Data collection using ideal and non-ideal channel estimation

Abstract

Certain aspects of the present disclosure provide wireless communications by a user equipment (UE), generally including: receiving signaling indicating a configuration of channel state information (CSI) reference signal (CSI-RS) resources; obtaining at least a first CSI-RS mode and a second CSI-RS mode from the configuration, wherein the first CSI-RS mode indicates a higher density of CSI-RS in a channel frequency range than the second CSI-RS mode; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; and sending at least the target model output data to an entity for training the ML model.

Description

Background Art

[0001] Aspects of the present disclosure relate to wireless communications, and more particularly to techniques for collecting data for training machine learning (ML) models.

[0002] Wireless communication systems are widely deployed to provide various telecommunication services such as telephony, video, data, messaging, broadcasts or other similar types of services. These wireless communication systems may employ multiple-access technologies capable of supporting communications with multiple users by sharing available wireless communication system resources with those users.

[0003] Although wireless communication systems have made tremendous technical progress over the years, challenges still exist. For example, complex and dynamic environments can still attenuate or block signals between wireless transmitters and wireless receivers. Therefore, there is a continuous desire to improve the technical performance of wireless communication systems, including, for example: improving the speed and data carrying capacity of communications, improving the efficiency of using shared communication media, reducing the power used by transmitters and receivers when performing communications, improving the reliability of wireless communications, avoiding redundant transmission and / or reception and related processing, improving the coverage area of ​​wireless communications, increasing the number and types of devices that can access wireless communication systems, increasing the ability of different types of devices to communicate with each other, increasing the number and types of wireless communication media available for use, etc. Therefore, there is a need to further improve wireless communication systems to overcome the above-mentioned technical challenges and other challenges. Summary of the invention

[0004] One aspect provides a method for wireless communication by a user equipment (UE). The method includes: receiving signaling indicating configuration of channel state information (CSI) reference signal (CSI-RS) resources; obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RS in a channel frequency range than the second CSI-RS pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; and sending at least the target model output data to an entity for training the ML model.

[0005] Another aspect provides a method for wireless communication by a UE. The method includes: receiving signaling indicating a configuration of a channel state information (CSI) reference signal (CSI-RS) pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the CSI-RS pattern; generating nominal data for training the ML model by adding artificial noise to the target model output data; and sending at least the target model output data to an entity for training the ML model.

[0006] On the other hand, a device for wireless communication is provided. The device includes a memory and a processor, which is coupled to the memory. The processor is configured to: receive signaling indicating the configuration of a channel state information (CSI) reference signal (CSI-RS) resource; obtain at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates that the density of the CSI-RS in the channel frequency range is higher than that of the second CSI-RS pattern; collect target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; collect nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; and send at least the target model output data to an entity for training the ML model.

[0007] On the other hand, a device for wireless communication is provided. The device includes a memory and a processor, which is coupled to the memory. The processor is configured to: receive signaling indicating the configuration of a channel state information (CSI) reference signal (CSI-RS) pattern; collect target model output data for a machine learning (ML) model based on channel measurements made according to the CSI-RS pattern; generate nominal data for training the ML model by adding artificial noise to the target model output data; and send at least the target model output data to an entity for training the ML model.

[0008] On the other hand, a device for wireless communication is provided. The device includes: a component for receiving signaling indicating the configuration of a channel state information (CSI) reference signal (CSI-RS) resource; a component for obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RS in a channel frequency range than the second CSI-RS pattern; a component for collecting target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; a component for collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; and a component for sending at least the target model output data to an entity for training the ML model.

[0009] Another aspect provides an apparatus for wireless communication. The apparatus includes: a component for receiving signaling indicating a configuration of a channel state information (CSI) reference signal (CSI-RS) pattern; a component for collecting target model output data for a machine learning (ML) model based on channel measurements made according to the CSI-RS pattern; a component for generating nominal data for training the ML model by adding artificial noise to the target model output data; and a component for sending at least the target model output data to an entity for training the ML model.

[0010] On the other hand, a non-transitory computer-readable medium is provided, on which instructions are stored, for receiving signaling indicating the configuration of a channel state information (CSI) reference signal (CSI-RS) resource; obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RS in a channel frequency range than the second CSI-RS pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; and sending at least the target model output data to an entity for training the ML model.

[0011] On the other hand, a non-transitory computer-readable medium is provided, on which instructions are stored, for receiving signaling indicating the configuration of a channel state information (CSI) reference signal (CSI-RS) pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the CSI-RS pattern; generating nominal data for training the ML model by adding artificial noise to the target model output data; and sending at least the target model output data to an entity for training the ML model.

[0012] Other aspects provide: an apparatus operable to, configured to, or otherwise adapted to perform any one or more of the foregoing methods and / or those described elsewhere herein; a non-transitory computer-readable medium comprising instructions that, when executed by a processor of the apparatus, cause the apparatus to perform the foregoing methods and those described elsewhere herein; a computer program product embodied on a computer-readable storage medium, the computer program product comprising code for performing the foregoing methods and those described elsewhere herein; and / or an apparatus comprising components for performing the foregoing methods and those described elsewhere herein. By way of example, an apparatus may include a processing system, a device having a processing system, or a processing system cooperating through one or more networks.

[0013] For purposes of illustration, the following description and drawings set forth certain features. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings depict certain features of the various aspects described herein and should not be considered to limit the scope of the disclosure.

[0015] Figure 1 An example wireless communication network is depicted.

[0016] Figure 2 An example disaggregated base station architecture is depicted.

[0017] Figure 3 Aspects of an example base station and example user equipment are depicted.

[0018] Figure 4A , Figure 4B , Figure 4C and Figure 4D Various example aspects of data structures for a wireless communication network are described.

[0019] Figure 5 The general functional framework applied to AI-enabled RAN intelligence is illustrated.

[0020] Figure 6 An example of an ML-based CSI feedback mechanism is depicted.

[0021] Figure 7 Describes the use Figure 6 Examples of encoder input and decoder output for the ML-based CSI feedback mechanism.

[0022] Figure 8 Depicted is an example training of an ML-based CSI feedback mechanism.

[0023] Fig. 9 Depicted is an example training of an ML-based CSI feedback mechanism in accordance with aspects of the present disclosure.

[0024] Fig. 10A and Fig. 10B An example of deriving a downlink precoder based on a high-density reference signal (RS) pattern and a low-density reference signal (RS) pattern in accordance with aspects of the present disclosure is depicted.

[0025] Fig.11 An example of a high-density RS mode in accordance with aspects of the present disclosure is depicted.

[0026] Fig.12 An example of a low-density RS pattern in accordance with aspects of the present disclosure is depicted.

[0027] Fig.13 Depicted is how an estimated precoder may be obtained by adding noise to a target precoder in accordance with aspects of the present disclosure.

[0028] Fig.14 A method for wireless communication is described.

[0029] Fig.15 A method for wireless communication is described.

[0030] Fig.16 Aspects of an example communications device are depicted. DETAILED DESCRIPTION

[0031] Aspects of the present disclosure relate to wireless communications, and more particularly to techniques for collecting data for training machine learning (ML) models.

[0032] Understanding the channel state between devices communicating in a wireless communication system is one aspect of improving wireless communication performance. Various techniques have been used to measure channel states and report feedback so that performance can be improved. For example, performance can be improved by measuring channel states and reporting feedback. Channel state information (CSI) feedback reports can be based on a codebook. The codebook is used as a pre-coding matrix indicator (PMI) dictionary, from which the user equipment (UE) can report the best PMI codeword for a given channel condition and use a bit sequence to report the PMI. In other words, instead of feeding back an actual value like a PMI codeword, the UE can only feed back a set of bits representing an entry in the codebook, thereby reducing signaling overhead. After receiving the feedback, the network entity can retrieve the PMI codeword from the codebook. Such techniques are typically relatively slow, power-consuming, and static in terms of methods.

[0033] Artificial intelligence (AI) or machine learning (ML) represents an opportunity to potentially improve the techniques used to measure channel state and report feedback. For example, a machine learning model can reduce the number of resource elements required to estimate the channel state (reducing signaling overhead) and improve the estimate of the value used when reporting the estimated channel state (e.g., by training an ML model to improve the accuracy of the channel estimate). In some cases, ML-based CSI feedback can be replaced by a CSI encoder and decoder (at least a decoder is required) instead of codebook-based processing. In such cases, the encoder can be similar to a PMI search algorithm, while the decoder can be similar to a PMI codebook used to convert CSI report bits into PMI codewords.

[0034] In some cases, the UE may be configured with a reference signal (RS) configuration to collect data for training the ML model. For example, the configuration may indicate resources (e.g., via resource ID, carrier ID, BWP ID, resource mapping, frequency band), meta information (e.g., for indicating transmit antenna configuration), and digital / analog beamforming. According to one or more examples, the UE performs channel estimation based on the RS, and performs processing of the channel estimation to obtain data for training (such as channel whitening, SVD of the channel estimate to obtain a target precoder V if necessary). The UE may upload the collected data to, for example, a training entity through its server via proprietary signaling.

[0035] In real-word deployment, the collected data may be generated via a channel estimation method based on an actual channel with noise. Since it is impossible to obtain an ideal channel estimate (without error), training may use a noisy channel estimate (i.e., a non-ideal channel). Models trained using such datasets may not achieve denoising capabilities because the input to the ML model and the ground truth are calculated based on the noisy channel estimate. The ground truth generally refers to the reality that will be modeled by the ML model (such as the target CSI in this example).

[0036] However, aspects of the present disclosure allow for more "ideal" ground truth values ​​(at least less noisy than the ML model input) to be used for ML model training. Therefore, the techniques presented herein can help improve ML model training and overall performance.

[0037] Introduction to wireless communication networks

[0038] The techniques and methods described herein can be used in various wireless communication networks. Although various aspects may be described herein using terms commonly associated with 3G, 4G and / or 5G wireless technologies, various aspects of the present disclosure may also be applicable to other communication systems and standards not explicitly mentioned herein.

[0039] Figure 1 An example of a wireless communication network 100 is depicted in which various aspects described herein may be implemented.

[0040] In general, the wireless communication network 100 includes various network entities (alternatively, network elements or network nodes). A network entity is typically a communication device and / or a communication function performed by a communication device (e.g., user equipment (UE), base station (BS), component of a BS, server, etc.). For example, various functions of a network and various devices associated with and interacting with the network may be considered network entities. In addition, the wireless communication network 100 includes ground aspects, such as ground-based network entities (e.g., BS 102), and non-ground aspects, such as satellites 140 and aircraft 145, which may include airborne network entities (e.g., one or more BSs) capable of communicating with other network elements (e.g., ground BSs) and user equipment.

[0041] In the depicted example, the wireless communication network 100 includes a BS 102, a UE 104, and one or more core networks (such as an evolved packet core (EPC) 160 and a 5G core (5GC) network 190) that interoperate to provide communication services over various communication links (including wired and wireless links).

