Channel state information compression based on low complexity machine learning

Abstract

The present disclosure provides systems, apparatuses, apparatuses, and methods, including computer programs encoded on a storage medium, for CSI compression (270a) based on low complexity ML. The UE (102) receives (240), from the network entity (104), a CSI-RS for the channel estimation (250). The UE (102) sends (285), to the network entity (104), a CSI report comprising the first PMI and the second PMI. The first PMI indicates a wideband precoder (355) associated with the channel estimate (250), and the second PMI indicates a compressed subband feature vector associated with the wideband precoder (355).

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of and priority to International Application No. PCT / CN2022 / 112193, filed on August 12, 2022, entitled “CSI Reports based on ML Techniques,” which is expressly incorporated herein by reference in its entirety. Technical Field

[0002] The present disclosure relates generally to wireless communications, and more particularly, to channel state information (CSI) reporting based on machine learning (ML) techniques. Background Art

[0003] The Third Generation Partnership Project (3GPP) has specified a radio interface called Fifth Generation (5G) New Radio (NR) (5G NR). The architecture of a 5G NR wireless communication system includes the 5G Core (5GC) network, the 5G Radio Access Network (5G-RAN), and user equipment (UE). Compared to previous generation cellular communication systems, the 5G NR architecture aims to provide increased data rates, reduced latency, and / or increased capacity.

[0004] Wireless communication systems generally provide various telecommunication services (e.g., telephony, video, data, messaging, broadcast, etc.) based on multiple access technologies (such as Orthogonal Frequency Division Multiple Access (OFDMA)) that support communication with multiple UEs. Improvements in mobile broadband continue to advance such wireless communication technologies. For example, user equipment (UE) and base stations can support more antenna configurations and multiple connections. However, one consequence is that channel state information (CSI) reports have become larger and more complex. Summary of the Invention

[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. This summary does not identify key or critical elements of all aspects, nor does it 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 will be presented later.

[0006] A machine learning (ML) model may be used to perform channel state information (CSI) compression. For example, a user equipment (UE) may receive a channel state information-reference signal (CSI-RS) from a network entity (such as a base station), and the UE may measure the channel state information-reference signal (CSI-RS) to estimate a channel associated with the CSI-RS. The UE may calculate an eigenvector for the channel in each subband such that the eigenvector may be input to an ML model to output a compressed eigenvector for a CSI report transmitted to the network entity. In an example, the first v columns of the eigenvector for the average channel associated with each subband may be used as input for the CSI compression. The UE reports the compressed CSI in a CSI report to the network entity, which decodes the CSI report to determine the compressed eigenvector, and then decompresses the compressed eigenvector to reconstruct the uncompressed eigenvector calculated by the UE for the channel in each subband.

[0007] The first v eigenvectors can be calculated based on a singular vector decomposition (SVD) technique of the average channel associated with each subband. However, calculating the increased number of eigenvectors for compression into the CSI report can result in increased complexity for the UE. That is, in some examples, performing eigenvector calculations for each subband can be computationally expensive, which can result in increased overhead for the UE. As the number of CSI-RS transmission ports increases, the SVD complexity also increases.

[0008] Aspects of the present disclosure address the above and other deficiencies by implementing techniques that reduce the UE complexity of ML-based CSI compression by reducing the complexity associated with eigenvector calculation. For example, the UE can select a wideband precoder to calculate eigenvectors and input the eigenvectors associated with the wideband precoder into the ML model for CSI compression. Compared to calculating / compressing eigenvectors for each subband, wideband precoder-based eigenvector compression reduces UE overhead / complexity.

[0009] According to some aspects, a UE receives a CSI-RS for channel estimation from a network entity. The UE sends a CSI report to the network entity that includes a first precoding matrix indicator (PMI) and a second PMI. The first PMI indicates a wideband precoder associated with the channel estimation, and the second PMI indicates a compressed subband eigenvector associated with the wideband precoder.

[0010] According to some aspects, a network entity transmits a CSI-RS to a UE to generate a channel estimate based on a CSI report. The network entity receives a CSI report from the UE including a first PMI and a second PMI. The first PMI indicates a wideband precoder associated with the channel estimate, and the second PMI indicates a compressed subband eigenvector associated with the wideband precoder. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A diagram illustrating a wireless communication system including multiple user equipments (UEs) and network entities communicating through one or more cells is shown.

[0012] Figure 2 Diagram showing an example machine learning (ML) based channel state information (CSI) encoder compression at a UE and an example ML based CSI decoder decompression at a network entity.

[0013] Figure 3 An example of a diagram showing low-complexity ML based CSI compression.

[0014] Figure 4 is a signaling diagram illustrating an example of a low-complexity ML based CSI reporting procedure.

[0015] Figure 5 is a flow chart of an example method for wireless communication at a UE for CSI compression and reporting.

[0016] Figure 6 is a flow chart of an example method for wireless communication for CSI decompression for low-complexity CSI reporting at a network entity.

[0017] 7A to 7D Table showing example ML-based CSI reporting.

[0018] Figure 8 is a diagram illustrating an example of a hardware implementation of an example UE device.

[0019] Figure 9 is a diagram illustrating an example of a hardware implementation of one or more example network entities. DETAILED DESCRIPTION

[0020] Figure 1Figure 100 shows a wireless communication system associated with multiple cells 190. The wireless communication system includes a user equipment (UE) 102 and a base station / network entity 104. Some base stations may include a converged base station architecture, while other base stations may include a disaggregated base station architecture. A converged base station architecture utilizes a radio protocol stack that is physically or logically integrated within a single radio access network (RAN) node. A disaggregated base station architecture utilizes a protocol stack that is physically or logically distributed between two or more units (e.g., a radio unit (RU) 106, a distributed unit (DU) 108, a central unit (CU) 110). For example, the CU 110 is implemented within a RAN node, and one or more DUs 108 may be co-located with the CU 110, or alternatively, may be geographically or virtually distributed in one or more other RAN nodes. The DU 108 may be implemented to communicate with one or more RUs 106. Any of the RU 106, DU 108, and CU 110 may be implemented as a virtual unit, such as a virtual radio unit (VRU), a virtual distributed unit (VDU), or a virtual central unit (VCU). The base station / network entity 104 (e.g., a converged base station or a disaggregated unit of a base station, such as the RU 106 or DU 108) may be referred to as a transmission reception point (TRP).

[0021] The operation and / or network design of the base station 104 can be based on the aggregated nature of base station functionality. For example, a disaggregated base station architecture is utilized in an integrated access backhaul (IAB) network, an open radio access network (O-RAN) network, or a virtualized radio access network (vRAN) (which may also be referred to as a cloud radio access network (C-RAN)). Disaggregation can include distributing functionality between two or more units located at various physical locations, as well as virtually distributing the functionality of at least one unit, which can enable flexibility in network design. Various units of a disaggregated base station architecture or disaggregated RAN architecture can be configured to communicate with at least one other unit in wired or wireless communication. For example, the base station 104d / 104e and / or RUs 106a-106d can communicate with UEs 102a-102d and 102s via one or more radio frequency (RF) access links based on a Uu interface. In an example, multiple RUs 106 and / or base stations 104 can simultaneously serve a UE 102, such as via intra-cell and / or inter-cell access links between the UE 102 and the RUs 106 / base stations 104.

[0022] The RU 106, DU 108, and CU 110 may include (or may be coupled to) one or more interfaces configured to transmit or receive information / signals via a wired or wireless transmission medium. For example, a wired interface may be configured to transmit or receive information / signals via a wired transmission medium, such as via a fronthaul link 160 between the RU 106d and a baseband unit (BBU) 112 of the base station 104d associated with the cell 190d. The BBU 112 includes the DU 108 and the CU 110, which may also have a wired interface (e.g., a mid-range link) configured between the DU 108 and the CU 110 to transmit or receive information / signals between the DU 108 and the CU 110. In a further example, a wireless interface, which may include a receiver, transmitter, or transceiver (such as an RF transceiver), is configured to transmit and / or receive information / signals via a wireless transmission medium, such as information transmitted between RU 106a of cell 190a and base station 104e of cell 190e via cross-cell communication beams 136-138 of RU106a and base station 104e.

[0023] The RU 106 may be configured to implement lower layer functions. For example, the RU 106 is controlled by the DU 108 and may correspond to a logical node that hosts RF processing functions or lower layer PHY functions, such as performing fast Fourier transform (FFT), inverse FFT (iFFT), digital beamforming, physical random access channel (PRACH) extraction and filtering, etc. The functions of the RU 106 may be based on functional partitioning, such as lower layer functional partitioning.

