Method and apparatus for channel state information reporting using autocoder

Abstract

A method of channel state information (CSI) reporting performed by a wireless device. The wireless device has one or more encoders of one or more autoencoders available. The method includes generating a CSI report using an automatic encoder. The method further includes segmenting an output of the autoencoder into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different portions of uplink control information (UCI). Also disclosed are related methods for a radio access node, and related wireless devices, radio access nodes, computer programs and computer program products.

Description

Technical Field

[0001] This disclosure broadly relates to wireless communications, and more specifically to methods for compressing channel state information. Related apparatus is also disclosed. Background Technology

[0002] The 3rd Generation Partnership Project (3GPP) has developed and is developing standards for fourth-generation (4G) (also known as Long Term Evolution (LTE)) and fifth-generation (5G) (also known as New Radio (NR)) wireless communication systems. These systems, among other features, provide broadband communication between network nodes (such as base stations) and mobile wireless devices, as well as communication between network nodes and between wireless devices. 3GPP is also developing standards for sixth-generation (6G) wireless communication networks.

[0003] NR uses Orthogonal Frequency Division Multiplexing (OFDM) with configurable bandwidth and subcarrier spacing to effectively support a diverse set of use cases and deployment scenarios. Compared to LTE, NR improves upon LTE in terms of deployment flexibility, user throughput, latency, and reliability. NR also brings enhanced support for spatial multiplexing, where time-frequency resources are spatially shared across users, commonly known as multi-user MIMO (MU-MIMO).

[0004] Figure 1 The diagram illustrates MU-MIMO operation, where... N TX A multi-antenna network node with multiple antenna ports is transmitting information in space to several wireless devices, where the sequence... S (1) Designed for use in wireless devices UE (1), sequence S (2) Intended for use with wireless devices (UE) (2), and so on. Precoding is performed before each transmission is modulated and transmitted. It is applied to each sequence to spatially separate the transmissions, thereby mitigating multiplexing interference.

[0005] On the receiver side, each wireless device demodulates its received signal and combines the received antenna signals to obtain an estimate of the transmitted sequence. The estimate It can be described mathematically as: .

[0006] In equation 1, the term It can be approximated as an identity matrix, and the second term represents the spatial multiplexing interference seen by the UE(i).

[0007] The goal of MU-MIMO network nodes is to build a set of precoders. This makes the norm Large norm Small. In other words, precoder. Channel observed by UE(i) It correlates well, but its correlation with other channels is poor.

[0008] To build a precoder for efficient MU-MIMO transmission, network nodes may need to acquire channels. Detailed knowledge of the channel is required. In deployments where channel reciprocity is established, channel knowledge can be obtained from Sounding Reference Signals (SRSs), which are transmitted on demand or periodically by active wireless devices. Based on these SRSs, network nodes estimate... However, when channel reciprocity is not met or SRS coverage is limited, active radio devices may need to feed back channel details to network nodes. In NR (and in LTE), this is accomplished by having network nodes periodically transmit a Channel State Information Reference Signal (CSI-RS), from which the radio device can estimate its channel. The radio device then reports the CSI, from which the network node can determine a suitable precoder for MU-MIMO.

[0009] However, transmitting CSI from wireless devices to network nodes is costly in terms of bandwidth and computing power. Therefore, developing methods for efficiently compressing and transmitting CSI is of great interest. Summary of the Invention

[0010] Some embodiments advantageously provide methods, systems, and apparatus for RRC signaling and CSI reporting for AI-based CSI compression and feedback.

[0011] This paper discloses a method for RRC signaling and for reporting AI-based CSI on the UCI, which includes the segmentation of CSI reports and the mapping order of CSI reports to UCI bit sequences.

[0012] This paper describes a method for feeding back CSI reports on UCI, in which a wireless device processes feature vectors of each transport layer using an AI model to generate CSI reports.

[0013] According to a first aspect, there is a method for reporting Channel State Information (CSI), the method being performed by a wireless device. The wireless device has one or more encoders of one or more available autoencoders. The method includes using the encoders to generate a CSI report. The method includes segmenting the output of the encoders into a first CSI report and a second CSI report. The first CSI report and the second CSI report are transmitted on different portions of Uplink Control Information (UCI).