[0042] Figure 1Various example UEs 104 are depicted, which may more generally include: a cellular phone, a smart phone, a Session Initiation Protocol (SIP) phone, a laptop, a personal digital assistant (PDA), a satellite radio, a global positioning system, a multimedia device, a video device, a digital audio player, a camera, a game console, a tablet computer, a smart device, a wearable device, a vehicle, an electric meter, a gas pump, a large or small kitchen appliance, a healthcare device, an implant, a sensor / actuator, a display, an Internet of Things (IoT) device, an Always-On (AON) device, an edge processing device, or other similar devices. UE 104 may also be more generally referred to as a mobile device, a wireless device, a wireless communication device, a station, a mobile station, a subscriber station, a mobile subscriber station, a mobile unit, a subscriber unit, a wireless unit, a remote unit, a remote device, an access terminal, a mobile terminal, a wireless terminal, a remote terminal, and a handset, etc.

[0043] BS 102 wirelessly communicates with (e.g., transmits signals to or receives signals from) UE 104 via communication link 120. Communication link 120 between BS 102 and UE 104 may include uplink (UL) (also referred to as a reverse link) transmissions from UE 104 to BS 102 and / or downlink (DL) (also referred to as a forward link) transmissions from BS 102 to UE 104. In various aspects, communication link 120 may utilize multiple-input multiple-output (MIMO) antenna technology, including spatial multiplexing, beamforming, and / or transmit diversity.

[0044] BS 102 may generally include: NodeB, enhanced NodeB (eNB), next generation enhanced NodeB (ng-eNB), next generation NodeB (gNB or gNodeB), access point, base transceiver station, radio base station, radio transceiver, transceiver function and / or transmission and reception point, etc. Each of BS 102 may provide communication coverage for a corresponding geographic coverage area 110, which may sometimes be referred to as a cell, and may overlap in some cases (e.g., a small cell 102' may have a coverage area 110' that overlaps with a coverage area 110 of a macro cell). For example, a BS may provide communication coverage for a macro cell (covering a relatively large geographic area), a pico cell (covering a relatively small geographic area, such as a stadium), a femto cell (relatively small geographic area (e.g., a home)), and / or other types of cells.

[0045] Although BS102 is depicted as a single communication device in various aspects, BS102 can be implemented in various configurations. For example, one or more components of the base station can be decomposed, including a central unit (CU), one or more distributed units (DU), one or more radio units (RU), a near real-time (near RT) RAN intelligent controller (RIC), or a non-real-time (non-RT) RIC, to name a few examples. In another example, various aspects of the base station can be virtualized. More generally, a base station (e.g., BS102) can include components located at a single physical location or components located at various physical locations. In an example in which the base station includes components located at various physical locations, the various components can each perform a function so that the various components jointly implement functionality similar to that of a base station located at a single physical location. In some aspects, a base station including components located at various physical locations can be referred to as a decomposed radio access network architecture (such as an open RAN (O-RAN) or virtualized RAN (VRAN) architecture). Figure 2 An example decomposed base station architecture is depicted and described.

[0046] Different BSs 102 within the wireless communication network 100 may also be configured to support different radio access technologies (such as 3G, 4G, and / or 5G). For example, a BS 102 configured for 4G LTE (collectively referred to as the Evolved Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network (E-UTRAN)) may interface with the EPC 160 via a first backhaul link 132 (e.g., an S1 interface). A BS 102 configured for 5G (e.g., 5G NR or Next Generation RAN (NG-RAN)) may interface with the 5GC 190 via a second backhaul link 184. The BSs 102 may communicate with each other directly or indirectly (e.g., via the EPC 160 or the 5GC 190) over a third backhaul link 134 (e.g., an X2 interface), which may be wired or wireless.

[0047] The wireless communication network 100 may subdivide the electromagnetic spectrum into various categories, frequency bands, channels, or other characteristics. In some aspects, subdivision is provided based on wavelength and frequency, where frequency may also be referred to as carrier, subcarrier, channel, tone, or subband. The communication link 120 between the BS 102 and, for example, the UE 104 may be through one or more carriers, which may have different bandwidths (e.g., 5 MHz, 10 MHz, 15 MHz, 20 MHz, 100 MHz, 400 MHz, and / or other MHz) and may be aggregated in various aspects. The carriers may or may not be adjacent to each other. The allocation of carriers may be asymmetric with respect to DL and UL (e.g., more or fewer carriers may be allocated for DL ​​compared to UL).

[0048] Communications using higher frequency bands may have higher path loss and shorter range than communications at lower frequencies. Figure 1 180) can utilize beamforming 182 with UE 104 to improve path loss and range. For example, BS 180 and UE 104 can each include multiple antennas, such as antenna elements, antenna panels and / or antenna arrays, to facilitate beamforming. In some cases, BS 180 can send beamformed signals to UE 104 in one or more transmit directions 182'. UE 104 can receive beamformed signals from BS 180 in one or more receive directions 182". UE 104 can also send beamformed signals to BS 180 in one or more transmit directions 182". BS 180 can also receive beamformed signals from UE 104 in one or more receive directions 182'. BS 180 and UE 104 can then perform beam training to determine the best receive direction and transmit direction for each of BS 180 and UE 104. It is worth noting that the transmit direction and receive direction of BS 180 may be the same or may not be the same. Similarly, the transmit direction and receive direction of UE 104 may or may not be the same.

[0049] Wireless communication network 100 further includes Wi-Fi AP 150 that communicates with Wi-Fi station (STA) 152 via communication link 154 in, for example, the 2.4 GHz and / or 5 GHz unlicensed spectrum.

[0050] Certain UEs 104 may communicate with each other using a device-to-device (D2D) communication link 158. The D2D communication link 158 may 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), a physical sidelink control channel (PSCCH), and / or a physical sidelink feedback channel (PSFCH).

[0051] The EPC 160 may include various functional components, including: a mobility management entity (MME) 162, other MMEs 164, a serving gateway 166, a multimedia broadcast multicast service (MBMS) gateway 168, a broadcast multicast service center (BM-SC) 170, and / or a packet data network (PDN) gateway 172, such as in the depicted example. The MME 162 may communicate with a home subscriber server (HSS) 174. The MME 162 is a control node that handles signaling between the UE 104 and the EPC 160. Generally speaking, the MME 162 provides bearer and connection management.

[0052] Generally, user Internet Protocol (IP) packets are delivered through the Serving Gateway 166, which itself is connected to the PDN Gateway 172. The PDN Gateway 172 provides UE IP address allocation and other functions. The PDN Gateway 172 and the BM-SC 170 are connected to IP services 176, which may include, for example, the Internet, an intranet, an IP Multimedia Subsystem (IMS), a Packet Switched (PS) streaming media service, and / or other IP services.

[0053] The BM-SC 170 may provide functionality for MBMS user service provisioning and delivery. The BM-SC 170 may serve as an entry point for content provider MBMS delivery, may be used to authorize and initiate MBMS bearer services within a public land mobile network (PLMN), and / or may be used to schedule MBMS delivery. The MBMS Gateway 168 may be used to distribute MBMS services to BSs 102 belonging to a multicast broadcast single frequency network (MBSFN) area of ​​a broadcast specific service, and / or may be responsible for session management (start / stop) and for collecting eMBMS related charging information.

[0054] 5GC 190 may include various functional components, including: access and mobility management function (AMF) 192, other AMFs 193, session management function (SMF) 194, and user plane function (UPF) 195. AMF 192 may communicate with unified data management (UDM) 196.

[0055] AMF 192 is a control node that handles signaling between UE 104 and 5GC 190. AMF 192 provides, for example, Quality of Service (QoS) flow and session management.

[0056] Internet Protocol (IP) packets are delivered through UPF 195, which is connected to IP services 197 and provides IP address allocation for UEs and other functions for 5GC 190. IP services 197 may include, for example, the Internet, intranet, IMS, PS streaming services, and / or other IP services.

[0057] The wireless communication network 100 includes a machine learning component 199 that can perform the operations described herein related to the machine learning timeline and / or machine learning concurrent processing. The wireless network 100 further includes a machine learning component 198 that can perform the operations described herein related to the machine learning timeline and / or machine learning concurrent processing.

[0058] In various aspects, a network entity or network node may be implemented as a converged base station, a decomposed base station, a component of a base station, an integrated access and backhaul (IAB) node, a relay node, a sidelink node, to name a few examples.

[0059] Figure 2 An example disaggregated base station 200 architecture is depicted. The disaggregated base station 200 architecture may include one or more central units (CUs) 210 that may communicate directly with a core network 220 via a backhaul link, or indirectly with the core network 220 through one or more disaggregated base station units, such as a near real-time (near-RT) RAN intelligent 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 CU 210 may communicate with one or more distributed units (DUs) 230 via respective midhaul links, such as an F1 interface. The DU 230 may communicate with one or more radio units (RUs) 240 via respective fronthaul links. The RU 240 may communicate with respective UEs 104 via one or more radio frequency (RF) access links. In some implementations, a UE 104 may be served simultaneously by multiple RUs 240.

[0060] Each of the units (e.g., CU 210, DU 230, RU 240, and near-RT RIC 225, non-RT RIC 215, and SMO framework 205) may include or be coupled to one or more interfaces configured to receive or send signals, data, or information (collectively referred to as signals) via a wired or wireless transmission medium. Each of the units or an associated processor or controller that provides instructions to the communication interface of the unit may be configured to communicate with one or more of the other units via a transmission medium. For example, a unit may include a wired interface configured to receive or send signals to one or more of the other units on a wired transmission medium. Additionally or alternatively, a unit may include a wireless interface that may include a receiver, transmitter, or transceiver (such as a radio frequency (RF) transceiver) configured to receive signals on a wireless transmission medium or send signals to one or more of the other units, or both.

[0061] In some aspects, CU 210 may host one or more higher layer control functions. Such control functions may include radio resource control (RRC), packet data convergence protocol (PDCP), or service data adaptation protocol (SDAP), etc. Each control function may be implemented using an interface configured to communicate signals with other control functions hosted by CU 210. CU 210 may be configured to handle user plane functionality (e.g., central unit-user plane (CU-UP)), control plane functionality (e.g., central unit-control plane (CU-CP)), or a combination thereof. In some specific implementations, CU 210 may 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 unit may communicate bidirectionally with the CU-CP unit via an interface (such as an E1 interface). As needed, CU 210 may be implemented to communicate with DU230 for network control and signaling.

[0062] DU 230 may correspond to a logical unit that includes one or more base station functions for controlling the operation of one or more RUs 240. In some aspects, DU 230 may 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 or modulation and demodulation, etc.) at least in part according to a functional split such as those defined by the 3rd Generation Partnership Project (3GPP). In some aspects, DU 230 may further host one or more low PHY layers. Each layer (or module) may be implemented using an interface that is configured to communicate signals with other layers (and modules) hosted by DU 230 or with control functions hosted by CU 210.

[0063] The lower layer functionality may be implemented by one or more RUs 240. In some deployments, a RU 240 controlled by a DU 230 may correspond to a logical node that hosts 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 RU 240 may 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 RU 240 may be controlled by the corresponding DU 230. In some scenarios, this configuration may enable the implementation of the DU 230 and the CU 210 in a cloud-based RAN architecture (such as a vRAN architecture).