[0024] The RU 106 may transmit or receive over-the-air (OTA) communications with one or more UEs 102. For example, the RU 106b of the cell 190b communicates with the UE 102b of the cell 190b via the first communication beam set 132 of the RU 106b and the second communication beam set 134b of the UE 102b, which may correspond to inter-cell communication beams or, in some examples, cross-cell communication beams. For example, the UE 102b of the cell 190b may communicate with the RU 106a of the cell 190a via the third communication beam set 134a of the UE 102b and the fourth communication beam set 136 of the RU 106a. The DU 108 may control the real-time and non-real-time characteristics of the control plane and user plane communications of the RU 106.

[0025] Any combination of RU 106, DU 108, and CU 110, or any reference thereto individually, may correspond to base station 104. Thus, base station 104 may include at least one of RU 106, DU 108, or CU 110. Base station 104 provides UE 102 with access to a core network. Base station 104 may relay communications between UE 102 and a core network (not shown). Base station 104 may be associated with a macro cell of a high-power cellular base station and / or a small cell of a low-power cellular base station. For example, cell 190e may correspond to a macro cell, while cells 190a-190d may correspond to small cells. Small cells include femto cells, pico cells, micro cells, and the like. A network including at least one macro cell and at least one small cell may be referred to as a "heterogeneous network."

[0026] Transmissions from a UE 102 to a base station 104 / RU 106 are referred to as uplink (UL) transmissions, while transmissions from a base station 104 / RU 106 to a UE 102 are referred to as downlink (DL) transmissions. Uplink transmissions may also be referred to as reverse link transmissions, while downlink transmissions may also be referred to as forward link transmissions. For example, RU 106 d utilizes an antenna of base station 104 d in cell 190 d to transmit downlink / forward link communications to UE 102 d or receive uplink / reverse link communications from UE 102 d over a Uu interface associated with an access link between UE 102 d and base station 104 d / RU 106 d.

[0027] The communication link between the UE 102 and the base station 104 / RU 106 can be based on multiple-input multiple-output (MIMO) antenna technology, including spatial multiplexing, beamforming, and / or transmit diversity. The communication link can be associated with one or more carriers. The UE 102 and the base station 104 / RU 106 can utilize a spectrum bandwidth of Y MHz (e.g., 5 MHz, 10 MHz, 15 MHz, 20 MHz, 100 MHz, 400 MHz, 800 MHz, 1600 MHz, 2000 MHz, etc.) per carrier, allocated in a carrier aggregation of up to a total of Yx MHz, with x component carriers (CCs) used for communication in each of the uplink and downlink directions. The carriers may or may not be adjacent to each other along the spectrum. In an example, uplink and downlink carriers can be allocated in an asymmetric manner, with more or fewer carriers allocated for the uplink or downlink. The component carriers can include a primary component carrier and one or more secondary component carriers. The primary component carrier may be associated with a primary cell (PCell), and the secondary component carrier may be associated with a secondary cell (SCell).

[0028] Some UEs 102 (such as UEs 102a and 102s) can perform device-to-device (D2D) communication via a sidelink. For example, the sidelink communication / D2D link utilizes the spectrum of a wireless wide area network (WWAN) associated with uplink and downlink communications. Such sidelink / D2D communication can be performed via various wireless communication systems, such as Wireless Fidelity (Wi-Fi) systems, Bluetooth systems, Long Term Evolution (LTE) systems, New Radio (NR) systems, and the like.

[0029] The electromagnetic spectrum is typically subdivided into different categories, bands, channels, etc., based on the different frequencies / wavelengths associated with the electromagnetic spectrum. Fifth-generation (5G) NR is typically associated with two operating frequency bands (FRs), referred to as Frequency Range 1 (FR1) and Frequency Range 2 (FR2). FR1 ranges from 410 MHz to 7.125 GHz, and FR2 ranges from 24.25 GHz to 71.0 GHz, including FR2-1 (24.25 GHz to 52.6 GHz) and FR2-2 (52.6 GHz to 71.0 GHz). Although a portion of FR1 is actually greater than 6 GHz, FR1 is often referred to as the "sub-6 GHz" band. In contrast, FR2 is often referred to as the "millimeter wave" (mmW) band. FR2 is distinct from the "extremely high frequency" (EHF) band, but is a close subset of it. The EHF band ranges from 30 GHz to 300 GHz and is sometimes also referred to as the "millimeter wave" band. Frequencies between FR1 and FR2 are often referred to as "mid-band" frequencies. The operating band for mid-band frequencies may be referred to as Frequency Range 3 (FR3), which ranges from 7.125 GHz to 24.25 GHz. Frequency bands within FR3 may include characteristics of FR1 and / or FR2. Thus, the features of FR1 and / or FR2 may be extended to mid-band frequencies. Higher operating bands have been identified to extend 5G NR communications above the 52.6 GHz associated with the upper limit of FR2. Three of these higher operating bands include FR2-2, which ranges from 52.6 GHz to 71.0 GHz, FR4, which ranges from 71.0 GHz to 114.25 GHz, and FR5, which ranges from 114.25 GHz to 300 GHz. The upper limit of FR5 corresponds to the upper limit of the EHF band. Therefore, unless otherwise specified herein, the term "sub-6 GHz" may refer to frequencies less than 6 GHz, frequencies within FR1, or frequencies that may include mid-band frequencies. Further, unless otherwise expressly stated herein, the term "millimeter wave" or mmW refers to frequencies that may include mid-band frequencies, frequencies that may be within FR2-1, FR4, FR2-2, and / or FR5, or frequencies that may be within the EHF band.

[0030] UE 102 and base station 104 / RU 106 may each include multiple antennas. The multiple antennas may correspond to antenna elements, antenna panels, and / or antenna arrays that may facilitate beamforming operations. For example, RU 106b transmits downlink beamformed signals to UE 102b based on a first communication beam set 132 in one or more transmit directions of RU 106b. UE 102b may receive downlink beamformed signals from RU 106b based on a second communication beam set 134b in one or more receive directions of UE 102b. In a further example, UE 102b may also transmit uplink beamformed signals (e.g., sounding reference signals (SRS)) to RU 106b based on the second communication beam set 134b in one or more transmit directions of UE 102b. RU 106b may receive uplink beamformed signals from UE 102b in one or more receive directions of RU 106b.

[0031] UE 102b may perform beam training to determine optimal receive and transmit directions for beamformed signals. The transmit and receive directions of UE 102 and base station 104 / RU 106 may be the same or different. In a further example, beamformed signals may be transmitted between a first base station / RU 106a and a second base station 104e. For example, base station 104e of cell 190e may transmit beamformed signals to RU 106a based on communication beam 138 in one or more transmit directions of base station 104e. RU 106a may receive beamformed signals from base station 104e of cell 190e based on RU communication beam 136 in one or more receive directions of RU 106a. In a further example, base station 104e transmits downlink beamformed signals to UE 102e based on communication beam 138 in one or more transmit directions of base station 104e. The UE 102e may receive downlink beamformed signals from the base station 104e in one or more receive directions of the UE 102e based on the UE communication beam 130. The UE 102e may also transmit uplink beamformed signals to the base station 104e in one or more transmit directions of the UE 102e based on the UE communication beam 130, so that the base station 104e may receive the uplink beamformed signals from the UE 102e in one or more receive directions of the base station 104e.

[0032] The base station 104 may include and / or be referred to as a network entity. That is, a “network entity” may refer to the base station 104 or at least one unit of the base station 104, such as the RU 106, the DU 108, and / or the CU 110. The base station 104 may also include and / or be referred to as a next generation evolved Node B (ng-eNB), a next generation NB (gNB), an evolved NB (eNB), an access point, a base transceiver station, a radio base station, a radio transceiver, a transceiver function, a basic service set (BSS), an extended service set (ESS), a TRP, a network node, a network device, or other related terms. The base station 104 or an entity at the base station 104 may be implemented as an IAB node, a relay node, a sidelink node, a converged (monolithic) base station, or a decomposed base station including one or more of the RU 106, the DU 108, and / or the CU 110. A converged or decomposed set of base stations may be referred to as a next generation radio access network (NG-RAN). In some examples, UE 102a operates in dual connectivity (DC) with base station 104e and base station / RU 106a. In such a case, base station 104e may be the primary node, while base station / RU 160a may be the secondary node.