[0014] According to one embodiment of the first aspect, the method further includes receiving an indication of obtaining a CSI report using an encoder via Radio Resource Control (RRC) signaling.

[0015] According to one embodiment of the first aspect, an encoder is used to generate a CSI report including a preprocessed estimated channel.

[0016] According to an embodiment of the first aspect, the preprocessing of the estimated channel includes: extracting feature vectors for each transport layer from the estimated channel; and reducing the dimensionality of the feature vectors extracted for each transport layer by applying preprocessing in the spatial, frequency, and / or time domains.

[0017] According to one embodiment of the first aspect, using an encoder to generate a CSI report includes compressing and quantizing the preprocessed feature vectors of each transport layer using an autoencoder.

[0018] According to one embodiment of the first aspect, the spatial, frequency, and / or time domain information forms part of the CSI report on the UCI.

[0019] According to one embodiment of the first aspect, the quantized bits form part of the CSI report described on the UCI.

[0020] According to an embodiment of the first aspect, the method further includes receiving parameters associated with the content and / or size of the encoded CSI via RRC signaling. The parameters include one or more of the following: a limitation on the rank of the encoded CSI; an indication of the spatial, frequency, and / or time-domain basis for preprocessing the feature vectors of each transport layer; an indication of the quantization bits that the wireless device can use to quantize the preprocessed feature vectors of each transport layer; an indication of the number of potential spatial coefficients of the encoded CSI for each transport layer; and an indication of whether the CSI report is generated at the wireless device using transport layer common processing or transport layer specific processing.

[0021] According to one embodiment of the first aspect, one or more of the parameters are explicitly associated with an autoencoder deployed at the wireless device.

[0022] According to one embodiment of the first aspect, a model identifier uniquely associated with one of the automatic encoders available to the wireless device is transmitted to the wireless device via a signal.

[0023] According to one embodiment of the first aspect, one or more of the parameters are dynamically configured on the downlink control information (DCI) and / or the media access control (MAC) control element (CE).

[0024] According to one embodiment of the first aspect, one or more of the parameters are configured by the wireless device and reported as part of the CSI report on the UCI.

[0025] According to one embodiment of the first aspect, the first CSI report includes one or more of the following: a regular CSI report volume; information about the autoencoder used to generate the CSI report; auxiliary information about the AI-based quantization bits; and a basis for preprocessing.

[0026] According to an embodiment of the first aspect, the second CSI report includes one or more of the following: an index of the spatial, frequency, and / or temporal dimensions of the feature vectors of each transport layer as a preprocessing basis; transport layer-specific information required by the network node to decode the quantized bits of each transport layer; and quantized bits of each transport layer from the output of the autoencoder.

[0027] According to a second aspect, there is a method for Channel State Information (CSI) reporting, performed by a network node. The network node has one or more decoders of one or more available autoencoders. The method includes instructing a radio device to transmit a CSI report compressed by the autoencoder via Radio Resource Control (RRC) signaling. The method includes receiving a first CSI report and a second CSI report from different portions of Uplink Control Information (UCI) from the radio device. The method includes decoding the first CSI report and the second CSI report using a decoder from one or more of the decoders.

[0028] According to an embodiment of the second aspect, the method further includes transmitting parameters associated with the content and / or size of the encoded CSI via RRC signaling. The parameters include one or more of the following: a limitation on the rank of the encoded CSI; an indication of the spatial, frequency, and / or time-domain basis for preprocessing the feature vectors of each transport layer; an indication of the quantization bits that the wireless device can use to quantize the preprocessed feature vectors of each transport layer; an indication of the number of potential spatial coefficients of the encoded CSI for each transport layer; and an indication of whether the CSI report is generated at the wireless device using transport layer common processing or transport layer specific processing.

[0029] According to one embodiment of the second aspect, one or more of the parameters are explicitly associated with an autoencoder deployed at the network node.

[0030] According to one embodiment of the second aspect, a model identifier uniquely associated with one of the automatic encoders available to the wireless device is sent to the wireless device via a signal.

[0031] According to one embodiment of the second aspect, one or more of the parameters are dynamically configured on the downlink control information (DCI) and / or the media access control (MAC) control element (CE).

[0032] According to one embodiment of the second aspect, one or more of the parameters are configured by the wireless device and received by the network node as part of the CSI report on the UCI.