[0064] The SMO framework 205 may be configured to support RAN deployment and provisioning of non-virtualized network elements and virtualized network elements. For non-virtualized network elements, the SMO framework 205 may be configured to support the deployment of dedicated physical resources for RAN coverage requirements, which may be managed via an operation and maintenance interface (such as an O1 interface). For virtualized network elements, the SMO framework 205 may be configured to interact with a cloud computing platform (such as an 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 O2 interface). Such virtualized network elements may include, but are not limited to, CU 210, DU 230, RU 240, and near-RT RIC 225. In some specific implementations, the SMO framework 205 may communicate with hardware aspects of the 4G RAN (such as an open eNB (O-eNB) 211) via the O1 interface. Additionally, in some specific implementations, the SMO framework 205 may communicate directly with one or more RUs 240 via the O1 interface. The SMO framework 205 may also include a non-RT RIC 215 configured to support the functionality of the SMO framework 205 .

[0065] The non-RT RIC 215 may be configured to include logic functions that enable non-real-time control and optimization of RAN elements and resources, artificial intelligence / machine learning (AI / ML) workflows including model training and updating, or policy-based guidance of applications / features in the near-RT RIC 225. The non-RT RIC 215 may be coupled to or communicate with the near-RT RIC 225 (such as via an A1 interface). The near-RT RIC 225 may be configured to include logic 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 the O-eNB with the near-RT RIC 225.

[0066] In some implementations, in order to generate an AI / ML model to be deployed in the near-RT RIC 225, the non-RT RIC 215 may receive parameters or external enrichment information from an external server. Such information may be utilized by the near-RT RIC 225 and may be received from a non-network data source or from a network function at the SMO framework 205 or the non-RT RIC 215. In some examples, the non-RT RIC 215 or the near-RT RIC 225 may be configured to tune RAN behavior or performance. For example, the non-RT RIC 215 may monitor long-term trends and patterns of performance and employ AI / ML models to perform corrective actions through the SMO framework 205 (such as via reconfiguration of O1) or via creation of RAN management policies (such as A1 policies).

[0067] Figure 3 Aspects of an example BS 102 and UE 104 are depicted.

[0068] In general, BS 102 includes various processors (e.g., 320, 330, 338, and 340), antennas 334a-334t (collectively 334), transceivers 332a-332t (collectively 332) including modulators and demodulators, and other aspects that enable wireless transmission of data (e.g., data source 312) and wireless reception of data (e.g., data sink 339). For example, BS 102 can transmit and receive data between BS 102 and UE 104. BS 102 includes a controller / processor 340 that can be configured to implement various functions described herein related to wireless communications.

[0069] Base station 102 includes a controller / processor 340 that can be configured to perform various functions related to wireless communication. In the depicted example, controller / processor 340 includes machine learning component 241, which can represent Figure 1 The machine learning component 199. It is worth noting that, although depicted as one aspect of the controller / processor 340, in other specific implementations, the machine learning component 341 can be additionally or alternatively implemented in various other aspects of the base station 102.

[0070] Generally speaking, the UE 104 includes various processors (e.g., 358, 364, 366, and 380), antennas 352a-352r (collectively 352), transceivers 354a-354r (collectively 354) including modulators and demodulators, and other aspects that enable wireless transmission of data (e.g., retrieved from a data source 362) and wireless reception of data (e.g., provided to a data sink 360). The UE 104 includes a controller / processor 380 that may be configured to implement various functions described herein related to wireless communications.

[0071] User equipment 104 includes a controller / processor 380 that can be configured to implement various functions related to wireless communication. In the depicted example, controller / processor 380 includes a machine learning component 381 that can represent Figure 1 The machine learning component 198. It is worth noting that, although depicted as one aspect of the controller / processor 380, in other specific implementations, the machine learning component 381 may additionally or alternatively be implemented in various other aspects of the user equipment 104.

[0072] Regarding example downlink transmissions, BS 102 includes a transmit processor 320 that can receive data from a data source 312 and control information from a controller / processor 340. The control information can be for a physical broadcast channel (PBCH), a physical control format indicator channel (PCFICH), a physical HARQ indicator channel (PHICH), a physical downlink control channel (PDCCH), and / or a group common PDCCH (GC PDCCH), etc. In some examples, the data can be for a physical downlink shared channel (PDSCH).

[0073] The transmit processor 320 may process (e.g., encode and symbol map) data and control information to obtain data symbols and control symbols, respectively. The transmit processor 320 may also generate reference symbols (such as for a primary synchronization signal (PSS), a secondary synchronization signal (SSS), a PBCH demodulation reference signal (DMRS), and a channel state information reference signal (CSI-RS)).

[0074] The transmit (TX) multiple-input multiple-output (MIMO) processor 330 may perform spatial processing (e.g., pre-decoding) on ​​data symbols, control symbols, and / or reference symbols where applicable, and may provide an output symbol stream to a modulator (MOD) in a transceiver 332a-332t. Each modulator in the transceiver 332a-332t may process a corresponding output symbol stream to obtain an output sample stream. Each modulator may further process (e.g., convert to analog, amplify, filter, and up-convert) the output sample stream to obtain a downlink signal. The downlink signals from the modulators in the transceivers 332a-332t may be transmitted via antennas 334a-334t, respectively.

[0075] To receive downlink transmissions, UE 104 includes antennas 352a-352r that can receive downlink signals from BS 102 and can provide received signals to demodulators (DEMODs) in transceivers 354a-354r, respectively. Each demodulator in transceivers 354a-354r can condition (e.g., filter, amplify, downconvert, and digitize) a corresponding received signal to obtain input samples. Each demodulator can further process the input samples to obtain received symbols.

[0076] A MIMO detector 356 may obtain received symbols from all demodulators in transceivers 354a-354r, perform MIMO detection on the received symbols if applicable, and provide detected symbols. A receive processor 358 may process (e.g., demodulate, deinterleave, and decode) the detected symbols, provide decoded data for UE 104 to a data sink 360, and provide decoded control information to a controller / processor 380.

[0077] With respect to example uplink transmissions, the UE 104 further includes a transmit processor 364 that may receive and process data from a data source 362 (e.g., for a PUSCH) and control information from a controller / processor 380 (e.g., for a physical uplink control channel (PUCCH)). The transmit processor 364 may also generate reference symbols for reference signals (e.g., for a sounding reference signal (SRS)). The symbols from the transmit processor 364 may be pre-decoded by a TX MIMO processor 366, if applicable, further processed by a modulator in the transceivers 354a-354r (e.g., for SC-FDM), and transmitted to the BS 102.

[0078] At BS 102, uplink signals from UE 104 may be received by antennas 334 a-334 t, processed by demodulators in transceivers 332 a-332 t, detected by MIMO detector 336 (if applicable), and further processed by receive processor 338 to obtain decoded data and control information transmitted by UE 104. Receive processor 338 may provide decoded data to data sink 339 and decoded control information to controller / processor 340.

[0079] Memory 342 and memory 382 may store data and program codes for BS 102 and UE 104, respectively.

[0080] The scheduler 344 may schedule UEs for data transmission on the downlink and / or uplink.

[0081] In various aspects, BS 102 may be described as sending and receiving various types of data associated with the methods described herein. In these contexts, "sending" may refer to various mechanisms for outputting data, such as outputting data from a data source 312, a scheduler 344, a memory 342, a transmit processor 320, a controller / processor 340, a TX MIMO processor 330, a transceiver 332a-332t, an antenna 334a-334t, and / or other aspects described herein. Similarly, "receiving" may refer to various mechanisms for obtaining data, such as obtaining data from an antenna 334a-334t, a transceiver 332a-332t, a RX MIMO detector 336, a controller / processor 340, a receive processor 338, a scheduler 344, a memory 342, and / or other aspects described herein.

[0082] In various aspects, the UE 104 may also be described as sending and receiving various types of data associated with the methods described herein. In these contexts, "sending" may refer to various mechanisms for outputting data, such as outputting data from a data source 362, a memory 382, ​​a transmit processor 364, a controller / processor 380, a TX MIMO processor 366, a transceiver 354a-354t, an antenna 352a-352t, and / or other aspects described herein. Similarly, "receiving" may refer to various mechanisms for obtaining data, such as obtaining data from an antenna 352a-352t, a transceiver 354a-354t, a RX MIMO detector 356, a controller / processor 380, a receive processor 358, a memory 382, ​​and / or other aspects described herein.

[0083] In some aspects, the processor may be configured to perform various operations, such as those associated with the methods described herein, and send (output) data to or receive (obtain) data from another interface configured to send or receive data, respectively.

[0084] Figure 4A , Figure 4B , Figure 4C and Figure 4D Describes a method for use in a wireless communication network such as Figure 1 Various aspects of the data structure of the wireless communication network 100).

[0085] Specifically, Figure 4A is a diagram 400 illustrating an example of a first subframe within a 5G (e.g., 5G NR) frame structure, Figure 4B is a diagram 430 illustrating an example of a DL channel within a 5G subframe, Figure 4C is a diagram 450 illustrating an example of a second subframe within a 5G frame structure, and Figure 4DFIG480 is a diagram illustrating an example of UL channels within a 5G subframe.

[0086] Wireless communication systems can utilize orthogonal frequency division multiplexing (OFDM) with a cyclic prefix (CP) on both the uplink and downlink. Such systems can also support half-duplex operation using time division duplex (TDD). OFDM and single carrier frequency division multiplexing (SC-FDM) can be used to maximize the system bandwidth (e.g., Figure 4B and Figure 4D The modulation symbols may be transmitted in the frequency domain using OFDM and / or in the time domain using SC-FDM.

[0087] The wireless communication frame structure may be frequency division duplex (FDD), where for a particular set of subcarriers, subframes within the subcarrier set are dedicated to either DL or UL. The wireless communication frame structure may also be time division duplex (TDD), where for a particular set of subcarriers, subframes within the subcarrier set are dedicated to both DL and UL.

[0088] exist Figure 4A and Figure 4C In the wireless communication frame structure, TDD is used, where D is DL, U is UL, and X is used flexibly between DL / UL. The UE can be configured with a time slot format (dynamically configured by DL control information (DCI) or semi-statically / statically configured by radio resource control (RRC) signaling) through the received time slot format indicator (SFI). In the depicted example, the 10ms frame is divided into 10 equally sized 1ms subframes. Each subframe may include one or more time slots. In some examples, each time slot may include 7 or 14 symbols, depending on the time slot format. The subframe may also include micro-slots, which typically have fewer symbols than the entire time slot. Other wireless communication technologies may have different frame structures and / or different channels.

[0089] In certain aspects, the number of slots within a subframe is based on the slot configuration and parameter set. For example, for slot configuration 0, different parameter sets (μ) 0 to 6 allow each subframe to have 1, 2, 4, 8, 16, 32, and 64 slots, respectively. For slot configuration 1, different parameter sets 0 to 2 allow each subframe to have 2, 4, and 8 slots, respectively. Thus, for slot configuration 0 and parameter set μ, there are 14 symbols / slot and 2μ slots / subframe. The subcarrier spacing and symbol length / duration are a function of the parameter set. The subcarrier spacing can be equal to 2 μ× 15kHz, where μ is parameter set 0 to 6. Thus, parameter set μ=0 has a subcarrier spacing of 15kHz, and parameter set μ=6 has a subcarrier spacing of 960kHz. The symbol length / duration is inversely related to the subcarrier spacing. Figure 4A , Figure 4B , Figure 4C and Figure 4D An example is provided for slot configuration 0 with 14 symbols per slot and parameter set μ=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 μs.