[0033] Uplink / downlink signaling may also be communicated via a satellite positioning system (SPS) 114. In an example, the SPS 114 of cell 190c may communicate with one or more UEs 102 (such as UE 102c) and one or more base stations 104 / RUs 106 (such as RU 106c). The SPS 114 may correspond to one or more of a global navigation satellite system (GNSS), a global positioning system (GPS), a non-terrestrial network (NTN), or other satellite positioning / location systems. The SPS 114 may be associated with LTE signals, NR signals (e.g., based on round-trip time (RTT) and / or multiple RTTs), wireless local area network (WLAN) signals, a terrestrial beacon system (TBS), sensor-based information, NR enhanced cell ID (NR E-CID) technology, downlink angle of departure (DL-AoD), downlink time difference of arrival (DL-TDOA), uplink time difference of arrival (UL-TDOA), uplink angle of arrival (UL-AoA), and / or other systems, signals, or sensors.

[0034] Still refer to Figure 1In certain aspects, the UE 102 may include a UE-based channel state information (CSI) processing component 140 configured to receive a channel state information-reference signal (CSI-RS) for channel estimation from a network entity; and send a CSI report to the network entity including a first precoding matrix indicator (PMI) and a second PMI, wherein the first PMI indicates a wideband precoder associated with the channel estimation and the second PMI indicates a compressed subband eigenvector associated with the wideband precoder.

[0035] In certain aspects, the base station 104 or a network entity of the base station 104 may include a network-based CSI processing component 150 configured to transmit a CSI-RS to a UE to generate a channel estimate based on a CSI report; and receive a CSI report from the UE including a first PMI and a second PMI, the first PMI indicating a wideband precoder associated with the channel estimate, and the second PMI indicating a compressed subband eigenvector associated with the wideband precoder. Although the following description may focus on 5G NR, the concepts described herein may be applicable to other similar areas, such as 5G-Advanced and future releases, LTE, LTE-Advanced (LTE-A), and other wireless technologies such as 6G.

[0036] Figure 2 Figure 200 illustrates an example machine learning (ML)-based CSI encoder compression at UE 102 and an example ML-based CSI decoder decompression at network entity 104. In a MIMO system, network entity 104 can use CSI to select a digital precoder for UE 102. Network entity 104 can configure CSI reporting 285 via RRC signaling (e.g., CSI-reportConfig), where UE 102 uses channel measurement resources (CMRs) to measure CSI-RS 240 to estimate 250 the downlink channel. Network entity 104 can also configure (e.g., via CSI-reportConfig) interference measurement resources (IMRs) for UE 102 to measure interference. Based on the CMRs and IMRs, UE 102 can identify CSI, which may include a rank indicator (RI), a precoding matrix indicator (PMI), a channel quality indicator (CQI), and / or a layer indicator (LI). RI and PMI are used to determine the digital precoder (also known as the precoding matrix), and CQI indicates the signal-to-interference-plus-noise (SINR), which is used to determine the transmitter's selection of the modulation and coding scheme (MCS). LI is used to identify the strongest layer, such as for multi-user (MU)-MIMO pairing with low-rank transmission and precoder selection, and phase tracking reference signal (PT-RS) precoder selection.

[0037] The UE 102 may indicate a CSI report 285 in two parts via a physical uplink control channel (PUCCH) / physical uplink shared channel (PUSCH), where CSI Part 1 may include the RI and CQI of the first transport block (TB), and CSI Part 2 may include the PMI, LI, and CQI of the second TB. The payload size of CSI Part 2 may be based on CSI Part 1, and both parts may be transmitted to the network entity 104 using separate channel coding operations.

[0038] The network entity 104 may configure the time domain behavior (e.g., periodic, semi-persistent, or aperiodic reporting) for CSI reporting 285 in CSI-reportConfig. The network entity 104 may activate or deactivate semi-persistent CSI reporting via a MAC control element (MAC CE). The network entity 104 may also trigger aperiodic CSI reporting via downlink control information (DCI). The UE 102 may report periodic CSI on the PUCCH resources configured in CSI-reportConfig. The UE 102 may report semi-persistent CSI on the PUCCH resources configured in CSI-reportConfig or on PUSCH resources triggered by DCI from the network entity 104. The UE 102 may report aperiodic CSI on PUSCH resources triggered by DCI from the network entity 104.

[0039] In resource element k of CSI-RS 240, the received signal in the frequency domain can be obtained as follows: in Indicates the effective channel including the simulated beamforming weights, whose dimension is N Rx × N Tx , Indicates the CSI-RS 240 at resource element k, Indicates interference plus noise, N Rx is the number of receiving ports, and N Tx Indicates the number of transport ports.

[0040] In resource element k of the physical downlink shared channel (PDSCH), the received signal in the frequency domain may correspond to: in Indicates the precoder. Typically, the precoder should be the same for subcarriers within a subband (e.g., bundled physical resource blocks (PRBs)).

[0041] The UE 102 may measure and report CSI using a Type 2 CSI codebook, where the precoder performs quantization based on: in Corresponding to dimension N Tx × 2L wideband precoder; corresponds to a subband precoder of dimension 2L × v; L corresponds to the number of beams; and v corresponds to the number of layers, which can be RI+1.

[0042] Can be quantized based on codebook, and The quantization can be done based on the power and angle of each element, which may result in large overhead because is based on subbands, and the CSI report 285 may have multiple subbands, which may be determined based on the bandwidth of the CSI-RS 240. In an example, the code selected by W1 may correspond to: in, represents the Kronecker product; L indicates the number of beams configured by RRC signaling; N1, N2, O1, and O2 correspond to the number of ports and oversampling factors in the horizontal and vertical domains that can be configured via RRC signaling. The candidate value can be based on the number of CSI-RS ports. The codebook includes precoders with different values ​​of m and n. In this example, the candidate values ​​are based on a predefined protocol.

[0043] ML is an example technique that UE 102 may implement to perform CSI compression 270a, where the first v columns of the eigenvector of the average channel for each subband may be used as input. As used herein, unless expressly stated otherwise, the terms "machine learning" and "artificial intelligence" may be used interchangeably.

[0044] Diagram 200 illustrates an example of ML-based CSI compression after UE 102 receives CSI-RS 240 from network entity 104. UE 102 may perform channel estimation 250 based on CSI-RS 240 and calculate 260a the eigenvectors of the channel in each subband. The eigenvectors may be input to a neural network for CSI encoder compression 270a. UE 102 transmits 280a a compressed CSI report 285 to network entity 104.

[0045] The network entity 104 performs CSI report detection 280b on the CSI report transmission 280a from the UE 102. The neural network at the network entity 104 decodes the compressed CSI report 285 via CSI decoder decompression 270b to recover the eigenvectors. The network entity 104 selects 260b a precoder for each subband based on the reported eigenvectors.

[0046] For each subband, the eigenvector V can be derived based on the singular vector decomposition (SVD) of the average channel in the subband as follows: Where N indicates the number of CSI-RS resource elements of subband S; indicates the estimated channel based on the CSI-RS 240 at resource element k.

[0047] ML-based CSI compression techniques may refer to the following terms:

[0048] Data collection refers to the process by which a network node, management entity, or UE 102 collects data for ML model training, data analysis, and inference.

[0049] An ML model is a data-driven algorithm that applies ML techniques to generate an output set based on an input set.

[0050] ML model training refers to the process of training an ML model in a data-driven manner (e.g., by learning input / output relationships) to obtain the trained ML model for inference.

[0051] ML model inference refers to the process of using a trained ML model to generate an output set based on an input set.

[0052] ML model validation refers to a sub-process of ML model training that uses a dataset different from the training dataset used for model training to evaluate the quality of the ML model. The different data can be used to select model parameters that generalize to data outside the dataset used for ML model training.

[0053] ML model testing refers to a sub-process of ML model training that uses a dataset different from the training dataset used for ML model training and validation to evaluate the performance of the trained ML model. Unlike ML model validation, testing does not assume subsequent adjustments to the ML model.

[0054] A UE-side ML model refers to an ML model whose inference is performed at the UE 102 .

[0055] A network-side ML model refers to an ML model whose inference is performed at the network / network entity 104 .

[0056] A one-sided ML model refers to a UE-side ML model or a network-side ML model.

[0057] A two-sided ML model refers to a paired ML model on which joint reasoning is performed, where the joint reasoning includes ML reasoning performed jointly across the UE 102 and the network entity 104 (e.g., a first portion of the reasoning is performed by the UE 102 and the remaining portion of the reasoning is performed by the network entity 104, or vice versa).

[0058] ML model delivery refers to delivering an ML model over the air interface based on the parameters of the model structure known at the receiving end or a new model with parameters. Delivery techniques can include the delivery of a complete ML model or a partial ML model.

