network node, a request for the information indicating the second spectral efficiency, and the information indicating the second spectral efficiency is transmitted in response to the request.

[0255] 29. The apparatus of any of clauses 27 or 28, and the information indicating the second spectral efficiency is transmitted to the network node when there is no request for the information indicating the second spectral efficiency.

[0256] 30. The apparatus of any of clauses 21 to 29, and the instructions are included in one of a MAC CE or an RRC signaling message.

[0257] 31. An apparatus for wireless communication at a network node, the apparatus comprising:

[0258] a memory; and

[0259] at least one processor coupled to the memory and configured to:

[0260] select one of an ML CSI feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to be used by a UE for CSI feedback;

[0261] transmit, to the UE, instructions indicating the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme; and

[0262] receive, from the UE, the CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

[0263] 32. The apparatus of clause 31, and the at least one processor is further configured to:

[0264] receiving, from the UE, a request to use the non-ML CSI feedback scheme for the CSI feedback, and the request indicates that the first spectral efficiency is less than the second spectral efficiency, and the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the request.

[0265] 33. The apparatus of clause 32, and the request is included in one of a MAC CE or a UAI.

[0266] 34. The apparatus of any of clauses 32 or 33, and the request is associated with a reference CSI feedback scheme that is based on at least one of a payload size associated with the ML CSI feedback scheme or an RRC configuration message sent to the UE.

[0267] 35. The apparatus of any of clauses 32 to 34, and the at least one processor is further configured to:

[0268] receive, from the UE, at least one measurement indicating the second spectral efficiency;

[0269] obtain a first estimate associated with the first spectral efficiency using HARQ feedback for signaling sent to the UE on a PDSCH;

[0270] obtain a second estimate associated with the first spectral efficiency using at least one of an output of a decoder that includes an ML model or a CQI report based on ML CSI feedback, and

[0271] the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the at least one measurement indicating the second spectral efficiency, the first estimate, and the second estimate.

[0272] 36. The apparatus of clause 35, and the non-ML CSI feedback scheme is selected when the at least one measurement indicating the second spectral efficiency is greater than the first estimate and the second estimate, and the ML CSI feedback scheme is selected when the at least one measurement indicating the second spectral efficiency is less than the first estimate and the second estimate.

[0273] 37. The apparatus of any of clauses 35 or 36, and the at least one processor is further configured to:

[0274] send, to the UE, a request for the at least one measurement indicating the second spectral efficiency, and the at least one measurement indicating the second spectral efficiency is received in response to the request for the at least one measurement.

[0275] 38. An apparatus as described in any of clauses 35 to 37, and the at least one measurement indicative of the second spectral efficiency is included in a message associated with at least one of periodicity or UAI.

[0276] 39. An apparatus as described in any of clauses 31 to 38, and the instruction is included in one of a MAC CE or an RRC signalling message.

[0277] 40. An apparatus according to any of clauses 31 to 39, and the at least one processor is further configured to:

[0278] Reconfiguring the ML model associated with the ML CSI feedback scheme.

[0279] The foregoing description is provided to enable one of ordinary skill in the art to practice the various aspects described herein. Various modifications to these aspects will be readily understood by one of ordinary skill in the art, and the general principles defined herein may be applied to other aspects. Accordingly, the claims are not intended to be limited to the aspects shown herein, but rather should be given the full scope consistent with the language. Accordingly, the language employed herein is not intended to limit the scope of the claims to only those aspects shown herein, but rather should be given the full scope consistent with the language of the claims.

[0280] As an example, the language "determine" may encompass a wide variety of actions and, therefore, may not be limited to the concepts and aspects explicitly described or illustrated by this disclosure. In some contexts, "determine" may include calculating, computing, processing, measuring, deriving, investigating, searching (e.g., searching in a table, database, or another data structure), ascertaining, resolving, selecting, choosing, establishing, and the like. In some other contexts, "determining" may include communication and / or memory operations / processes such as "receiving" (e.g., receiving information), "accessing" (e.g., accessing data in a memory), "detecting," and the like, by obtaining information or values.

[0281] As another example, reference to an element in the singular is not intended to mean “one and only one” unless specifically stated, but rather “one or more.” Furthermore, terms such as “if’ and “when” are to be interpreted as meaning “under the condition that’ unless otherwise specifically stated. That is, these phrases (e.g., “when,” “if,” “in response to determining”) are not to be construed as requiring a temporal or chronological relationship among the actions or events described, but rather are to be construed as requiring a causal relationship among the actions or events described. The word “exemplary” is used herein to mean “serving as an example, instance, or illustration.” Any aspect described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other aspects. Unless specifically stated otherwise, the term “some” refers to one or more. Combinations such as “at least one of A, B, or C,” “one or more of A, B, or C,” “at least one of the group consisting of A, B, and C,” “one or more of the group consisting of A, B, and C,” and “A, B, and / or C or any combination thereof’ include number one (1) A, number one (1) B, number one (1) C, or a combination of more than one (1) A, more than one (1) B, more than one (1) C, or more than one (1) A and more than one (1) B, more than one (1) A and more than one (1) C, more than one (1) B and more than one (1) C, or more than one (1) A, more than one (1) B, and more than one (1) C. Specifically, combinations such as “at least one of A, B, or C,” “one or more of A, B, or C,” “at least one of the group consisting of A, B, and C,” “one or more of the group consisting of A, B, and C,” and “A, B, and / or C or any combination thereof’ can be A only, B only, C only, A and B, A and C, B and C, or A and B and C, where any such combination can contain one or more members. All structural and functional equivalents to the elements of the various aspects described throughout this disclosure that are known or later come to be known to those of ordinary skill in the art are expressly incorporated herein by reference and are intended to be encompassed by the claims. Moreover, nothing disclosed herein is intended to be dedicated to the public regardless of whether such disclosure is explicitly recited in the claims. The words “module,” “mechanism,” “element,” “device,” and the like do not require that all of these components be implemented in a single physical device. Rather, any of these components can be implemented in a single physical device or multiple physical devices. Furthermore, if used herein, the terms “determining” and “displaying” generally can refer to actions performed by one or more computing devices and are not intended to be limited to actions performed by a human being.