[0090] like Figure 4A , Figure 4B , Figure 4C and Figure 4D As depicted in , a resource grid can be used to represent a frame structure. Each time slot includes a resource block (RB) (also referred to as a physical RB (PRB)) extending, for example, over 12 consecutive subcarriers. The resource grid is divided into a plurality of resource elements (REs). The number of bits carried by each RE depends on the modulation scheme.

[0091] like Figure 4A As illustrated, some REs carry Figure 1 and Figure 3 The RS may include a demodulation RS (DMRS) and / or a channel state information reference signal (CSI-RS) for channel estimation at the UE. The RS may also include a beam measurement RS (BRS), a beam refinement RS (BRRS), and / or a phase tracking RS (PT-RS).

[0092] Figure 4B Examples of various DL 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 comprising, for example, nine RE Groups (REGs), each REG comprising, for example, four consecutive REs in an OFDM symbol.

[0093] The primary synchronization signal (PSS) may be in symbol 2 of a particular subframe of a frame. The PSS is transmitted by a UE (e.g., Figure 1 and Figure 3 104) is used to determine subframe / symbol timing and physical layer identity.

[0094] A Secondary Synchronization Signal (SSS) may be within symbol 4 of a specific subframe of a frame. The SSS is used by the UE to determine the physical layer cell identity group number and radio frame timing.

[0095] Based on the physical layer identifier and the physical layer cell identifier group number, the UE can determine the physical cell identifier (PCI). Based on the PCI, the UE can determine the location of the aforementioned DMRS. 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. The MIB provides the system frame number (SFN) and the number of RBs in the system bandwidth. The physical downlink shared channel (PDSCH) carries user data, broadcast system information (such as system information blocks (SIBs)) that is not sent via the PBCH, and / or paging messages.

[0096] like Figure 4C As illustrated, some REs carry DMRS for channel estimation at the base station (indicated as R for a specific configuration, but other DMRS configurations are possible). The UE may send DMRS for PUCCH and DMRS for PUSCH. The PUSCH DMRS may be sent, for example, in the first one or two symbols of the PUSCH. The PUCCH DMRS may be sent in different configurations depending on whether a short PUCCH or a long PUCCH is sent and depending on the specific PUCCH format used. UE104 may send a sounding reference signal (SRS). The SRS may be sent, for example, in the last symbol of a subframe. The SRS may have a comb structure, and the UE may send the SRS on one of the teeth of the comb. The SRS may be used by the base station for channel quality estimation to achieve frequency-dependent scheduling of the UL.

[0097] Figure 4D Examples of various UL channels within a subframe of a frame are illustrated. The PUCCH may be located at a position as indicated in one configuration. The PUCCH carries uplink control information (UCI), such as scheduling requests, channel quality indicator (CQI), precoding matrix indicator (PMI), rank indicator (RI), and HARQ ACK / NACK feedback. The PUSCH carries data and may additionally be used to carry a buffer status report (BSR), a power headroom report (PHR), and / or UCI.

[0098] Introduction to mmWave Wireless Communications

[0099] In wireless communications, the electromagnetic spectrum is often subdivided into various classes, bands, channels, or other characteristics. The subdivision is usually provided based on wavelength and frequency, where frequency may also be referred to as a carrier, subcarrier, channel, tone, or subband.

[0100] In 5G NR, two initial operating frequency bands have been identified as frequency range designations FR1 (410 MHz-7125 MHz) and FR2 (24,250 MHz-71,000 MHz). In some cases, FR2 may be further defined according to sub-ranges, such as a first sub-range FR2-1 including 24,250 MHz-52,600 MHz and a second sub-range FR2-2 including 52,600 MHz-71,000 MHz. It should be understood that, although portions of FR1 are greater than 6 GHz, FR1 is often (interchangeably) referred to as the “sub-6 GHz” band in various documents and articles. A similar naming issue sometimes occurs with respect to FR2, which is often (interchangeably) referred to as the “millimeter wave” band in documents and articles, despite being different from the extremely high frequency (EHF) band (30 GHz-300 GHz) identified as the “millimeter wave” band by the International Telecommunication Union (ITU).

[0101] The frequencies between FR1 and FR2 are generally referred to as mid-band frequencies. Recent 5G NR research has identified the operating bands for these mid-band frequencies as frequency range designation FR3 (7.125GHz-24.25GHz). The bands falling within FR3 can inherit FR1 characteristics and / or FR2 characteristics, and therefore the features of FR1 and / or FR2 can be effectively extended to mid-band frequencies. In addition, currently higher frequency bands are being explored to extend 5G NR operation to more than 52.6GHz. For example, three higher operating bands have been identified as frequency range designation FR4a or FR4-1 (52.6GHz-71GHz), FR4 (52.6GHz-114.25GHz) and FR5 (114.25GHz-300GHz). Each of these higher frequency bands falls within the EHF band.

[0102] In view of the above aspects, unless otherwise specifically stated, it should be understood that if the term "below 6 GHz" or the like is used in this document, it can be broadly referred to as a frequency that may be less than 6 GHz, may be within FR1, or may include mid-band frequencies. In addition, unless otherwise specifically stated, it should be understood that if the term "millimeter wave" or the like is used in this document, it can be broadly referred to as a frequency that may include mid-band frequencies, may be within FR2, FR4, FR4-a or FR4-1 and / or FR5, or may be within the EHF band.

[0103] Communications using mmWave / near mmWave radio frequency bands (e.g., 3 GHz-300 GHz) may have higher path loss and shorter range than lower rate communications. Figure 1As described, a base station (e.g., 180) configured to communicate using mmWave / near-mmWave radio frequency bands can utilize beamforming (e.g., 182) with a UE (e.g., 104) to improve path loss and range.

[0104] In addition, as described herein, the UE may use a machine learning model to estimate the channel or generate channel state information in mmWave bands and / or other frequency bands.

[0105] Overview of AI / ML functional framework for RAN intelligence

[0106] The AI / ML functional framework includes a data collection function 502, a model training function 504, a model inference function 506, and an actor function 508, which interoperate to provide a platform for collaboratively applying AI / ML to various procedures in the RAN.

[0107] The data collection function 502 typically provides input data to the model training function 504 and the model inference function 506. AI / ML algorithm specific data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) may not be performed in the data collection function 502.

[0108] Examples of input data to the data collection function 502 (or other functions) may include measurements from UEs or different network entities, feedback from auto-author functions, and outputs from AI / ML models. In some cases, data analysis required at the model training function 504 and the model inference function 506 may be performed at the data collection function 502. As illustrated, the data collection function 502 may deliver training data to the model training function 504 and deliver inference data to the model inference function 506.

[0109] The model training function 504 can perform AI / ML model training, validation, and testing, which can generate model performance metrics as part of the model testing procedure. If necessary, the model training function 504 can also be responsible for data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) based on the training data delivered by the data collection function 502.

[0110] The model training function 504 may provide model deployment / update data to the model interface function 506. The model deployment / update data may be used to initially deploy a trained, validated, tested AI / ML model to the model inference function 506 or to deliver an updated model to the model inference function 506.

[0111] As illustrated, the model inference function 506 can provide AI / ML model inference output (e.g., prediction or decision) to the actor function 508, and sometimes also provide model performance feedback to the model training function 504. The model inference function 506 can also sometimes be responsible for data preparation (e.g., data preprocessing and cleaning, formatting and transformation) based on the inference data delivered by the data collection function 502.

[0112] Inference output of the AI / ML model may be produced by the model inference function 506. The specific details of this output may be specific in terms of the use case. Sometimes, model performance feedback may be used to monitor the performance of the AI / ML model. In some cases, for example, if certain information derived from the model inference function is applicable to improving the AI / ML model trained in the model training function 504, the model performance feedback may be delivered to the model training function 504.

[0113] The model inference function 506 may signal the outputs of the model to nodes that have requested them (e.g., via subscriptions) or to nodes that take actions based on the outputs from the model inference function. Prior to deployment, the AI / ML models used in the model inference function 506 may require initial training, validation, and testing by the model training function. The model training function 504 and the model inference function 506 may be able to request specific information for training or executing the AI / ML algorithm and avoid receiving unnecessary information. The nature of such information may depend on the use case and the AI / ML algorithm.

[0114] The actor function 508 may receive output from the model inference function 506, which may trigger or perform corresponding actions. The actor function 508 may trigger actions for other entities or for itself. The feedback generated by the actor function 508 may provide information for deriving training data, inferring data, or for monitoring the performance of the AI / ML model. As described above, the input data for the data collection function 502 may include this feedback from the author function 508. Feedback from the author function 508 or other network entities (via the data collection function) may also be used at the model inference function 506.

[0115] The AI / ML functional framework 500 can be deployed in various RAN intelligence-based use cases. Such use cases can include CSI feedback enhancement, enhanced beam management (BM), positioning and location (Pos-Loc) accuracy enhancement, and various other use cases.

[0116] In certain aspects, the UE or BS may use continuous measurements or reported L1-RSRP in the time domain to perform ML-based beam prediction. In some cases, a pre-trained deep neural network (DNN) model may be used for such ML-based predictive beam management.

[0117] Traditionally, beam quality and failure are identified through measurement reports carried by relevant downlink (DL) and uplink (UL) reference signals (e.g., SSB, CSI-RS, RSRP), which increases beam selection latency and beam management overhead, while in the same case, beam selection accuracy may be limited due to power and overhead constraints, which may lead to poor system performance.

[0118] In contrast, AI / ML-based predictive beam management can reduce the amount of reference signal transmission used to predict unmeasured beam quality and the future possibility of beam blocking / failure. In predictive beam management, beam prediction can be a highly nonlinear problem that can be efficiently solved by a pre-trained DNN model that can predict future beam quality, for example, based on UE mobility speed and trajectory that are difficult to model through statistical processing methods.

[0119] Overview of ML model-based techniques

[0120] Various techniques may be used to help determine the channel state between wireless communication devices so that those devices can optimize their wireless communication configurations (e.g., select the best beam for sending and receiving data). For example, a channel state information reference signal (CSI-RS) may be transmitted by one device and measured by another device in order to estimate the channel state and provide channel state information (CSI) feedback useful for optimizing wireless communication between the two devices.

[0121] However, due to the increasing complexity and capabilities of wireless communication devices, such as those capable of transmitting and receiving through multiple input and output antenna ports (e.g., implementing multiple-input multiple-output (MIMO) technology), certain techniques may require significant processing power and time, which reduces the performance of both the device and the entire wireless communication network. These technical problems are exacerbated by the specific use cases and environments for wireless communication, which are often dynamic. In other words, because channel states frequently change, channel estimation and feedback procedures are often performed frequently, resulting in high power usage and significant network overhead for wireless communication systems (e.g., in terms of time and frequency resources dedicated to channel estimation). One way to alleviate such problems is to implement machine learning models that can perform various functions related to channel state estimation and feedback more accurately and efficiently.