[0059] Model download refers to the transfer of an ML model from the network entity 104 to the UE 102 .

[0060] Model upload refers to the transfer of ML models from the UE 102 to the network entity 104 .

[0061] Federated learning / federated training refers to a machine learning technique that trains ML models across multiple distributed edge nodes (e.g., UEs, network entities, etc.), with each edge node performing local model training using local data samples. Federated learning / training can be based on multiple interactions with the ML model without exchanging local data samples.

[0062] Offline field data refers to data collected from the field and used for offline training of ML models.

[0063] Online field data refers to data collected from the field and used for online training of ML models.

[0064] Model monitoring refers to the process used to monitor the inference performance of ML models.

[0065] Supervised learning refers to the process of training a model based on inputs and corresponding labels.

[0066] Unsupervised learning refers to the process of training a model without the need for labeled data.

[0067] Semi-supervised learning refers to the process of training a model based on a mixture of labeled and unlabeled data.

[0068] Reinforcement learning (RL) is the process of training an ML model based on inputs (or states) and feedback signals (or rewards) generated from the model’s outputs (or actions) in the environment with which the model interacts.

[0069] Model activation refers to enabling an ML model for a specific feature.

[0070] Model deactivation refers to disabling an ML model for a specific feature.

[0071] Model switching refers to deactivating the currently active ML model and activating a different ML model for a specific function.

[0072] The UE complexity resulting from the eigenvector calculation may be associated with an increase in the UE's processing power. For each subband, the calculation of the first eigenvector V1 may be based on a singular vector decomposition (SVD) of the average channel, which may correspond to: where N indicates the number of CSI-RS resource elements for subband S, and Corresponds to the estimated channel based on the CSI-RS at resource element k. In some examples, the number of transmission ports of the CSI-RS may be large (e.g., N Tx =32). Therefore, the complexity of SVD may be high. Therefore, a method for reducing the UE complexity of ML-based CSI compression is proposed, including reducing the eigenvector calculation complexity.

[0073] Figure 3 FIG300 shows CSI compression based on low complexity ML. Elements 240, 250, 260b, 270a-270b, 280a-280b and 285 have been described with respect to Figure 2 Provide a description.

[0074] After the UE 102 estimates 250 the channel based on the received CSI-RS 240, the UE 102 may select 355 a wideband precoder W1 for CSI compression. In an example, the codebook for wideband precoder W1 selection 355 may be based on: in, Indicates the Kronecker product; L indicates the number of beams configured by RRC signaling; N1, N2, O1, and O2 correspond to the number of ports and oversampling factors in the horizontal and vertical domains that can be configured by RRC signaling. The candidate values ​​of the codebook can be based on the number of CSI-RS ports. The codebook can include precoders with different values ​​of m and n.

[0075] The UE 102 performs eigenvector calculation 365a based on the channel estimate and the wideband precoder for each subband. The dimension can correspond to N Rx× 2L, where L is configured by the RRC parameters. In an example, L can be equal to 2, so that the dimension of the input matrix of the SVD can be much smaller than the input of the neural network in Figure 200. For example, in Figure 300, the input of the neural network for CSI encoder compression 270a can be determined based on the following:

[0076] The first v columns of the second eigenvector V2 can be input to the neural network, where v indicates the number of layers. W1 corresponds to a wideband precoder, which can be quantized based on a predefined codebook or a codebook configured by the network entity 104 (e.g., a Type 1 or Type 2 codebook). After the compressed CSI encoder is decompressed by the neural network for CSI decoder decompression 270b at the network entity 104, the network entity 104 performs a similar eigenvector calculation 365b for each subband, as performed 365a at the UE 102, to determine the precoder selection 260b for each subband based on the reported eigenvectors.

[0077] Figure 4 4 is a signaling diagram illustrating a low-complexity ML-based CSI reporting procedure. The network entity 104 transmits 402 a configuration for ML-based CSI reporting (e.g., CSI-reportConfig) to the UE 102 based on RRC signaling. When the codebookType in the CSI-reportConfig is set to a first specific value (e.g., 'ai-M1' or 'type3'), or the reportQuantity in the CSI-reportConfig is set to a second specific value (e.g., 'ri-compressedPmi-cqi'), the CSI report may be ML-based. The network entity 104 may also transmit 402 a configuration for non-ML-based CSI reporting, which may be based on a specific codebook (e.g., Type 1 or Type 2 codebook).

[0078] The network entity 104 may transmit 404 to the UE 102 a trigger indication for triggering CSI reporting from the UE 102. However, for periodic CSI reporting, the network entity 104 may skip transmitting 404 the trigger indication because the UE 102 may report periodic CSI reporting via uplink resource configuration 402 from the network entity 104 via RRC signaling. The trigger indication for semi-persistent and aperiodic CSI reporting may be a MAC-CE or a DCI. For example, the network entity 104 may transmit 404 a MAC-CE to activate semi-persistent CSI reporting. In other examples, the network entity 104 may transmit 404 a DCI to trigger aperiodic CSI reporting.

[0079] The UE 102 receives 240 a CSI-RS associated with the triggered or configured CSI report from the network entity 104. The UE 102 may perform 470a CSI measurement and compression based on the CSI-RS reception 240. For example, the UE 102 may compress the CSI based on a low-complexity technique for ML-based compression, as shown in diagram 300, where, compared to diagram 200 (e.g., the first eigenvector (eigenvector V1) is used as input to the ML model), the second eigenvector (eigenvector V2) is used as input to the ML model.

[0080] After the UE 102 performs 470a CSI measurement and compression, the UE 102 may transmit 285 the compressed CSI to the network entity 104. That is, the UE 102 may transmit 285 a low-complexity CSI report to the network entity 104 based on the CSI compression. The network entity 104 receives 285 the low-complexity CSI report from the UE 102 and decompresses 470b the CSI based on the ML model. The network entity 104 may transmit 495 a PDSCH to the UE 102 based on the CSI decompression 470b (e.g., based on the wideband precoder and the uncompressed / reconstructed subband eigenvectors). Figures 5 and 6 Shown for implementation Figures 2 to 4 Specifically, Figure 5 UE 102 is shown Figures 2 to 4 Implementation of one or more aspects. Figure 6 The network entity 104 is shown Figures 2 to 4 Implementation of one or more aspects.

[0081] Figure 5 A flow chart 500 is shown of a method of wireless communication at a UE. Figures 1 to 4 and Figure 8 , the method can be performed by UE 102, UE device 802, etc., which may include memory 826', 806', 816 and may correspond to the entire UE 102 or the entire UE device 802, or components of the UE 102 or UE device 802, such as the wireless baseband processor 826 and / or the application processor 806.

[0082] The UE 102 receives 502 a configuration from a network entity indicating at least one of: an ML model for compressing subband feature vectors, a codebook for a wideband precoder, or a number of beams for a wideband precoder. Figure 4 , the UE 102 receives 402 a configuration of ML-based CSI reporting from the network entity 104 .

[0083] UE 102 receives 504 a trigger indication of a CSI report from a network entity, the CSI report including a first PMI for a wideband precoder and a second PMI for a compressed subband eigenvector. Figure 4 , the UE 102 receives 404 a trigger indication for CSI reporting from the network entity 104 .

[0084] UE 102 receives 540 CSI-RS for channel estimation from the network entity—a wideband precoder is associated with the channel estimation and a subband eigenvector is compressed into a compressed subband eigenvector associated with the wideband precoder. Figures 2 to 4 , the UE 102 receives 240 CSI-RS from the network entity 104 to estimate 250 the channel to then select 355 a wideband precoder, calculate 365a subband eigenvectors, and compress 270a the subband eigenvectors using the wideband precoder.

[0085] UE 102 calculates 565a the subband eigenvector based on the wideband precoder and channel estimation. Figure 3 , UE 102 calculates 365a an eigenvector based on the channel and the wideband precoder for each subband.

[0086] UE 102 compresses 570a the subband eigenvector using a wideband precoder to generate a compressed subband eigenvector. Figures 2 to 4 , UE 102 performs 270a / 470a CSI encoder compression using a neural network.

[0087] UE 102 sends 585 a CSI report to the network entity, the CSI report including a first PMI indicating a wideband precoder and a second PMI indicating a compressed subband eigenvector. Figures 2 to 4 , the UE 102 sends 285 a CSI report to the network entity 104 .