[0033] According to an embodiment of the second aspect, the first CSI report includes one or more of the following: a regular CSI report volume; information about the autoencoder used to generate the CSI report; auxiliary information about the AI-based quantization bits; and a basis for preprocessing.

[0034] According to an embodiment of the second aspect, the second CSI report includes one or more of the following: an index of the spatial, frequency, and / or temporal dimensions of the feature vectors of each transport layer as a preprocessing basis; transport layer-specific information required by the network node to decode the quantized bits of each transport layer; and quantized bits of each transport layer from the output of the autoencoder.

[0035] According to a third aspect, there exists a wireless device configured to perform a method for reporting Channel State Information (CSI). The wireless device has one or more encoders of one or more available autoencoders. The method includes using the encoders to generate CSI reports. The method also includes segmenting the output of the encoders into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different portions of Uplink Control Information (UCI).

[0036] According to one embodiment of the third aspect, the wireless device is also configured to perform the method according to any embodiment of the first aspect.

[0037] According to a fourth aspect, there exists a wireless device configured to perform a method for reporting Channel State Information (CSI). The wireless device includes a processing circuit module and a memory. The wireless device has one or more encoders of one or more available autoencoders. The method includes using the encoders to generate CSI reports. The method includes segmenting the output of the encoders into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different portions of Uplink Control Information (UCI).

[0038] According to one embodiment of the fourth aspect, the wireless device is also configured to perform the method according to any embodiment of the first aspect.

[0039] According to a fifth aspect, there exists a radio access node in a communication network configured to perform a method for reporting Channel State Information (CSI). The network node has one or more decoders of one or more available autoencoders. The method includes instructing a radio device to transmit a CSI report compressed by an autoencoder via Radio Resource Control (RRC) signaling. The method includes receiving a first CSI report and a second CSI report from different portions of Uplink Control Information (UCI) from the radio device. The method includes decoding the first CSI report and the second CSI report using a decoder from one or more of the decoders.

[0040] According to one embodiment of the fifth aspect, the radio access node is further configured to perform the method according to any embodiment of the second aspect.

[0041] According to a sixth aspect, there exists a radio access node in a communication network configured to perform a method for reporting Channel State Information (CSI). The network node includes a processing circuit module and a memory. The network node has one or more decoders of one or more available autoencoders. The method includes instructing a radio device to transmit a CSI report compressed by an autoencoder via Radio Resource Control (RRC) signaling. The method includes receiving a first CSI report and a second CSI report from different portions of Uplink Control Information (UCI) from the radio device. The method includes decoding the first CSI report and the second CSI report using a decoder from one or more of the decoders.

[0042] According to one embodiment of the sixth aspect, the radio access node is further configured to perform the method according to any embodiment of the second aspect.

[0043] According to the seventh aspect, there exists a computer program including machine-readable instructions that, when executed by a processor of a wireless device, cause the wireless device to perform a method according to any embodiment of the first aspect.

[0044] According to the eighth aspect, there exists a computer program product comprising a non-transitory computer-readable storage medium on which the computer program according to the seventh aspect is stored.

[0045] According to the ninth aspect, there is a computer program including machine-readable instructions that, when executed by a processor of a radio access node, cause the radio access node to perform a method according to any embodiment of the second aspect.

[0046] According to the tenth aspect, there exists a computer program product comprising a non-transitory computer-readable storage medium on which the computer program according to the ninth aspect is stored. Attached Figure Description