[0122] For example, a wireless communication system may use time division multiplexing (TDM), code division multiplexing (CDM), and / or frequency division multiplexing (FDM) to allocate N packets per resource block. t Multiplex N resource elements tSuch a system may typically achieve a resource block density between 0.5 and 1, such that a resource element is transmitted in every other or every single resource block. In contrast, a machine learning model deployed by a transmitting device (e.g., a base station) may be trained to perform N on L resource elements per resource block. t Multiplexing of ports, where L <N t , which therefore reduces the number of resource elements required for channel estimation, leaving more resource elements available for data transmission. In addition to reducing the number of resource elements required for channel estimation (which reduces power and enhances resource utilization), such models can also operate at reduced resource block densities (e.g., below 0.5) and can also implement non-uniform resource block patterns, which further improves the aforementioned benefits. On the receiving device (e.g., user equipment) side, a machine learning-based channel estimator can be trained to recover the complete channel over all resource blocks (e.g., N t ports), while receiving a reduced number of resources L. In various aspects, the CSI-RS multiplexing models at the transmitter side and the receiver side may be trained jointly or sequentially.

[0123] As another example, the CSI reporting configuration may rely on a precoding matrix indicator (PMI) search algorithm and a PMI codebook to determine the best PMI codeword (e.g., CSI feedback) and report it to the network. However, machine learning-based models (such as encoders and decoders) may be trained to directly generate CSI feedback, which eliminates the need for a PMI search algorithm (replaced by an encoder) and a PMI codebook (replaced by a decoder). In various aspects described herein, a CSI encoder at the user equipment side may be trained to compress a channel estimate to a few bits that are then reported to a network entity (e.g., a base station), while a CSI decoder at the network entity side is trained to use the reported bits to recover the channel or precoding matrix.

[0124] Thus, in general, machine learning models can be trained to perform many functions related to channel estimation and feedback, and such models can often be more accurate, faster, more power efficient, and better able to maintain performance in very dynamic radio environments.

[0125] There are various options for what type of information is fed back as channel state feedback (CSF), and the decision on what to feed back may depend on the specific implementation (e.g., whether AI / ML based). The type of feedback may vary on a continuum from relatively sparse information such as rank indicator (RI), PMI, and channel quality indicator (CQI) to detailed full channel information.

[0126] For CSI feedback, the CSI report configuration may include a codebook. The codebook is used as a precoding matrix indicator (PMI) dictionary from which the UE can select and report the best PMI codeword, thereby reporting the PMI using a bit sequence.

[0127] like Figure 6 As illustrated in the example block diagram 600 of , in some cases, the AI-based CSI feedback may be replaced by a CSI encoder 602 (at the UE) and a decoder 604 at the network (e.g., gNB) to replace the codebook. In some cases, the encoder may not be required and only the decoder may be used. Generally speaking, the encoder is similar to the PMI search algorithm in current systems, while the decoder is typically similar to the PMI codebook, which is used to convert the CSI report bits into PMI codewords.

[0128] like Figure 7 As illustrated in Table 700 of , the input to the decoder includes a downlink channel matrix (H), a downlink pre-decoder (V), and an interference covariance matrix (Rnn). The output of the decoder may be a downlink channel matrix (H), a transmit covariance matrix, a downlink pre-decoder (V), and an interference covariance matrix (Rnn), where H or V may correspond to the original channel or a channel pre-whitened by the UE based on its demodulation filter.

[0129] In a specific ML model for CSI measurement and reporting, the UE and network entity (e.g., gNB) can collaborate by sharing assistance information. For example, in one form of collaboration, inter-node assistance can help the corresponding node improve the ML-based algorithm. This can apply to the UE receiving assistance information from the gNB (e.g., for training, adaptation, etc.) and the UE receiving assistance information from the gNB. In some cases, the assistance information can be exchanged without exchanging information about the actual ML model. In other cases (e.g., for joint ML operation), the UE and gNB can exchange information about the ML model or ML model instructions.

[0130] The ML-based CSI scheme involving low-density CSI-RS patterns is an example of a scenario where information can be exchanged between the UE and the gNB. For example, due to the association between the low-density CSI-RS pattern and the UE channel estimation (CHEST) ML model. In some cases, the CHEST ML model may need to be changed. For example, when the environment changes, the CSI-RS pattern may be changed. In such cases, the UE may need auxiliary information from the gNB for training and adaptation, etc.

[0131] The dimensionality reduction along the frequency dimension can be based on a low-density resource block (RB) pattern. For example, K RBs out of N RBs (K < N) can be selected to transmit CSI-RS. In such cases, the pattern can be a uniform RB pattern, a random RB pattern, or a learned RB pattern (e.g., a certain form of pattern can be dedicated to certain scenarios).

[0132] Non-orthogonal covering codes can be used to multiplex N_t CSI-RS ports onto L resource elements (REs) within an RB, where L < N_t. The UE can use an ML model associated with the low-density CSI-RS pattern to recover the channel matrix h^.

[0133] The dimensionality reduction along the frequency dimension can be based on a low-density resource block (RB) pattern. For example, K RBs out of N RBs (K < N) can be selected to transmit CSI-RS. In such cases, the pattern can be a uniform RB pattern, a random RB pattern, or a learned RB pattern (e.g., a certain form of pattern can be dedicated to certain scenarios).

[0134] Non-orthogonal covering codes can be used to multiplex N_t CSI-RS ports onto L resource elements (REs) within an RB, where L < N_t. The UE can use an ML model associated with the low-density CSI-RS pattern to recover the channel matrix h^.

[0135] The UE can receive signaling from a network entity indicating at least a first CSI-RS pattern. The first CSI-RS pattern can indicate the time and frequency resources (e.g., RBs and / or REs within an RB) for CSI-RS transmission. The first CSI-RS pattern can indicate that the density of CSI-RS resources in the channel frequency range is lower compared to a second (e.g., full-density) CSI-RS pattern.

[0136] The UE can update the ML model based on the first CSI-RS pattern. For example, as described above, during training, the UE can apply a loss function by comparing the output of the ML model generated based on measurements obtained from CSI-RS transmitted according to the first CSI-RS pattern with the ground truth estimated channel obtained using the full-density CSI-RS pattern. The first CSI-RS pattern can also be used when the UE uses the ML model for channel estimation (e.g., after the ML model is well-trained and ready to be used).

[0137] For example, using the updated ML model, the UE may generate CSI for the channel frequency range based on measurements made by the UE according to the first CSI-RS pattern. For example, the UE may input measurements made from CSI-RS sent according to the first CSI-RS pattern into the trained ML model to generate a channel estimate. The UE may then send a report indicating the CSI for the channel frequency range to a network entity.

[0138] Aspects related to data collection using ideal and non-ideal channel estimates

[0139] As described above, the UE may be configured with a reference signal (RS) configuration to collect data for training the ML model. For example, the configuration may indicate resources (e.g., via resource ID, carrier ID, BWP ID, resource mapping, frequency band), meta information (e.g., for indicating transmit antenna configuration), and digital / analog beamforming. The UE performs channel estimation based on the RS, and performs processing of the channel estimation to obtain data for training (such as channel whitening, SVD of the channel estimate to obtain the target precoder V if necessary). The UE may upload the collected data to, for example, a training entity through its server via proprietary signaling.

[0140] There are various types of ML model training. For example, a first type of training, known as joint training (or centralized training), may be performed at a single training entity (e.g., a UE / gNB-side server or a 3rd party). In this case, the UE may provide input data and target output data (e.g., ground truth) to the training entity where the training occurs. In some cases, the UE may provide the input data and target output data to its server, and the server forwards the input data and target output data to the training entity via proprietary signaling. After training, the model may be tested, compiled (e.g., confirmed to meet performance targets), and stored in a model repository, and the UE / gNB downloads the model from the model repository.

[0141] The second type of training may be joint training performed at both the UE-side server and the gNB-side server. In this case, a single training session may occur across both sides. In each iteration, the UE-side server may provide activations to the gNB-side server, and the gNB-side server may provide gradients to the UE-side server for the UE-side servers to update their ML models. In this case, the target output data (ground truth) is provided to the gNB-side server (e.g., data flows from UE->UE server->gNB server), and the input data is provided to the UE server, both of which may be provided via proprietary signaling.

[0142] The second type of training may be separate training at the UE side server and the gNB side server. In this case, training occurs first at the UE side and then at the gNB side (or vice versa). For example, the UE may train its encoder-decoder pair, providing the potential information (e.g., encoder output / decoder input) and the target decoder output (or just the decoder output) to the gNB side. The gNB side server may use this potential information as input and the target decoder output (or UE side decoder output) as ground truth to train its decoder. In this method, the target output data and input data may be provided via proprietary signaling to the UE server where the UE side training occurs. After the UE side training is complete, the potential information and the target decoder output or the output of the UE side decoder are provided to the gNB side server via proprietary signaling.

[0143] like Figure 8 As illustrated, the collected (training) data may be generated via a channel estimation method. Figure 8 The example shown in shows an ML model involving an encoder 802 at a UE and a decoder 804 at a network entity (gNB).

[0144] Since it is impossible to obtain an ideal channel estimate (without error), training may use a noisy channel estimate (i.e., a non-ideal channel). A model trained using such a dataset compares the pre-decoder output (Vout) from the decoder 804 with the model data input (Vest) to the encoder 802, which may not achieve denoising capabilities because the input to the ML model and the ground truth are calculated based on the noisy channel estimate.

[0145] like Fig. 9 As illustrated, an example ML model involving an encoder 902 at a UE and a decoder 904 at a network entity (gNB) is shown, however, aspects of the present disclosure allow for more "ideal" target ground truth values ​​(at least less noisy than the ML model input) to be used for ML model training. In this case, the model can be trained with such a dataset, thereby comparing the pre-decoder output (Vout) from the decoder 904 with the more ideal target model output data Vtarget. Therefore, the techniques presented herein can help improve ML model training and overall performance.

[0146] In some cases, the target model output can be obtained from the measurement of the high-density reference signal (RS_high). Fig. 10A As illustrated in the module 1002 of , Vtarget can be determined according to RS_high (eg, Vtarget=f(RS_high)). Fig. 10BAs illustrated in module 1004 of , a normal training data set may be obtained from the measurement of a low-density reference signal RS_low (eg, Vest=f(RS_low)).

[0147] In some cases, a network entity (e.g., a gNB) may send a configuration of a high density RS mode and / or a low density mode to a UE. In some cases, a training entity may take nominal data as input and take target output data as their respective ground truth values. As used herein, nominal data generally refers to data input to an ML model to generate an ML data output. Nominal data and target model output data may be provided for training an ML model. For training purposes, actual ML data outputs may be compared to expected target model output data.

[0148] According to certain aspects, a high-density RS (e.g., CSI-RS) pattern (used to determine Vtarget) may have a CSI-RS density greater than 1 (density>1). CSI-RS density generally refers to the number of CSI-RS observations (receptions) per resource block (RB), for example, the number of REs allocated for CSI-RS per CSI-RS port per RB. A lower-density RS (CSI-RS) with fewer CSI-RS observations per RB compared to a high-density RS may be obtained by zeroing the observations on the corresponding tone (ignoring reception). In other cases, instead of zeroing, multiple CSI-RS resources may be configured, and the UE may obtain multiple CSI-RS patterns from the multiple resources.