[0088] The input to the neural network at the UE 102 for CSI encoder compression 270a may be based on the selected wideband precoder 355 and the channel 250 estimated from the CSI-RS 240. In an example, the input may be based on the matrix The first v columns of the second eigenvector V2 calculated by the SVD of W1 are obtained. The number of beams for W1 and the codebook to search can be configured 402 based on higher layer signaling (e.g., RRC signaling). The codebook can correspond to a Type 1 or Type 2 CSI codebook. In an example, the codebook used for precoder W1 selection 355 can be based on: in, Indicates the Kronecker product; L indicates the number of beams that can be configured by RRC signaling (e.g., numberOfBeams); N1, N2, O1, and O2 correspond to the number of ports and oversampling factors that can be configured by RRC signaling (e.g., n1-n2) in the horizontal and vertical domains. The candidate values ​​of the codebook can be based on the number of CSI-RS ports. The codebook can include precoders with different values ​​of m and n.

[0089] Figure 6 is a flow chart 600 of a method of wireless communication at a network entity. Figures 1 to 4 and Figure 9 The method may be performed by one or more network entities 104, which may correspond to a base station or a unit of a base station, such as RU 106, DU 108, CU 110, RU processor 906, DU processor 926, CU processor 946, etc. The one or more network entities 104 may include a memory 906' / 926' / 946', which may correspond to the entirety of the one or more network entities 104, or a component of the one or more network entities 104, such as RU processor 906, DU processor 926, or CU processor 946.

[0090] The network entity 104 transmits 602 a configuration to the UE, the configuration indicating at least one of the following: an ML model, a codebook for a wideband precoder, or a number of beams for a wideband precoder. Figure 4 , the network entity 104 transmits 402 to the UE 102 a configuration of ML-based CSI reporting.

[0091] The network entity 104 transmits 604 a trigger indication of a CSI report to the UE, the CSI report including a first PMI for the wideband precoder and a second PMI for the compressed subband eigenvector. Figure 4 , the network entity 104 transmits 404 a trigger indication of a CSI report to the UE 102.

[0092] The network entity 104 transmits 640 a CSI-RS to the UE to generate a channel estimate based on the CSI feedback in the CSI report. Figures 2 to 4 , the network entity 104 transmits 240 a CSI-RS to the UE 102 to estimate the channel based on the CSI feedback in the CSI report 285 .

[0093] The network entity 104 receives 685 a CSI report from the UE, the CSI report including a first PMI indicating a wideband precoder associated with the channel estimate and a second PMI indicating a compressed subband eigenvector associated with the wideband precoder. Figures 2 to 4 , the network entity 104 receives 285 a CSI report based on the low complexity technique for CSI compression from the UE 102 .

[0094] The network entity 104 decompresses 670b the compressed sub-band feature vector to reconstruct the uncompressed sub-band feature vector. Figures 2 to 4 , the network entity 104 performs 270b / 470b CSI decoder decompression using a neural network.

[0095] The network entity 104 applies 690 the uncompressed subband eigenvectors to a wideband precoder to generate a channel estimate associated with transmitting the CSI-RS to the UE. Figures 2 to 3 , the network entity 104 selects 260b a precoder for each subband based on the reported eigenvectors used to estimate the channel.

[0096] The network entity 104 transmits 695 a PDSCH signal to the UE based on the wideband precoder and the uncompressed subband eigenvector. Figure 4 , the network entity 104 transmits 495 the PDSCH to the UE 102 based on the CSI decompression 470b.

[0097] The ML model used for CSI encoder compression 270a can be configured 402 via higher layer signaling (e.g., RRC signaling) from the network entity 104, or can be predefined or preconfigured. Different ML models can be used for different codebooks. For example, a first ML model can be used for a codebook with a certain number of configured beams, while a second ML model can be used for a codebook with a configured number of beams N1, N2, O1, and O2. ML models can also be used for codebooks with a certain number of subbands and a certain number of configured beams N1, N2, O1, and O2.

[0098] The RRC configuration of the codebook for ML-based CSI reporting may be based on the UE capabilities of the codebook associated with ML-based CSI reporting reported by the UE 102 to the network entity 104. For ML-based CSI reporting, the UE 102 may report 285 a first precoder matrix index (PMI) of W1 and an output of the ML model for CSI encoder compression 270a, which may be indicated based on the compressed eigenvectors. In a further example, the UE 102 may report a second PMI for each subband to the network entity 104. In an example, the first PMI may be reported as two indices, including a first index indicating the horizontal beam index m and a second index indicating the vertical beam index n. The second PMI may correspond to the output of the ML model for CSI encoder compression 270a.

[0099] 7A to 7D Tables 700-730 illustrate example ML-based CSI reporting. Tables 700-730 include a CSI report number field that indicates CSI report #n and the CSI part associated with the ML-based CSI report (e.g., CSI part 1 or CSI part 2). The network entity uses CSI part 1 to decode CSI part 2, which carries channel estimation information. Tables 700-730 also include CSI fields that indicate information associated with CSI-RS resource indicator (CRI), RI, wideband CQI, subband (differential) CQI, PMI, etc.

[0100] For short PUCCH-based CSI reporting, PMI and other CSI may be reported in a single CSI part (eg, CSI part 1). Figure 7A Table 700 shows an example ML-based CSI reporting format associated with short PUCCH.

[0101] For CSI parts based on PUSCH or long PUCCH, both PMIs can be reported in CSI part 2, where the bit width of CSI part 2 is based on the information reported in CSI part 1 (e.g., CRI / RI and CQI of the first transport block (TB)). Figure 7B Table 710 shows an example ML-based CSI reporting format with long PUCCH and CSI part 2 in PUSCH.

[0102] For CSI reporting based on PUSCH or long PUCCH, both PMIs may be reported in CSI Part 2, or the first PMI may be reported in CSI Part 1 and the second PMI may be reported in CSI Part 2. Figure 7C Table 720 and Figure 7DTable 730 shows an example of an ML-based CSI reporting format with CSI part 1 and CSI part 2 in long PUCCH and PUSCH.

[0103] Low-complexity CSI compression may be enabled via RRC signaling from the network entity 104 (e.g., an RRC parameter in CSI-reportConfig) that may be based on a UE capability report. The codebook may have a codebookType configured as 'ML' or 'Type3'. The RRC signaling of the ML codebook may include at least one of the following: the number of beams (L), the number of ports in the horizontal and vertical domains (N1, N2), the number of oversampling factors in the horizontal and vertical domains (O1, O2), the number of subbands used for CSI compression, the number of subbands per CQI calculation, which indicates the number of subbands used for CSI compression for CQI calculation, an RI limit that may be used to indicate a candidate rank for precoder selection, or an ML model indicating an ML model for CSI compression. The subtype of low-complexity ML-based CSI compression may be indicated in the codebook configuration to indicate whether the ML-based CSI report is a low-complexity ML-based CSI report (e.g., type III-hybrid or other ML-based report).

[0104] Unless otherwise specified herein, the RRC signaling may indicate an RRC reconfiguration message from the network entity 104 to the UE 102, or a system information block (SIB), where the SIB may be a predefined SIB (e.g., SIB1) or a different SIB transmitted by the network entity 104. The network entity 104 may obtain the UE capabilities via UE capability report signaling or from the UE 102, another network entity / base station, or a core network entity (e.g., AMF).

[0105] Figure 8 8 is a diagram illustrating an example of a hardware implementation of a UE device 802. The UE device 802 may be the UE 102, a component of the UE 102, or may implement UE functionality. The UE device 802 may include an application processor 806, which may have on-chip memory 806′. In an example, the application processor 806 may be coupled to a secure digital (SD) card 808 and / or a display 810. The application processor 806 may also be coupled to a sensor module 812, a power supply 814, an additional memory module 816, a camera 818, and / or other related components. For example, the sensor module 812 may control a barometric pressure sensor / altimeter, a motion sensor (such as an inertial management unit (IMU)), a gyroscope, an accelerometer, a light detection and ranging (LIDAR) device, a radio-aided detection and ranging (RADAR) device, a sound navigation and ranging (SONAR) device, a magnetometer, an audio device, and / or other technologies for positioning.

[0106] The UE device 802 may further include a wireless baseband processor 826, which may be referred to as a modem. The wireless baseband processor 826 may have on-chip memory 826'. Like the application processor 806, the wireless baseband processor 826 may also be coupled to a sensor module 812, a power supply 814, an additional memory module 816, a camera 818, and / or other related components. The wireless baseband processor 826 may also be coupled to one or more subscriber identity module (SIM) cards 820 and / or one or more transceivers 830 (e.g., wireless RF transceivers).