[0047] A more complete understanding of the presented embodiments and their accompanying advantages and features will be more readily obtained by referring to the following detailed description in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating multi-MU-MIMO operation; Figure 2 This is a diagram of CSI Type II feedback; Figure 3 This is a diagram of a fully connected automatic encoder; Figure 4 This is a diagram illustrating CSI compression using an automatic encoder; Figure 5 This is a flowchart of performing quantization operations on the encoder output to adapt the CSI payload to the air interface; Figure 6 This is a diagram illustrating the implicit feedback preprocessing based on the eigenvectors of the estimated transport rank. Figure 7 This is a diagram of the common model of the transport layer; Figure 8 This is a diagram of a specific model of the transport layer; Figure 9 This is a schematic diagram of an example network architecture based on the principles of this disclosure, illustrating a communication system connected to a host computer via an intermediate network; Figure 10 This is a block diagram illustrating how a host computer communicates with a wireless device via a network node through at least a partial wireless connection, according to some embodiments of this disclosure. Figure 11 This is a flowchart illustrating an example method for executing a client application at a wireless device, implemented in a communication system according to some embodiments of the present disclosure, the communication system including a host computer, a network node, and a wireless device; Figure 12 This is a flowchart illustrating an example method for receiving user data at a wireless device, implemented in a communication system according to some embodiments of the present disclosure, the communication system including a host computer, a network node, and a wireless device; Figure 13 This is a flowchart illustrating an example method for receiving user data from a wireless device at a host computer, implemented in a communication system according to some embodiments of the present disclosure, the communication system including a host computer, a network node, and a wireless device; Figure 14This is a flowchart illustrating an example method for receiving user data at a host computer, implemented in a communication system according to some embodiments of the present disclosure, the communication system including a host computer, a network node, and a wireless device; Figure 15A This is a flowchart of an example process in a network node according to some embodiments of this disclosure; Figure 15B This is a flowchart of an example process in a network node according to some embodiments of this disclosure; Figure 16A This is a flowchart of an example process in a wireless device according to some embodiments of the present disclosure; Figure 16B This is a flowchart of an example process in a wireless device according to some embodiments of the present disclosure; and Figure 17 This is a schematic diagram of an example architecture of a channel feature vector feedback method according to some embodiments of the present disclosure. Detailed Implementation

[0048] In the figures, hardware components are indicated by conventional symbols in their appropriate places. Only those specific details relevant to understanding the embodiments are shown so as not to obscure this disclosure due to details that will be readily understood by one of ordinary skill in the art upon which the description herein is given. Similar reference numerals are used throughout the description to indicate similar elements.

[0049] As used herein, relational terms (such as “first” and “second”, “top” and “bottom”, and the like) may be used only to distinguish one entity or element from another, and do not necessarily require or imply any physical or logical relationship or order between such entities or elements. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the concepts described herein. As used herein, the singular forms “a” (“a”, “an”) and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It will be further understood that the terms “comprises” and / or “comprising” (“includes” and / or “including”) as used herein specify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or combinations thereof.

[0050] In the embodiments described herein, the connecting terms “communicating with” and similar expressions can be used to indicate electrical or data communication, which may be accompanied, for example, by physical contact, induction, electromagnetic radiation, radio signaling, infrared signaling, or optical signaling. Those skilled in the art will appreciate that multiple components can interoperate, and modifications and variations in electrical and data communication are possible.

[0051] In some embodiments described herein, the terms “coupled,” “connected,” and the like may be used herein to indicate a connection, although not necessarily a direct one, and may include wired and / or wireless connections.

[0052] As used herein, the term "network node" can refer to any type of network node included in a radio network, and may also include any of the following: base station (BS), radio base station, base transceiver station (BTS), base station controller (BSC), radio network controller (RNC), g node B (gNB), evolved node B (eNB or eNodeB), node B, multi-standard radio (MSR) radio node (such as MSR BS), multi-cell / multicast coordination entity (MCE), integrated access and backhaul (IAB) node, relay node, donor node control relay, radio access point (AP), transport point, transport node, remote radio unit (RRU), remote radio head (RRH), core network node (e.g., mobility management entity (MME), ad hoc network (SON) node, coordination node, location node, MDT node, etc.), external node (e.g., third-party node, node outside the current network), node in distributed antenna system (DAS), spectrum access system (SAS) node, element management system (EMS), etc. Network nodes may also include test equipment. The term “radio node” as used in this article can also be used to refer to a wireless device (WD), such as a wireless device or a node in a radio network.

[0053] In some embodiments, the non-limiting terms "wireless device" and "user equipment (UE)" are used interchangeably. A wireless device as used herein can be any type of wireless device capable of communicating with a network node or another wireless device via radio signals, such as a wireless device-to-wireless device. A wireless device can also be a radio communication device, a target device, a device-to-device (D2D) wireless device, a machine-type wireless device or a wireless device capable of machine-to-machine (M2M) communication, a low-cost and / or low-complexity wireless device, a sensor equipped with a wireless device, a tablet computer, a mobile terminal, a smartphone, a laptop embedded device (LEE), a laptop installed device (LME), a USB dongle, a client premises equipment (CPE), an Internet of Things (IoT) device, or a narrowband IoT (NB-IoT) device, etc.