Claims

1. A method of wireless communication at a user equipment (UE), the method comprising: receiving, from a network node, an instruction indicating one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback; and reporting the CSI feedback to the network node using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

2. The method of claim 1, the method further comprising: comparing a first measurement associated with the first spectral efficiency and a second measurement associated with the second spectral efficiency, wherein the instruction indicating the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is based on comparing the first measurement and the second measurement.

3. The method of claim 2, the method further comprising: obtaining the first measurement based on at least one of a rank, a modulation and coding scheme (MCS), or a block error rate (BLER) associated with signaling received from the network node on a physical downlink shared channel (PDSCH); and obtaining the second measurement based on signaling received on at least one of a channel measurement resource (CMR) or an interference measurement resource (IMR) associated with a reference CSI feedback scheme.

4. The method of claim 3, wherein the reference CSI feedback scheme is based on at least one of a payload size associated with the ML CSI feedback scheme or a radio resource control (RRC) configuration message received from the network node.

5. The method of claim 4, the method further comprising: when the first measurement is less than the second measurement, sending a request to the network node that the non-ML CSI feedback scheme is to be used for the CSI feedback, wherein the instruction is received in response to the request.

6. The method of claim 5, wherein the request is included in one of a medium access control (MAC) control element (CE) or UE assistance information (UAI).

7. The method of claim 1, the method further comprising: sending information indicating the second spectral efficiency to the network node, wherein the instruction is associated with the information indicating the second spectral efficiency.

8. The method of claim 7, the method further comprising: receiving, from the network node, a request for the information indicating the second spectral efficiency, wherein the information indicating the second spectral efficiency is sent in response to the request.

9. The method of claim 7, wherein the information indicating the second spectral efficiency is sent to the network node when there is no request for the information indicating the second spectral efficiency. ​ 10. The method of claim 1, wherein the instructions are included in one of a medium access control (MAC) control element (CE) or a radio resource control (RRC) signaling message.

11. A method of wireless communication at a network node, the method comprising: selecting one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to be used by a user equipment (UE) for CSI feedback; transmitting, to the UE, instructions indicating the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme; and receiving, from the UE, the CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

12. The method of claim 11, the method further comprising: receiving, from the UE, a request to use the non-ML CSI feedback scheme for the CSI feedback, wherein the request indicates that the first spectral efficiency is less than the second spectral efficiency, and wherein the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the request.

13. The method of claim 12, wherein the request is included in one of a medium access control (MAC) control element (CE) or UE assistance information (UAI).

14. The method of claim 12, wherein the request is associated with a reference CSI feedback scheme based on at least one of a payload size associated with the ML CSI feedback scheme or a radio resource control (RRC) configuration message transmitted to the UE.

15. The method of claim 12, the method further comprising: receiving, from the UE, at least one measurement indicating the second spectral efficiency; obtaining a first estimate associated with the first spectral efficiency using hybrid automatic repeat request (HARQ) feedback for signaling transmitted to the UE on a physical downlink shared channel (PDSCH); obtaining a second estimate associated with the first spectral efficiency using at least one of an output of a decoder comprising a ML model or a channel quality indicator (CQI) report based on ML CSI feedback, wherein the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the at least one measurement indicating the second spectral efficiency, the first estimate, and the second estimate.

16. The method of claim 15, wherein the non-ML CSI feedback scheme is selected when the at least one measurement indicating the second spectral efficiency is greater than the first estimate and the second estimate, and wherein the ML CSI feedback scheme is selected when the at least one measurement indicating the second spectral efficiency is less than the first estimate and the second estimate.

17. The method of claim 15, the method further comprising: transmitting, to the UE, a request for the at least one measurement indicating the second spectral efficiency, wherein the at least one measurement indicating the second spectral efficiency is received in response to the request for the at least one measurement.

18. The method of claim 15, wherein the at least one measurement indicating the second spectral efficiency is included in a message associated with at least one of periodic or UE assistance information (UAI).

19. The method of claim 11, wherein the instructions are included in one of a medium access control (MAC) control element (CE) or a radio resource control (RRC) signaling message.