[0107] Within one or more transceivers 830, the UE device 802 may include a Bluetooth module 832, a WLAN module 834, an SPS module 836 (e.g., a GNSS module), and / or a cellular module 838. The Bluetooth module 832, the WLAN module 834, the SPS module 836, and the cellular module 838 may each include an on-chip transceiver (TRX), or in some cases, include only a transmitter (TX) or only a receiver (RX). The Bluetooth module 832, the WLAN module 834, the SPS module 836, and the cellular module 838 may each include a dedicated antenna and / or utilize an antenna 840 to communicate with one or more other nodes. For example, the UE device 802 may communicate with another UE (e.g., sidelink communication) and / or communicate with a network entity 104 (e.g., uplink / downlink communication) via the transceiver 830 via the antenna 840, where the network entity 104 may correspond to a base station or a unit of a base station, such as a RU 106, a DU 108, or a CU 110.

[0108] The wireless baseband processor 826 and the application processor 806 may each include computer-readable media / memory 826′, 806′, respectively. Additional modules of the memory 816 may also be considered computer-readable media / memory. Each computer-readable medium / memory 826′, 806′, 816 may be non-transitory. The wireless baseband processor 826 and the application processor 806 may each be responsible for general processing, including executing software stored on the computer-readable media / memory 826′, 806′, 816. This software, when executed by the wireless baseband processor 826 / application processor 806, enables the wireless baseband processor 826 / application processor 806 to perform the various functions described herein. The computer-readable media / memory may also be used to store data manipulated by the wireless baseband processor 826 / application processor 806 when executing the software. The wireless baseband processor 826 / application processor 806 may be a component of the UE 102. The UE device 802 may be a processor chip (e.g., a modem and / or applications) and include only the wireless baseband processor 826 and / or application processor 806. In other examples, the UE device 802 may be the entire UE 102 and include additional modules of the device 802.

[0109] like Figure 1 Discussion and about Figure 5 In one embodiment, the UE-based CSI processing component 140 is configured to receive a CSI-RS for channel estimation from a network entity and to send a CSI report to the network entity including a first PMI and a second PMI, wherein the first PMI indicates a wideband precoder associated with the channel estimation and the second PMI indicates a compressed subband eigenvector associated with the wideband precoder. The UE-based CSI processing component 140 may be within the application processor 806 (e.g., at 140a), within the wireless baseband processor 826 (e.g., at 140b), or within both the application processor 806 and the wireless baseband processor 826. The UE-based CSI processing components 140a-140b may be one or more hardware components explicitly configured to perform the process / algorithm, implemented by one or more processors configured to perform the process / algorithm, stored in a computer-readable medium for implementation by one or more processors, or a combination thereof.

[0110] Figure 99 is a diagram illustrating an example of a hardware implementation of one or more network entities 104. The one or more network entities 104 may be a base station, a component of a base station, or may implement base station functionality. The one or more network entities 104 may include or may correspond to at least one of a RU 106, a DU 108, or a CU 110. The CU 110 may include a CU processor 946, which may have on-chip memory 946'. In some aspects, the CU 110 may further include an additional memory module 956 and / or a communication interface 948, both of which may be coupled to the CU processor 946. The CU 110 may communicate with the DU 108 via a medium-range link 162, such as an F1 interface between the communication interface 948 of the CU 110 and the communication interface 928 of the DU 108.

[0111] The DU 108 may include a DU processor 926, which may have on-chip memory 926'. In some aspects, the DU 108 may further include an additional memory module 936 and / or a communication interface 928, both of which may be coupled to the DU processor 926. The DU 108 may communicate with the RU 106 via a fronthaul link 160 between the communication interface 928 of the DU 108 and the communication interface 908 of the RU 106.

[0112] The RU 106 may include a RU processor 906, which may have on-chip memory 906'. In some aspects, the RU 106 may further include an additional memory module 916, a communication interface 908, and one or more transceivers 930, all of which may be coupled to the RU processor 906. The RU 106 may further include an antenna 940, which may be coupled to the one or more transceivers 930, such that the RU 106 may communicate with the UE 102 via the antenna 940 through the one or more transceivers 930.

[0113] The on-chip memories 906', 926', 946' and the additional memory modules 916, 936, 956 can each be considered a computer-readable medium / memory. Each computer-readable medium / memory can be non-transitory. Each of the processors 906, 926, 946 is responsible for general processing, including executing software stored on the computer-readable medium / memory. The software, when executed by the corresponding processor 906, 926, 946, causes the processor 906, 926, 946 to perform the various functions described herein. The computer-readable medium / memory can also be used to store data manipulated by the processors 906, 926, 946 when executing the software. In an example, the network-based CSI processing component 150 can be located at any one of the one or more network entities 104, such as at the CU 110; at both the CU 110 and the DU 108; at each of the CU 110, DU 108, and RU 106; at both the DU 108 and the RU 106; or at the RU 106.

[0114] like Figure 1 Discussion and about Figure 6 In one embodiment, a network-based CSI processing component 150 is configured to transmit a CSI-RS to a UE to generate a channel estimate from a CSI report; and receive a CSI report from the UE including a first PMI and a second PMI, wherein the first PMI indicates a wideband precoder associated with the channel estimate and the second PMI indicates a compressed subband eigenvector associated with the wideband precoder. The network-based CSI processing component 150 may be located within one or more processors of one or more network entities 104, such as the RU processor 906 (e.g., at 150a), the DU processor 926 (e.g., at 150b), and / or the CU processor 946 (e.g., at 150c). The network-based CSI processing components 150a-150c may be one or more hardware components explicitly configured to perform the described process / algorithm, implemented by one or more processors 906, 926, 946 configured to perform the described process / algorithm, stored on a computer-readable medium for implementation by one or more processors 906, 926, 946, or a combination thereof.

[0115] The specific order or hierarchy of the blocks in the processes and flow charts disclosed herein is illustrative of example methods. Therefore, the specific order or hierarchy of the blocks in the processes and flow charts may be rearranged. Some blocks may also be merged or deleted. The accompanying method claims present elements of the blocks in an example order and are not limited to the specific order or hierarchy presented in the claims, processes, and flow charts.

[0116] The detailed description set forth herein, in conjunction with the accompanying drawings, describes various configurations and does not represent the only configuration in which the concepts described herein may be practiced. The detailed description includes specific details to provide a comprehensive explanation of the various concepts. However, these concepts may be practiced without using these specific details. In some cases, well-known structures and components are shown in block diagram form to avoid obscuring such concepts.

[0117] Various aspects of wireless communication systems (such as telecommunication systems) are presented with reference to various apparatuses and methods. These apparatuses and methods are described in the detailed description that follows and are illustrated in the accompanying drawings by various blocks, components, circuits, processes, call flows, systems, algorithms, etc. (collectively, "elements"). These elements can be implemented using electronic hardware, computer software, or a combination thereof. Whether such elements are implemented as hardware or software depends on the specific application and design constraints imposed on the overall system.

[0118] Element, or any part of an element or any combination of elements can be implemented as a "processing system" including one or more processors. Examples of processors include microprocessors, microcontrollers, graphics processing units (GPUs), central processing units (CPUs), application processors, digital signal processors (DSPs), reduced instruction set computing (RISC) processors, systems on chip (SoCs), baseband processors, field programmable gate arrays (FPGAs), programmable logic devices (PLDs), state machines, gating logic, discrete hardware circuits, and other similar hardware configured to perform the various functions described throughout this disclosure. One or more processors in a processing system can execute software, which can be referred to as software, firmware, middleware, microcode, hardware description language, or other. Software should be broadly interpreted as meaning instructions, instruction sets, codes, code segments, program codes, programs, subroutines, software components, applications, software applications, software packages, routines, subroutines, objects, executables, execution threads, processes, functions, or any combination thereof.

[0119] If the functions described herein are implemented in software, these functions may be stored on a computer-readable medium (such as a non-transitory computer-readable storage medium) or encoded as one or more instructions or codes on the computer-readable medium. Computer-readable media include computer storage media and may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable ROM (EEPROM), optical disk storage devices, magnetic disk storage devices, other magnetic storage devices, combinations of these types of computer-readable media, or any other medium that can be used to store computer-executable code in the form of computer-accessible instructions or data structures. The storage medium can be any available medium that is accessible to the computer.

[0120] The various aspects, implementations, and / or use cases described herein can be implemented across many different platform types, devices, systems, shapes, sizes, and packaging arrangements. For example, the various aspects, implementations, and / or use cases can be generated via integrated chip implementations and other non-module component-based devices such as end-user devices, vehicles, communication devices, computing devices, industrial equipment, retail / procurement devices, medical devices, artificial intelligence (AI)-enabled devices, machine learning (ML)-enabled devices, and the like. The various aspects, implementations, and / or use cases can range from chip-level or modular components to non-modular or non-chip-level implementations, and further to aggregated, distributed, or original equipment manufacturer (OEM) devices or systems incorporating one or more of the techniques described herein.