[0149] In some cases, the UE may receive a configuration of a high-density RS (e.g., density = 3) and determine a zeroing pattern or zeroing tone index for a low-density RS (e.g., density = 1). In some cases, the zeroing tone index (or pattern) may be cell-specific (e.g., configured per cell or determined based on a cell ID). In other words, a lower density RS may be obtained by applying the zeroing tone index or pattern to obtain a subset of the high-density RS (resources) for use as a low-density RS.

[0150] In some cases, the higher density RS pattern may be an aggregation of multiple low density RS resources. In other words, a higher density RS may be obtained by aggregating multiple lower density RS (resources). In such cases, the UE may receive a configuration of N lower density RS resources for data collection. N resources may be bundled and triggered together, which typically means that the target output is obtained from measurements of the N bundled resources, while the nominal data set may be obtained from measurements of a single resource.

[0151] The N resources may be multiplexed in the frequency domain or the time domain. In some cases, each resource may have a specific RE position or symbol / slot position. In some cases, the N resources may have the same quasi-co-located (QCL) reference signal and type (e.g., type A, B, C, and / or D), including Doppler shift, average gain, delay spread, maximum delay, Rx spatial information, or they may be QCL to each other (which means sharing one or more common QCL parameters).

[0152] like Fig.11 As illustrated, in some cases, a new CSI-RS pattern may be required to facilitate high-density RS (>1) and to facilitate multiplexing of three low-density (e.g., nominal CSI-RS density = 1) CSI-RS resources in one time slot 1100. For example, for a 32-port CSI-RS transmission, a total of 96 REs may be required.

[0153] like Fig.11 As illustrated, a new 4x8 CSI-RS pattern with a total of 4 starting symbol positions in the time domain can be defined. In this example, each starting symbol generally means two consecutive symbols. The 4 starting symbol positions can be evenly distributed (as shown in the figure), or can be non-uniformly distributed.

[0154] In some cases, for 32-port CSI-RS transmission, 8 symbols may be used (e.g., spanning 4 illustrated starting symbol positions). For other cases, for 24-port CSI-RS transmission, 6 symbols (3 starting symbol positions) may be used. For other cases, for 16-port CSI-RS transmission, 4 symbols (2 starting symbol positions) may be used. For 32-port CSI-RS transmission, the first 16 ports may be mapped to the first 4 symbols, and the remaining 16 ports may be mapped to the last 4 symbols.

[0155] exist Fig.11 In the example configuration shown, there are 3 possible RE positions (0, 4, 8). In the case of a CSI-RS density of 3, all three possible RE positions may be occupied. On the other hand, a density of 1 would mean that only one of these RE positions is occupied (e.g., 3 resources may occupy 3 RE positions). According to one example, each RB may occupy 3 REs per port.

[0156] refer to Fig.12, consider a 32-port CSI-RS, the first copy of these 32 ports can occupy 4x8 REs with subcarriers 0-3 and symbols 2-9 (marked as mode 1). The second copy can occupy 4x8 REs with subcarriers 4-7 and symbols 2-9 (marked as mode 2). The third copy can occupy 4x8 REs with subcarriers 8-11 and symbols 2-9 (marked as mode 3). Fig.12 As illustrated, for low-density RSs collecting nominal data for ML training, the UE may ignore the 2nd and 3rd copies on subcarriers 4-11. In some cases, a time domain (TD) orthogonal cover code (OCC) 8 may be used, or a frequency domain (e.g., FD2-TD8 or TD-OCC8) code division multiplexing (CDM) pattern may be defined.

[0157] In some cases, the same (e.g., 32) ports are sent in different (high-density and low-density) modes (each copy of the CSI-RS observation / reception of a single resource, or each CSI-RS resource in multiple CSI-RS resources to be aggregated together). In such cases, aggregation can produce better channel estimates by having higher processing gain than the channel estimates produced by measuring one of the low-density resources or zeroing the CSI-RS observation / reception of the high-density resources. In some other cases, each CSI-RS resource is sent using a different 32 ports, and the UE can then use a single resource to perform port prediction in the spatial domain (the output is 64 ports if there are two resources, or 96 ports if there are three resources). In this case, the UE can use the aggregated resources to generate a target output of a total of 64 (or 96) ports. The channel estimate obtained using the aggregated resources can have better channel estimate quality than using one of the resources for port prediction / port extrapolation.

[0158] In some cases, such as Fig.13 As illustrated, the UE may collect target model output data for a machine learning (ML) model based on channel measurements made according to a CSI-RS pattern, and generate nominal data for training the ML model by adding artificial noise to the target model output data.

[0159] In other words, the target model output can be generated using the measurement of the reference signal (RS) used / configured for data collection, and the UE can obtain (Vtarget=f(RS)). Fig.13 As illustrated, the dataset used in training can be generated by adding artificial noise to the target model output (Vest=f(RS)+noise).

[0160] The UE shall include in the data collection the statistics of the artificial noise (eg, variance) or the ratio between the channel estimate power and the noise variance (ie, SNR).

[0161] The statistics for each data sample can be different, and there can be multiple noisy versions added to the same sample:

[0162] Vest1=f(RS)+noise1, Vest2=f(RS_high)=f(RS)+noise2, and so on.

[0163] The two noises can have different statistics.

[0164] There are various options for delivering training data (regardless of which of the above options are used). For example, for joint training at a single entity, the UE may upload (Vest, Vtarget) or (Hest, Vtarget) to the training entity or model server (e.g., where the model server may forward the data to the training entity). In such cases, the training entity considers Vest or Hest as input, and Vtarget may be the ground truth value as an approximate target.

[0165] For joint training across UE and gNB, the UE can upload Vtarget to its model server, and the model server exchanges with the gNB side server via proprietary signaling, or the UE can directly upload Vtarget to the gNB side server. In this case, the UE side server can treat Vest as input, and the gNB side server uses Vtarget as the ground truth value as the approximate target.

[0166] In some cases, in separate training using a UE-driven approach, the UE may upload (Vest, Vtarget) or (Hest, Vtarget) to its model server, which uses Vest / Hest as input and Vtarget as target to train the UE-side encoder and decoder pair. In such cases, the encoder output z and Vtarget (or the output of the UE-side decoder) may be shared with the gNB server via proprietary signaling. The gNB may use z as input and Vtarget (or the output of the UE-side decoder) as target to train its decoder.

[0167] In some cases, a gNB-driven approach may be used. In this case, the UE may upload an indication of (Vest, Vtarget) or (Hest, Vtarget) to its model server, and the server may share (e.g., Vest, Vtarget) or (e.g., Hest, Vtarget) with the gNB server. The gNB server may then use Vest / Hest as input and Vtarget as a target to train its encoder and decoder pair. Although the examples in this document refer to ML models based on encoder / decoder pairs, the techniques presented in this document may be applied to ML models that do not use encoder / decoder pairs. The gNB server may deliver the encoder output z to the UE, and the UE may use Vest / Hest as input to train its encoder and use z as the target output of the encoder. In some cases, the received signal-to-noise ratio (SNR) or reference signal received power (RSRP) of the CSI-RS may be uploaded with each sample (e.g., when ML model training data is provided).

[0168] Example Operation of User Equipment

[0169] Fig.14 It is shown that a UE (such as Figure 1 and Figure 3 An example of a method 1400 for performing wireless communications with a UE 104).

[0170] Method 1400 begins at step 1405: receiving signaling indicating the configuration of CSI-RS resources. In some cases, the operation of this step refers to the following steps: Fig.16 The described circuits for receiving and / or codes for receiving may be performed by or may be executed by the circuits and / or codes.

[0171] Method 1400 then proceeds to step 1410: obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RS in the channel frequency range than the second CSI-RS pattern. In some cases, the operation of this step refers to as described in reference Fig.16 The described circuit for obtaining and / or code for obtaining, or can be executed by the circuit and / or the code.

[0172] Method 1400 then proceeds to step 1415: collecting target model output data for the ML model based on channel measurements made according to the first CSI-RS pattern. In some cases, the operation of this step refers to the operation described in reference to Fig.16 The described circuit for collecting and / or code for collecting, or can be executed by the circuit and / or the code.

[0173] Method 1400 then proceeds to step 1420: collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern. In some cases, the operation of this step refers to the operation described in reference to Fig.16 The described circuit for collecting and / or code for collecting, or can be executed by the circuit and / or the code.

[0174] Method 1400 then proceeds to step 1425: sending at least the target model output data to an entity for training the ML model. In some cases, the operation of this step refers to the operation of Fig.16 The described circuits for transmitting and / or codes for transmitting, or can be performed by the circuits and / or the codes.

[0175] In some aspects, the second CSI-RS pattern includes a CSI-RS subset of the first CSI-RS pattern.

[0176] In some aspects, obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration includes: obtaining the first CSI-RS pattern from the configuration; and obtaining the second CSI-RS pattern from the first CSI-RS pattern and a zeroing tone pattern.

[0177] In some aspects, the nulling tone pattern is cell-specific.

[0178] In some aspects, the first CSI-RS pattern has a CSI-RS density greater than 1, where the CSI-RS pattern density indicates a number of CSI-RS per RB.

[0179] In some aspects, the first CSI-RS pattern is configured with: up to four starting symbol positions in a slot; and up to eight symbols total in the slot.

[0180] In some aspects, the first CSI-RS pattern comprises a superset of CSI-RS resources of the second CSI-RS pattern.

[0181] In some aspects, obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration includes: obtaining the second CSI-RS pattern from a plurality of CSI-RS patterns indicated by the configuration for data collection; and obtaining the first CSI-RS pattern by aggregating the plurality of CSI-RS patterns.

[0182] In some aspects, the method 1400 further includes: receiving a resource bundling configuration, wherein the aggregation is based on the resource bundling configuration. In some cases, the operation of this step refers to the Fig.16The described circuits for receiving and / or codes for receiving may be performed by or may be executed by the circuits and / or codes.

[0183] In some aspects, the multiple CSI-RS patterns are multiplexed in at least one of frequency or time.

[0184] In some aspects, the multiple CSI-RS patterns have the same QCL reference signal and type, or are QCL to each other.

[0185] In some aspects, a common CSI-RS port is sent in both the first CSI-RS pattern and the second CSI-RS pattern.

[0186] In some aspects, a first set of CSI-RS ports is transmitted in the first CSI-RS pattern and a second set of CSI-RS ports is transmitted in the second CSI-RS pattern.

[0187] In some aspects, the nominal data includes at least one of a downlink predecoder (Vest) or a downlink channel matrix (Hest), which is estimated based on channel measurements made according to the second CSI-RS pattern; and the target model output data includes a target downlink predecoder (Vtarget), which is estimated based on channel measurements made according to the first CSI-RS pattern.

[0188] In some aspects, the entity comprises a model server or a training entity; and sending at least the target model output data to the entity for training the ML model comprises sending Vtarget and at least one of Vest or Hest.