20. The method of claim 11, the method further comprising: reconfiguring a ML model associated with the ML CSI feedback scheme.

21. An apparatus for wireless communication at a user equipment (UE), the apparatus comprising: a memory; and at least one processor coupled to the memory and configured to: receive, from a network node, instructions indicating one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to use for CSI feedback; and report the CSI feedback to the network node using the indicated one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

22. The apparatus of claim 21, wherein the at least one processor is further configured to: compare a first measurement associated with the first spectral efficiency and a second measurement associated with the second spectral efficiency, wherein the instructions indicating the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is based on comparing the first measurement and the second measurement.

23. The apparatus of claim 22, wherein the at least one processor is further configured to: obtain the first measurement based on at least one of a rank, a modulation and coding scheme (MCS), or a block error rate (BLER) associated with signaling received from the network node on a physical downlink shared channel (PDSCH); and obtain the second measurement based on signaling received on at least one of a channel measurement resource (CMR) or an interference measurement resource (IMR) associated with a reference CSI feedback scheme.

24. The apparatus of claim 23, wherein the reference CSI feedback scheme is based on at least one of a payload size associated with the ML CSI feedback scheme or a radio resource control (RRC) configuration message received from the network node.

25. The apparatus of claim 24, wherein the at least one processor is further configured to: transmit, to the network node, a request to use the non-ML CSI feedback scheme for the CSI feedback when the first measurement is less than the second measurement, wherein the instructions are received in response to the request.

26. The apparatus of claim 25, wherein the request is included in one of a medium access control (MAC) control element (CE) or UE assistance information (UAI).

27. The apparatus of claim 21, wherein the at least one processor is further configured to: transmit, to the network node, information indicating the second spectral efficiency, wherein the instructions are associated with the information indicating the second spectral efficiency.

28. The apparatus of claim 27, wherein the at least one processor is further configured to: receive, from the network node, a request for the information indicating the second spectral efficiency, wherein the information indicating the second spectral efficiency is transmitted in response to the request.

29. The apparatus of claim 27, wherein the information indicating the second spectral efficiency is transmitted, to the network node, when there is no request for the information indicating the second spectral efficiency.

30. The apparatus of claim 21, wherein the instructions are included in one of a medium access control (MAC) control element (CE) or a radio resource control (RRC) signaling message.

31. An apparatus for wireless communication at a network node, the apparatus comprising: a memory; and at least one processor coupled to the memory and configured to: select one of a machine learning (ML) channel state information (CSI) feedback scheme associated with a first spectral efficiency or a non-ML CSI feedback scheme associated with a second spectral efficiency to be used by a user equipment (UE) for CSI feedback; transmit, to the UE, instructions indicating the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme; and receive, from the UE, the CSI feedback associated with the selected one of the ML CSI feedback scheme or the non-ML CSI feedback scheme.

32. The apparatus of claim 31, wherein the at least one processor is further configured to: receive, from the UE, a request to use the non-ML CSI feedback scheme for the CSI feedback, wherein the request indicates that the first spectral efficiency is less than the second spectral efficiency, and wherein the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the request.

33. The apparatus of claim 32, wherein the request is included in one of a medium access control (MAC) control element (CE) or UE assistance information (UAI). ​ 34. The apparatus of claim 32, wherein the request is associated with a reference CSI feedback scheme that is based on at least one of a payload size associated with the ML CSI feedback scheme or a radio resource control (RRC) configuration message sent to the UE.

35. The apparatus of claim 32, wherein the at least one processor is further configured to: receive, from the UE, at least one measurement indicative of the second spectral efficiency; obtain a first estimate associated with the first spectral efficiency using hybrid automatic repeat request (HARQ) feedback for signaling sent to the UE on a physical downlink shared channel (PDSCH); obtain a second estimate associated with the first spectral efficiency using at least one of an output of a decoder comprising an ML model or a channel quality indicator (CQI) report based on ML CSI feedback, wherein the one of the ML CSI feedback scheme or the non-ML CSI feedback scheme is selected based on the at least one measurement indicative of the second spectral efficiency, the first estimate, and the second estimate.

36. The apparatus of claim 35, wherein the non-ML CSI feedback scheme is selected when the at least one measurement indicative of the second spectral efficiency is greater than the first estimate and the second estimate, and wherein the ML CSI feedback scheme is selected when the at least one measurement indicative of the second spectral efficiency is less than the first estimate and the second estimate.

37. The apparatus of claim 35, wherein the at least one processor is further configured to: transmit, to the UE, a request for the at least one measurement indicative of the second spectral efficiency, wherein the at least one measurement indicative of the second spectral efficiency is received in response to the request for the at least one measurement.

38. The apparatus of claim 35, wherein the at least one measurement indicative of the second spectral efficiency is included in a message associated with at least one of periodic or UE assistance information (UAI).

39. The apparatus of claim 31, wherein the instructions are included in one of a medium access control (MAC) control element (CE) or a radio resource control (RRC) signaling message.

40. The apparatus of claim 31, wherein the at least one processor is further configured to: reconfigure an ML model associated with the ML CSI feedback scheme.

Images