[0054] Furthermore, in some embodiments, the generic term "radio network node" is used. It can be any type of radio network node, which may include any of the following: base station, radio base station, base transceiver station, base station controller, network controller, RNC, evolved Node B (eNB), Node B, gNB, multi-cell / multicast coordination entity (MCE), IAB node, relay node, access point, radio access point, remote radio unit (RRU), remote radio head (RRH).

[0055] Note that the concepts of "network" and / or network node can be understood as a general network node, gNB, base station, unit within a base station used to process at least some ML operations, relay node, core network node, core network node that processes at least some ML operations, or device that supports D2D communication. Nodes can be deployed in 5G or 6G networks.

[0056] Note that although terms from a particular wireless system (such as, for example, 3GPP LTE and / or New Radio (NR)) may be used in this disclosure, this should not be construed as limiting the scope of this disclosure to only the aforementioned systems. Other wireless systems, including but not limited to Wideband Code Division Multiple Access (WCDMA), Global Microwave Access Interoperability (WiMax), Ultra Mobile Broadband (UMB), and Global System for Mobile Communications (GSM), may also benefit from utilizing the concepts covered in this disclosure.

[0057] In some embodiments, the general description element in the form of "one of A and B" corresponds to A or B. In some embodiments, at least one of A and B corresponds to A, B, or AB, or to one or more of A and B, or to one or both of A and B. In some embodiments, at least one of A, B, and C corresponds to one or more of A, B, and C, and / or to A, B, C, or a combination thereof.

[0058] It should be noted further that the functions performed by wireless devices or network nodes as described herein can be distributed across multiple wireless devices and / or network nodes. In other words, it is conceivable that the functions of the network nodes and wireless devices described herein are not limited to being performed by a single physical device, and can actually be distributed across several physical devices.

[0059] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It will also be further understood that terms used herein should be interpreted as having the same meaning as they have in the context of this specification and the relevant field, and will not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0060] Some implementations provide AI-based CSI compression and feedback for RRC signaling and CSI reporting.

[0061] In NR, the CSI feedback mechanism for Multi-User Multiple-Input Multiple-Output (MU-MIMO) operation is called CSI Type II, where the radio device reports CSI feedback with high CSI resolution (e.g., as specified in 3GPP TS 38.214). It is based on a specified set of Direct Fourier Transform (DFT) basis functions (a grid of beams), from which the radio device selects those beams that best suit its channel conditions (e.g., Classical Codebook Precoding Matrix Indicators (PMI)). The number of beams reported by the radio device is configurable via Radio Resource Control (RRC) signaling and can be 2 or 4 (e.g., as specified in 3GPP Technical Specification (TS) Rel-15 Type II), or 2, 4, or 6 (e.g., as specified in TS Rel-16 Type II). In Rel-16 Type II, the CSI report can be further compressed in the frequency domain (FD), where the set of FD DFT basis vectors is selected by the radio device. The number of selected FD basis vectors is a function of the following terms: the number of CQI subbands, the number of PMI subbands per Channel Quality Indicator (CQI) subband, and the ratio that determines FD compression (which can be called...). For example, as specified in 3GPP TS 38.214, where v This is a layer index), configured by the network node via RRC signaling. Additionally, the radio device reports non-zero coefficients (NZCs) associated with the selected beams of Rel-15 Type II, informing the network node how these beams should be combined according to the relative amplitude scaling and in-phase of each subband. In Rel-16, the reported NZCs are then associated with the selected beam and FD basis vectors. In Rel-16, to further compress CSI reports, the network node also configures a layer index (CSI) for the radio device via RRC signaling. β The ratio determines the maximum number of NZCs to be reported. For example, for a single-layer transmission configured with 2L beams and M FD basis vectors by network nodes, there are a total of 2LM linear combination coefficients. Then, at most ⌈ 2LMβ ⌉NZC, the rest 2LM -⌈ 2LMβ ⌉ is considered zero and is not reported. The selected beam is typically used for all subbands and all transport layers, while NZC (for Rel-15 and Rel-16 Type II) and FD basis vectors (for Rel-16 Type II) are layer-specific.