[0121] Devices incorporating aspects and features described herein may also include additional components and features for implementing and practicing the aspects and features claimed and described. For example, the transmission and reception of wireless signals necessarily include many components for analog and digital purposes, such as hardware components, antennas, RF chains, power amplifiers, modulators, buffers, processors, interleavers, adders / summers, etc. The techniques described herein can be practiced in a wide variety of devices, chip-level components, systems, distributed arrangements, aggregated or disaggregated components, end-user devices, etc., in various configurations.

[0122] The description herein is provided to enable those skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects. Therefore, the claims are not limited to the various aspects described herein, but should be interpreted in view of the full scope of the disclosure consistent with the language of the claims.

[0123] Unless explicitly stated, references to singular elements do not mean "one and only one", but rather "one or more". Terms such as "if", "when..." and "at..." do not imply an immediate temporal relationship or reaction. That is, these phrases (e.g., "when...") do not imply immediate action in response to the occurrence of an action or during the occurrence of an action, but simply mean that if a certain condition is met, an action will occur, but no specific or immediate time constraint is required for the occurrence of the action. Unless explicitly stated otherwise, the term "some" refers to one or more. Combinations such as "at least one of A, B, or C" or "one or more of A, B, or C" include any combination of A, B, and / or C, such as A and B, A and C, B and C, or A and B and C, and may include multiple A, multiple B, and / or multiple C, or may include only A, only B, or only C. A set should be interpreted as a set of elements whose number of elements is one or more.

[0124] Structural equivalents and functional equivalents of the elements of various aspects described in the entire present disclosure that are known or later learned by those of ordinary skill in the art are expressly incorporated herein by reference and are covered by the claims. The words "module", "mechanism", "element", "device" and the like may not be substitutes for the word "component". Therefore, unless the phrase "component for ..." is used to expressly state the claim elements, any claim element shall not be interpreted as a means plus function. As used herein, the phrase "based on" should not be interpreted as a reference to a closed information set, one or more conditions, one or more factors, etc. In other words, unless explicitly stated differently, the phrase "based on A" (wherein "A" can be information, conditions, factors, etc.) should be interpreted as "at least based on A".

[0125] The following examples are illustrative only and may be combined with other examples or teachings described herein without limitation.

[0126] Example 1 is a method for wireless communication at a UE, comprising: receiving a CSI-RS for channel estimation from a network entity; and sending a CSI report including a first PMI and a second PMI to the network entity, the first PMI indicating a wideband precoder associated with the channel estimation, and the second PMI indicating a compressed subband eigenvector associated with the wideband precoder.

[0127] Example 2 can be combined with Example 1 and further include calculating a subband eigenvector based on the wideband precoder and the channel estimate.

[0128] Example 3 can be combined with any one of Examples 1 to 2, and further include compressing the subband feature vector using the wideband precoder to generate the compressed subband feature vector.

[0129] Example 4 can be combined with any one of Examples 1 to 3, and the compressing including the subband feature vector is based on using an ML model to generate the compressed subband feature vector.

[0130] Example 5 can be combined with Example 4 and include a predefined protocol indicating the ML model for the compression of the sub-band feature vector.

[0131] Example 6 can be combined with any one of Examples 1 to 4 and further includes receiving a configuration indicating at least one of the following from the network entity: the ML model for the compression of the subband feature vector, the codebook for the wideband precoder, or the number of beams for the wideband precoder.

[0132] Example 7 can be combined with any one of Examples 1 to 6, and further includes receiving a trigger indication for the CSI report from the network entity, the CSI report including the first PMI indicating the wideband precoder and the second PMI indicating the compressed subband eigenvector.

[0133] Example 8 can be combined with any one of Examples 1 to 7, and includes sending the CSI report further comprising: transmitting the CSI report to the network entity using a PUCCH format, the PUCCH format including the first PMI and the second PMI in the same CSI part of the PUCCH format.

[0134] Example 9 can be combined with any one of Examples 1 to 7, and includes that sending the CSI report further includes: transmitting the CSI report to the network entity using a PUCCH format, the PUCCH format including the first PMI and the second PMI in different CSI parts of the PUCCH format.

[0135] Example 10 can be combined with any one of Examples 1 to 7, and including that sending the CSI report further includes: transmitting the CSI report to the network entity using a PUSCH transmission, the PUSCH transmission including the first PMI and the second PMI in the same CSI part of the PUSCH transmission.

[0136] Example 11 can be combined with any one of Examples 1 to 7, and including that sending the CSI report further includes: transmitting the CSI report to the network entity using a PUSCH transmission, the PUSCH transmission including the first PMI and the second PMI in different CSI parts of the PUSCH transmission.

[0137] Example 12 is a method for wireless communication at a network entity, comprising: transmitting a CSI-RS to a UE to generate a channel estimate from a CSI report; and receiving the CSI report from the UE including a first PMI and a second PMI, the first PMI indicating a wideband precoder associated with the channel estimate, and the second PMI indicating a compressed subband feature vector associated with the wideband precoder.

[0138] Example 13 can be combined with Example 12, and further include decompressing the compressed sub-band feature vector to reconstruct an uncompressed sub-band feature vector.

[0139] Example 14 can be combined with any one of Examples 12 to 13, and include that decompressing the compressed sub-band feature vector is based on reconstructing the uncompressed sub-band feature vector using an ML model.

[0140] Example 15 can be combined with Example 14, and include a predefined protocol indicating the ML model for decompressing the compressed sub-band feature vector.

[0141] Example 16 can be combined with any one of Examples 12 to 14, and further includes: transmitting to the UE a configuration indicating at least one of: the ML model, the codebook for the wideband precoder, or the number of beams for the wideband precoder.

[0142] Example 17 can be combined with any one of Examples 12 to 16, and further includes: transmitting a trigger indication for the CSI report to the UE, the CSI report including the first PMI for the wideband precoder and the second PMI for the compressed subband eigenvector.

[0143] Example 18 can be combined with any one of Examples 13 to 17, and further include: applying the uncompressed subband eigenvector to the wideband precoder to generate the channel estimate associated with transmitting the CSI-RS to the UE.

[0144] Example 19 can be combined with any one of Examples 12 to 18, and further include: transmitting a PDSCH signal to the UE based on the wideband precoder and the uncompressed subband eigenvector.

[0145] Example 20 can be combined with any one of Examples 12 to 19, and including receiving the CSI report further comprising: receiving the CSI report from the UE based on a PUCCH format, the PUCCH format including the first PMI and the second PMI in the same CSI part of the PUCCH format.

[0146] Example 21 can be combined with any one of Examples 12 to 19, and including receiving the CSI report further includes: receiving the CSI report from the UE based on a PUCCH format, the PUCCH format including the first PMI and the second PMI in different CSI parts of the PUCCH format.

[0147] Example 22 can be combined with any one of Examples 12 to 19, and including receiving the CSI report further comprising: receiving the CSI report from the UE via a PUSCH transmission, the PUSCH transmission including the first PMI and the second PMI in the same CSI part of the PUSCH transmission.

[0148] Example 23 can be combined with any one of Examples 12 to 19, and including receiving the CSI report further comprising: receiving the CSI report from the UE via a PUSCH transmission, the PUSCH transmission including the first PMI and the second PMI in different CSI parts of the PUSCH transmission.

[0149] Example 24 is a method for performing wireless communication at a UE, comprising: receiving a CSI-RS from a base station; selecting a wideband precoder W1 based on the CSI-RS; compressing a subband feature vector based on the CSI-RS and the wideband precoder W1; and sending a first PMI indicating the wideband precoder W1 and a second PMI indicating the compressed subband feature vector to the base station.

[0150] Example 25 can be combined with Example 24, and further includes receiving an RRC message from the base station, the RRC message configuring a wideband precoder codebook and the number of beams of the wideband precoder W1.

[0151] Example 26 can be combined with any one of Examples 24 to 25, and include that the RRC message is an RRC reconfiguration message.

[0152] Example 27 can be combined with any one of Examples 24 to 26, and further comprising, before compression, determining the subband feature vector as The first v columns of the eigenvector of , where N indicates the number of CSI-RS resource elements of subband S; is the channel estimated based on the CSI-RS at resource element k.