[0189] In some aspects, the method 1400 further includes: transmitting at least one of an SNR or an RSRP corresponding to a CSI-RS received according to at least one of the first CSI-RS pattern or the second CSI-RS pattern. In some cases, the operation of this step refers to as described in reference Fig.16 The described circuits for transmitting and / or codes for transmitting, or can be performed by the circuits and / or the codes.

[0190] In some aspects, the nominal data is used as input to the ML model; and the target model output data is labeled on a sample-by-sample basis as ground truth for the ML model.

[0191] In one aspect, method 1400 or any aspect related thereto may be performed by an apparatus such as Fig.16The method 1400 is performed by a communication device 1600 that includes various components operable, configured, or adapted to perform the method 1400. The communication device 1600 is described in more detail below.

[0192] Please note that Fig.14 This is merely one example of a method, and other methods including fewer, additional, or alternative steps may also be consistent with the present disclosure.

[0193] Fig.15 It is shown that a UE (such as Figure 1 and Figure 3 An example of a method 1500 for performing wireless communications with a UE 104).

[0194] Method 1500 begins at step 1505: receiving signaling indicating the configuration of a CSI-RS pattern. In some cases, the operation of this step refers to the following steps. Fig.16 The described circuits for receiving and / or codes for receiving may be performed by or may be executed by the circuits and / or codes.

[0195] Method 1500 then proceeds to step 1510: collecting target model output data for the ML model based on channel measurements made according to the CSI-RS pattern. In some cases, the operation of this step refers to the operation described in reference to Fig.16 The described circuit for collecting and / or code for collecting, or can be executed by the circuit and / or the code.

[0196] Method 1500 then proceeds to step 1515: generating nominal data for training the ML model by adding artificial noise to the target model output data. In some cases, the operation of this step refers to the Fig.16 The described circuit for generating and / or code for generating, or can be executed by the circuit and / or the code.

[0197] Method 1500 then proceeds to step 1520: sending at least the target model output data to an entity for training the ML model. In some cases, the operation of this step refers to the operation of Fig.16 The described circuits for transmitting and / or codes for transmitting, or can be performed by the circuits and / or the codes.

[0198] In some aspects, the method 1500 further includes: sending an indication of the statistics of the artificial noise. In some cases, the operation of this step refers to the Fig.16 The described circuits for transmitting and / or codes for transmitting, or can be performed by the circuits and / or the codes.

[0199] In some aspects, the statistic comprises at least one of a variance or a ratio between a channel estimate power and a noise variance.

[0200] In some aspects, the statistics are different for different data samples of the nominal data.

[0201] In some aspects, generating the nominal data includes generating a plurality of sets of nominal data from the common target output data, wherein each set is generated using particular noise statistics.

[0202] In some aspects, the nominal data includes at least one of a downlink predecoder (Vest) or a downlink channel matrix (Hest), which is estimated based on channel measurements made according to the CSI-RS pattern; and the target model output data includes a target downlink predecoder (Vtarget), which is estimated based on channel measurements made according to the CSI-RS pattern.

[0203] In some aspects, the entity comprises a model server or a training entity; and sending at least the target model output data to the entity for training the ML model comprises sending Vtarget and at least one of Vest or Hest.

[0204] In some aspects, the method 1500 further includes: transmitting at least one of an SNR or an RSRP corresponding to the CSI-RS received according to the CSI-RS pattern. In some cases, the operation of this step refers to as described in reference Fig.16 The described circuits for transmitting and / or codes for transmitting, or can be performed by the circuits and / or the codes.

[0205] In some aspects, the nominal data is used as input to the ML model; and the target model output data is labeled on a sample-by-sample basis as ground truth for the ML model.

[0206] In one aspect, method 1500 or any aspect related thereto may be performed by an apparatus such as Fig.16 The method 1500 is performed by a communication device 1600 that includes various components operable, configured, or adapted to perform the method 1500. The communication device 1600 is described in more detail below.

[0207] Please note that Fig.15 This is merely one example of a method, and other methods including fewer, additional, or alternative steps may also be consistent with the present disclosure.

[0208] Example Communication Device

[0209] Fig.16 Various aspects of an example communications device 1600 are depicted. In some aspects, the communications device 1600 is a user equipment (UE), such as described above with respect to Figure 1 and Figure 3 UE 104 is described.

[0210] The communication device 1600 includes a processing system 1605 coupled to a transceiver 1675 (e.g., a transmitter and / or a receiver). The transceiver 1675 is configured to transmit and receive signals of the communication device 1600, such as various signals as described herein, via an antenna 1680. The processing system 1605 may be configured to perform processing functions of the communication device 1600, including processing signals received and / or to be transmitted by the communication device 1600.

[0211] The processing system 1605 includes one or more processors 1610. In various aspects, such as with respect to Figure 3 As described, the one or more processors 1610 may represent one or more of the receive processor 358, the transmit processor 364, the TX MIMO processor 366, and / or the controller / processor 380. The one or more processors 1610 are coupled to the computer-readable medium / memory 1640 via the bus 1670. In some aspects, the computer-readable medium / memory 1640 is configured to store instructions (e.g., computer-executable code) that, when executed by the one or more processors 1610, cause the one or more processors 1610 to perform operations related to the processing of the data transmission. Fig.14 The method 1400 described herein or any aspect thereof, and / or any aspect thereof Fig.15 The described method 1500 or any aspect related thereto. Note that references to a processor performing a function of the communication device 1600 may include one or more processors 1610 performing that function of the communication device 1600 .

[0212] In the depicted example, computer readable medium / memory 1640 stores code (e.g., executable instructions) such as code for receiving 1645, code for obtaining 1650, code for collecting 1655, code for sending 1660, and code for generating 1665. Processing of code for receiving 1645, code for obtaining 1650, code for collecting 1655, code for sending 1660, and code for generating 1665 may enable communication device 1600 to perform operations related to Fig.14 The method 1400 described herein or any aspect thereof, and / or any aspect thereof Fig.15 The described method 1500 or any aspect related thereto.

[0213] The one or more processors 1610 include circuits configured to implement (e.g., execute) code stored in the computer-readable medium / memory 1640, including circuits such as circuits for receiving 1615, circuits for obtaining 1620, circuits for collecting 1625, circuits for sending 1630, and circuits for generating 1635. Processing using the circuits for receiving 1615, circuits for obtaining 1620, circuits for collecting 1625, circuits for sending 1630, and circuits for generating 1635 may enable the communication device 1600 to perform operations related to Fig.14 The method 1400 described herein or any aspect thereof, and / or any aspect thereof Fig.15 The described method 1500 or any aspect related thereto.

[0214] The various components of the communication device 1600 may provide for performing Fig.14 The method 1400 described herein or any aspect thereof, and / or any aspect thereof Fig.15 The components of the described method 1500 or any aspect related thereto. For example, the components for sending, transmitting, or outputting for sending may include Figure 3 The illustrated transceiver 354 and / or antenna 352 and / or Fig.16 The transceiver 1675 and antenna 1680 of the communication device 1600 in FIG. 1 may include Figure 3 The illustrated transceiver 354 and / or antenna 352 and / or Fig.16 The transceiver 1675 and antenna 1680 of the communication device 1600 in FIG.

[0215] Sample Clauses

[0216] Specific implementation examples are described in the following numbered clauses:

[0217] Clause 1: A method of wireless communication by a UE, the method comprising: receiving signaling indicating a configuration of CSI-RS resources; obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RS in a channel frequency range than the second CSI-RS pattern; collecting target model output data for an ML model based on channel measurements made according to the first CSI-RS pattern; collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; and sending at least the target model output data to an entity for training the ML model.

[0218] Clause 2: The method of clause 1, wherein the second CSI-RS pattern comprises a CSI-RS subset of the first CSI-RS pattern.

[0219] Clause 3: A method according to clause 2, wherein obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration includes: obtaining the first CSI-RS pattern from the configuration; and obtaining the second CSI-RS pattern from the first CSI-RS pattern and a zero tone pattern.

[0220] Clause 4: The method of clause 3, wherein the null tone pattern is cell-specific.

[0221] Clause 5: The method of clause 3, wherein the first CSI-RS pattern has a CSI-RS density greater than 1, wherein the CSI-RS pattern density indicates a number of CSI-RS per RB.

[0222] Clause 6: The method of clause 3, wherein the first CSI-RS pattern is configured with: up to four starting symbol positions in a slot; and up to eight symbols total in the slot.

[0223] Clause 7: A method as set forth in any of clauses 1 to 6, wherein the first CSI-RS pattern comprises a superset of CSI-RS resources of the second CSI-RS pattern.

[0224] Clause 8: A method according to clause 7, wherein obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration comprises: obtaining the second CSI-RS pattern from a plurality of CSI-RS patterns indicated by the configuration for data collection; and obtaining the first CSI-RS pattern by aggregating the plurality of CSI-RS patterns.

[0225] Clause 9: The method of clause 8, further comprising: receiving a resource bundling configuration, wherein the aggregation is based on the resource bundling configuration.

[0226] Clause 10: The method of clause 8, wherein the plurality of CSI-RS patterns are multiplexed in at least one of frequency or time.

[0227] Clause 11: The method of clause 8, wherein the multiple CSI-RS patterns have the same QCL reference signal and type, or are QCL to each other.

[0228] Clause 12: The method of clause 8, wherein a common CSI-RS port is sent in both the first CSI-RS pattern and the second CSI-RS pattern.

[0229] Clause 13: The method of clause 8, wherein a first set of CSI-RS ports is transmitted in the first CSI-RS pattern and a second set of CSI-RS ports is transmitted in the second CSI-RS pattern.

[0230] Clause 14: A method according to any one of clauses 1 to 13, wherein: the nominal data includes at least one of a downlink predecoder (Vest) or a downlink channel matrix (Hest), the downlink predecoder and the downlink channel matrix are estimated based on channel measurements performed according to the second CSI-RS pattern; and the target model output data includes a target downlink predecoder (Vtarget), the target downlink predecoder is estimated based on channel measurements performed according to the first CSI-RS pattern.

[0231] Clause 15: A method according to clause 14, wherein: the entity comprises a model server or a training entity; and sending at least the target model output data to the entity for training the ML model comprises sending Vtarget and at least one of Vest or Hest.

[0232] Clause 16: The method of clause 15, further comprising sending at least one of an SNR or an RSRP corresponding to a CSI-RS received according to at least one of the first CSI-RS pattern or the second CSI-RS pattern.

[0233] Clause 17: A method according to any one of clauses 1 to 16, wherein: the nominal data is used as input to the ML model; and the target model output data is labeled on a sample-by-sample basis as the ground truth value of the ML model.

[0234] Clause 18: A method of wireless communication by a UE, the method comprising: receiving signaling indicating a configuration of a CSI-RS pattern; collecting target model output data for an ML model based on channel measurements made according to the CSI-RS pattern; generating nominal data for training the ML model by adding artificial noise to the target model output data; and sending at least the target model output data to an entity for training the ML model.

[0235] Clause 19: The method of clause 18, further comprising: sending an indication of statistics of the artificial noise.

[0236] Clause 20: The method of clause 19, wherein the statistical data comprises at least one of a variance or a ratio between a channel estimate power and a noise variance.

[0237] Clause 21: The method of clause 19, wherein the statistical data is different for different data samples of the nominal data.

[0238] Clause 22: The method of any of clauses 18 to 21, wherein generating the nominal data comprises generating a plurality of sets of nominal data from common target output data, wherein each set is generated using specific noise statistics.

[0239] Clause 23: A method according to any one of clauses 18 to 22, wherein: the nominal data includes at least one of a downlink predecoder (Vest) or a downlink channel matrix (Hest), the downlink predecoder and the downlink channel matrix are estimated based on channel measurements performed according to the CSI-RS pattern; and the target model output data includes a target downlink predecoder (Vtarget), the target downlink predecoder is estimated based on channel measurements performed according to the CSI-RS pattern.

[0240] Clause 24: A method according to clause 23, wherein: the entity comprises a model server or a training entity; and sending at least the target model output data to the entity for training the ML model comprises sending Vtarget and at least one of Vest or Hest.

[0241] Clause 25: The method of clause 24, further comprising transmitting at least one of an SNR or an RSRP corresponding to a CSI-RS received according to the CSI-RS pattern.

[0242] Clause 26: A method according to any one of clauses 18 to 25, wherein: the nominal data is used as input to the ML model; and the target model output data is labeled on a sample-by-sample basis as the ground truth value of the ML model.

[0243] Clause 27: An apparatus, the apparatus comprising: a memory including executable instructions; and a processor configured to execute the executable instructions and cause the apparatus to perform the method according to any one of clauses 1 to 26.

[0244] Clause 28: An apparatus comprising means for performing the method according to any one of clauses 1 to 26.

[0245] Clause 29: A non-transitory computer readable medium comprising executable instructions which, when executed by a processor of an apparatus, cause the apparatus to perform the method of any one of clauses 1 to 26.

[0246] Clause 30: A computer program product embodied on a computer readable storage medium, the computer program product comprising code for performing the method according to any one of clauses 1 to 26.

[0247] Additional considerations

[0248] The foregoing description is provided to enable any person skilled in the art to practice the various aspects described herein. The examples discussed herein are not limited to the scope, applicability or aspects set forth in the claims. 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. For example, without departing from the scope of the present disclosure, the functions and arrangements of the elements discussed may be changed. Various examples may omit, replace or add various procedures or components as appropriate. For example, the described methods may be performed in an order different from the order described, and various actions may be added, omitted or combined. In addition, features equivalent to some example descriptions may be combined in some other examples. For example, any number of aspects set forth herein may be used to implement a device or practice method. In addition, the scope of the present disclosure is intended to cover such devices or methods practiced using other structures, functionality or structures and functionality that are supplementary or alternative to the various aspects of the present disclosure set forth herein. It should be understood that any aspect of the disclosure disclosed herein may be embodied by one or more elements of the present invention.

[0249] The various illustrative logical blocks, modules, and circuits described in conjunction with the present disclosure may be implemented or performed using a general purpose processor, a digital signal processor (DSP), an ASIC, a field programmable gate array (FPGA) or other programmable logic device (PLD), discrete gate or transistor logic components, discrete hardware components, or any combination thereof, designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in an alternative, the processor may be any commercially available processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, a system on a chip (SoC), or any other such configuration.

[0250] As used herein, a phrase referring to "at least one of" a list of items refers to any combination of those items (including single members). For example, "at least one of a, b, or c" is intended to cover a, b, c, ab, ac, bc, and abc, as well as any combination with multiple identical elements (e.g., aa, aaa, aab, aac, abb, acc, bb, bbb, bbc, cc, and ccc, or any other ordering of a, b, and c).

[0251] As used herein, the term "determining" encompasses a wide variety of actions. For example, "determining" may include calculating, computing, processing, deriving, investigating, searching (e.g., searching in a table, database, or other data structure), ascertaining, etc. In addition, "determining" may include receiving (e.g., receiving information), accessing (e.g., accessing data in a memory), etc. In addition, "determining" may include parsing, selecting, choosing, establishing, etc.

[0252] The method disclosed herein includes one or more actions for implementing the method. The method actions can be interchangeable with each other without departing from the scope of the claims. In other words, unless a specific order of actions is specified, the order and / or use of specific actions can be modified without departing from the scope of the claims. In addition, the various operations of the method described above can be performed by any appropriate component that can perform the corresponding function. The component may include various hardware and / or software components and / or modules, including but not limited to circuits, application specific integrated circuits (ASICs) or processors.

[0253] The following claims are not intended to be limited to the various aspects shown herein, but should be given the full scope consistent with the language of the claims. Within the claims, unless otherwise specified, reference to an element in the singular form is not intended to mean "one and only one", but "one or more". Unless otherwise specified, the term "some" refers to one or more. Any claim element is not interpreted according to the provisions of 35 U.S.C. § 112 (f) unless the element is explicitly stated using the phrase "parts for...". All structural and functional equivalents of the elements of the various aspects described throughout this disclosure that are known or will be known later to a person of ordinary skill in the art are expressly incorporated herein by reference and are intended to be covered by the claims. In addition, nothing disclosed herein is intended to be dedicated to the public, regardless of whether such disclosure is explicitly stated in the claims.

Claims

1. A device for wireless communication, the device comprising: Memory; and a processor coupled to the memory, the processor being configured to: receiving signaling indicating configuration of a channel state information (CSI) reference signal (CSI-RS) resource; obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RSs in a channel frequency range than the second CSI-RS pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; as well as At least the target model output data is sent to an entity for training the ML model.

2. The apparatus of claim 1, wherein the second CSI-RS pattern comprises a CSI-RS subset of the first CSI-RS pattern.

3. The apparatus of claim 2, wherein to obtain at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, the processor is configured to: Obtaining the first CSI-RS pattern from the configuration; and The second CSI-RS pattern is obtained from the first CSI-RS pattern and a nulling tone pattern.

4. The apparatus of claim 3, wherein the nulling tone pattern is cell-specific. 5 . The apparatus of claim 3 , wherein the first CSI-RS pattern has a CSI-RS density greater than 1, wherein the CSI-RS pattern density indicates the number of CSI-RSs per resource block (RB).

6. The apparatus of claim 3, wherein the first CSI-RS pattern is configured with: Up to four starting symbol positions in a slot; and There are up to eight symbols in total in the slot.

7. The apparatus of claim 1, wherein the first CSI-RS pattern comprises a superset of CSI-RS resources of the second CSI-RS pattern.

8. The apparatus of claim 7, wherein to obtain at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, the processor is configured to: deriving the second CSI-RS pattern from a plurality of CSI-RS patterns indicated by the configuration for data collection; and The first CSI-RS pattern is obtained by aggregating the multiple CSI-RS patterns.

9. The apparatus according to claim 8, wherein the processor is further configured to: A resource bundling configuration is received, wherein the aggregation is based on the resource bundling configuration.

10. The apparatus of claim 8, wherein the plurality of CSI-RS patterns are multiplexed in at least one of frequency or time.

11. The apparatus of claim 8, wherein the plurality of CSI-RS patterns have the same quasi-co-located (QCL) reference signal and type, or share one or more common QCL parameters with each other.

12. The apparatus of claim 8, wherein a common CSI-RS port is transmitted in both the first CSI-RS pattern and the second CSI-RS pattern.

13. The apparatus of claim 8, wherein a first set of CSI-RS ports is transmitted in the first CSI-RS pattern, and a second set of CSI-RS ports is transmitted in the second CSI-RS pattern.

14. The device according to claim 1, wherein: The nominal data includes at least one of a downlink pre-decoder (Vest) or a downlink channel matrix (Hest), the downlink pre-decoder and the downlink channel matrix estimated based on channel measurements made according to the second CSI-RS pattern; and The target model output data includes a target downlink precoder (Vtarget), which is estimated based on channel measurements made according to the first CSI-RS pattern.

15. The device according to claim 14, wherein: The entity comprises a model server or a training entity; and For sending, the processor is configured to send at least the target model output data to an entity for training the ML model, including sending Vtarget and at least one of Vest or Hest.

16. The apparatus of claim 15, wherein the processor is further configured to send at least one of a signal-to-noise ratio (SNR) or a reference signal received power (RSRP) corresponding to a CSI-RS received according to at least one of the first CSI-RS pattern or the second CSI-RS pattern.

17. The device according to claim 1, wherein: The nominal data is used as input to the ML model; and The target model output data is labeled sample by sample as the ground truth of the ML model.

18. An apparatus for wireless communication, the apparatus comprising: Memory; and a processor coupled to the memory, the processor being configured to: receiving signaling indicating a configuration of a channel state information (CSI) reference signal (CSI-RS) pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the CSI-RS pattern; generating nominal data for training the ML model by adding artificial noise to the target model output data; as well as At least the target model output data is sent to an entity for training the ML model.

19. The apparatus of claim 18, the processor further configured to send an indication of statistics of the artificial noise.

20. The apparatus of claim 19, wherein the statistical data comprises at least one of a variance or a ratio between a channel estimate power and a noise variance.

21. The apparatus of claim 19, wherein the statistical data is different for different data samples of the nominal data.

22. The apparatus of claim 18, wherein to generate the nominal data, the processor is configured to generate a plurality of sets of nominal data from common target output data, wherein each set is generated using specific noise statistics.

23. The apparatus of claim 18, wherein: The nominal data includes at least one of a downlink pre-decoder (Vest) or a downlink channel matrix (Hest), the downlink pre-decoder and the downlink channel matrix being estimated based on channel measurements made according to the CSI-RS pattern; and The target model output data includes a target downlink precoder (Vtarget), which is estimated based on channel measurements made according to the CSI-RS pattern.

24. The apparatus of claim 23, wherein: The entity comprises a model server or a training entity; and To send at least the target model output data to an entity for training the ML model, the processor is configured to send Vtarget and at least one of Vest or Hest.

25. The apparatus of claim 24, the processor further configured to transmit at least one of a signal-to-noise ratio (SNR) or a reference signal received power (RSRP) corresponding to the CSI-RS received according to the CSI-RS pattern.

26. The apparatus of claim 18, wherein: The nominal data is used as input to the ML model; and The target model output data is labeled sample by sample as the ground truth of the ML model.

27. A method of wireless communication by a user equipment (UE), the method comprising: receiving signaling indicating configuration of a channel state information (CSI) reference signal (CSI-RS) resource; obtaining at least a first CSI-RS pattern and a second CSI-RS pattern from the configuration, wherein the first CSI-RS pattern indicates a higher density of CSI-RSs in a channel frequency range than the second CSI-RS pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the first CSI-RS pattern; collecting nominal data for training the ML model based on channel measurements made according to the second CSI-RS pattern; as well as At least the target model output data is sent to an entity for training the ML model.

28. A method of wireless communication by a user equipment (UE), the method comprising: receiving signaling indicating a configuration of a channel state information (CSI) reference signal (CSI-RS) pattern; collecting target model output data for a machine learning (ML) model based on channel measurements made according to the CSI-RS pattern; generating nominal data for training the ML model by adding artificial noise to the target model output data; as well as At least the target model output data is sent to an entity for training the ML model.

Images