[0153] Example 28 can be combined with any one of Examples 24 to 27, and further include compressing the subband feature vector based on an AI / ML model.

[0154] Example 29 can be combined with any one of Examples 24 to 28, and further includes receiving an RRC message for configuring the AI / ML model from the base station.

[0155] Example 30 can be combined with any one of Examples 24 to 29, and include that the AI / ML model is a predefined AI / ML model or is predetermined by the UE.

[0156] Example 31 can be combined with any one of Examples 24 to 30, and includes sending the first PMI and the second PMI further comprising: generating a PUCCH transmission using a short PUCCH format, and including the first PMI and the second PMI in a single part of the short PUCCH format; and transmitting the PUCCH transmission to the base station.

[0157] Example 32 can be combined with any one of Examples 24 to 31, and includes sending the first PMI and the second PMI further including: generating a PUCCH transmission using a long PUCCH format, and including the first PMI and the second PMI in CSI part 2 of the long PUCCH format; and transmitting the PUCCH transmission to the base station.

[0158] Example 33 can be combined with any one of Examples 24 to 32, and includes sending the first PMI and the second PMI further including: generating a PUCCH transmission using a long PUCCH format, and including the first PMI in CSI part 1 of the short PUCCH format and including the second PMI in CSI part 2; and transmitting the PUCCH transmission to the base station.

[0159] Example 34 can be combined with any one of Examples 24 to 33, and includes sending the first PMI and the second PMI further including: generating a PUSCH transmission and including the first PMI and the second PMI in CSI part 2 of the PUSCH transmission; and transmitting the PUSCH transmission to the base station.

[0160] Example 35 can be combined with any one of Examples 24 to 34, and includes sending the first PMI and the second PMI further including: generating a PUSCH transmission, and including the first PMI in CSI part 1 of the PUSCH transmission and including the second PMI in CSI part 2; and transmitting the PUSCH transmission to the base station.

[0161] Example 36 is a method for performing wireless communication at a base station, comprising: transmitting a CSI-RS to a UE; transmitting a first RRC message to the UE, the first RRC message configuring a wideband precoder codebook and the number of beams of the wideband precoder W1; receiving a first precoder matrix indicator (PMI) indicating the wideband precoder W1 and a second PMI indicating a compressed subband eigenvector from the UE; decompressing the compressed subband eigenvector to obtain an uncompressed subband eigenvector; and transmitting a PDSCH signal using a precoder based on the uncompressed subband eigenvector.

[0162] Example 37 can be combined with Example 36, and include deriving the precoder based on the uncompressed subband eigenvector and the wideband PMI.

[0163] Example 38 can be combined with any one of Examples 36 to 37, and further include decompressing the compressed subband feature vector based on the ML model to obtain an uncompressed subband feature vector based on the ML model.

[0164] Example 39 can be combined with any one of Examples 36 to 38, and further include transmitting a second RRC message configuring the AI / ML model to the UE.

[0165] Example 40 can be combined with any one of Examples 36 to 39, and include that the AI / ML model is a predefined AI / ML model or is predetermined by the base station.

[0166] Example 41 may be combined with any one of Examples 36 to 40, and including receiving the first PMI and the second PMI further including receiving a PUCCH transmission including the first PMI and the second PMI according to a short PUCCH format.

[0167] Example 42 may be combined with any one of Examples 36 to 41, and including receiving the first PMI and the second PMI further including receiving a PUCCH transmission including the first PMI and the second PMI in CSI part 2 according to a long PUCCH format.

[0168] Example 43 can be combined with any one of Examples 36 to 42, and includes receiving the first PMI and the second PMI further including: receiving a PUCCH transmission including the first PMI in CSI part 1 and the second PMI in CSI part 2 according to a long PUCCH format.

[0169] Example 44 can be combined with any one of Examples 36 to 43, and including receiving the first PMI and the second PMI further including: receiving a PUSCH transmission including the first PMI and the second PMI in CSI part 2.

[0170] Example 45 can be combined with any one of Examples 36 to 44, and including receiving the first PMI and the second PMI further includes: receiving a PUSCH transmission including the first PMI in CSI part 1 and the second PMI in CSI part 2.

[0171] Example 46 is an apparatus for wireless communication implementing the method of any one of Examples 1 to 45.

[0172] Example 47 is an apparatus for wireless communication including means for implementing the method of any one of Examples 1 to 45.

[0173] Example 48 is a non-transitory computer-readable medium storing computer-executable code that, when executed by at least one processor, causes the at least one processor to implement the method of any one of Examples 1 to 45.

Claims

1. A method of wireless communication at a user equipment (UE) (102), comprising: receiving (240) a channel state information-reference signal (CSI-RS) for channel estimation (250) from a network entity (104); as well as A channel state information (CSI) report is sent (285) to the network entity (104) including a first precoding matrix indicator (PMI) and a second PMI, wherein the first PMI indicates a wideband precoder (355) associated with the channel estimate (250) and the second PMI indicates a compressed subband eigenvector associated with the wideband precoder (355).

2. The method of claim 1, further comprising: Subband eigenvectors are calculated (365a) based on the wideband precoder (355) and the channel estimate (250).

3. The method according to any one of claims 1 to 2, further comprising: The subband feature vector is compressed (270a) using the wideband precoder (355) to produce the compressed subband feature vector.

4. The method of any one of claims 1 to 3, wherein the compression (270a) of the sub-band feature vectors is based on using a machine learning (ML) model to generate the compressed sub-band feature vectors.

5. The method of claim 4, wherein a predefined protocol dictates the ML model used for the compression (270a) of the sub-band feature vectors.

6. The method according to any one of claims 1 to 4, further comprising: A configuration is received (402) from the network entity (104) indicating at least one of: the ML model used for the compression (270a) of the subband feature vectors, a codebook used for the wideband precoder (355), or a number of beams used for the wideband precoder (355).

7. The method according to any one of claims 1 to 6, further comprising: A trigger indication for the CSI report is received (404) from the network entity (104), the CSI report comprising the first PMI indicating the wideband precoder (355) and the second PMI indicating the compressed subband eigenvector.

8. The method of any one of claims 1 to 7, wherein sending the CSI report further comprises: The CSI report is transmitted (285) to the network entity (104) using a physical uplink control channel (PUCCH) format, the PUCCH format including the first PMI and the second PMI in a same CSI portion of the PUCCH format.

9. The method of any one of claims 1 to 7, wherein sending the CSI report further comprises: The CSI report is transmitted (285) to the network entity (104) using a physical uplink control channel (PUCCH) format, the PUCCH format including the first PMI and the second PMI in different CSI parts of the PUCCH format.

10. The method of any one of claims 1 to 7, wherein sending the CSI report further comprises: The CSI report is transmitted (285) to the network entity (104) using a physical uplink shared channel (PUSCH) transmission, the PUSCH transmission including the first PMI and the second PMI in a same CSI portion of the PUSCH transmission.

11. The method of any one of claims 1 to 7, wherein sending the CSI report further comprises: The CSI report is transmitted (285) to the network entity (104) using a physical uplink shared channel (PUSCH) transmission, the PUSCH transmission including the first PMI and the second PMI in different CSI parts of the PUSCH transmission.

12. A method for wireless communication at a network entity (104), comprising: transmitting (240) a channel state information-reference signal (CSI-RS) to a user equipment (UE) (102) to generate a channel estimate based on a channel state information (CSI) report; as well as The CSI report including a first precoding matrix indicator (PMI) and a second PMI is received (285) from the UE (102), the first PMI indicating a wideband precoder associated with the channel estimate and the second PMI indicating a compressed subband eigenvector associated with the wideband precoder.

13. The method of claim 12, further comprising: The compressed sub-band feature vector is decompressed (270b / 470b) to reconstruct an uncompressed sub-band feature vector.

14. The method of claim 13, further comprising: The uncompressed subband eigenvector is applied to the wideband precoder to generate the channel estimate associated with transmitting (240) the CSI-RS to the UE (102).

15. The method of any one of claims 12 to 14, further comprising: A physical downlink shared channel (PDSCH) signal is transmitted (495) to the UE (102) based on the wideband precoder and the uncompressed subband eigenvector.

16. The method of any one of claims 12 to 15, wherein the CSI report (285) includes the first PMI and the second PMI in the same CSI part.

17. The method of any one of claims 12 to 15, wherein the CSI report (285) includes the first PMI and the second PMI in different CSI parts.

18. An apparatus for wireless communication, comprising a memory, a transceiver and a processor, the processor being coupled to the memory and the transceiver, the apparatus being configured to implement the method according to any one of claims 1 to 17.

Images