[0062] To further explain the structure of Type II CSI, an example of Rel-15 Type II CSI is shown in Figure 2(In the CSI Type II feedback diagram, the DFT beam vector is shown.) b n and their relative amplitude The choice of in-phase parameter is determined from a broadband perspective, while in-phase parameter is per-subband. Here, broadband means that the selected DFT beam vector is the same for all subcarriers used in OFDM transmission, while subband means that the in-phase parameter is determined on a subset of consecutive subcarriers. The in-phase parameter is quantized such that… Obtained from QPSK or 8PSK signal constellations.

[0063] by k To represent the subband index, the precoder reported by the wireless device can be represented as... .

[0064] Note that the reporting overhead for Type II CSI is typically large, especially compared to Type I CSI. The majority of the reporting overhead comes from subband reporting, such as layer-specific NZC. For example, reporting the phase and amplitude of a coefficient requires approximately 7 bits (the actual number depends on the release and parameter configuration).

[0065] CSI Report in NR In NR, a wireless device can be configured with one or more CSI reporting settings, each configured by a higher-level parameter CSI-ReportConfig. Each CSI-ReportConfig is associated with a Bandwidth Part (BWP) and contains one or more of the following: CSI resource configuration for channel measurements CSI Interference Measurement (CSI-IM) Resource Configuration for Interference Measurement Report configuration type, namely aperiodic CSI (on the Physical Uplink Shared Channel (PUSCH), periodic CSI (on the Physical Uplink Control Channel (PUCCH)), or semi-persistent CSI on PUCCH or PUSCH. Reporting quantities specify what needs to be reported, such as the Rank Indicator (RI), PMI, and CQI. Codebook configuration, such as Type I or Type II CSI Frequency domain configuration, i.e., subband relative to broadband CQI or PMI, and subband size. CQI table to use The wireless device may be configured with one or more CSI resource configurations for channel measurements and one or more CSI-IM resources for interference measurements. Each CSI resource configuration for channel measurements may contain one or more NZP CSI-RS resource sets. For each NZP CSI-RS resource set, it may further contain one or more NZP CSI-RS resources. NZP CSI-RS resources may be periodic, semi-persistent, or aperiodic.

[0066] Similarly, each CSI-IM resource configuration used for interference measurements may contain one or more sets of CSI-IM resources. Each CSI-IM resource set may further contain one or more CSI-IM resources. CSI-IM resources may be periodic, semi-persistent, or aperiodic.

[0067] Type II CSI Report on PUSCH After successfully decoding downlink control information (DCI) format 0_1 ​​or DCI format 0_2, the wireless device uses PUSCH to perform an aperiodic CSI report, which triggers the aperiodic CSI trigger state.

[0068] When two PUSCHs are allocated in DCI format 0_1, the aperiodic CSI report is carried on the second allocated PUSCH. When more than two PUSCHs are allocated in DCI format 0_1, the aperiodic CSI report is carried on the penultimate allocated PUSCH.

[0069] After successfully decoding DCI format 0_1 ​​or DCI format 0_2, the wireless device performs a semi-persistent CSI report on the PUSCH, which activates the semi-persistent CSI trigger state. DCI format 0_1 ​​and DCI format 0_2 contain a CSI request field that indicates whether the semi-persistent CSI trigger state should be activated or deactivated. PUSCH resources and MCS are allocated by the uplink DCI in a semi-persistent manner.

[0070] CSI reports on the PUSCH can be multiplexed with uplink data on the PUSCH. CSI reports on the PUSCH can also be performed without any multiplexing with uplink data from the wireless device.

[0071] Type II CSI Report, Parts 1 and 2 For CSI feedback on Rel-15 Type II and Rel-16 Type II (also known as Enhanced Type II or e Type II) on the PUSCH, the CSI report consists of two parts: Part 1 and Part 2. One reason for dividing the CSI report into Part 1 and Part 2 is to address dynamically changing CSI payloads. For example, based on a time-varying channel, a wireless device may report different ranks throughout the connection, which significantly impacts the actual required CSI payload size. To allow network nodes to understand the actual payload size, the network node first decodes Part 1, which has a fixed payload size and carries information used to calculate the payload size for Part 2. For Type II CSI feedback in Rel-15, Part 1 contains the RI (if reported), CQI, and an indication of the number of non-zero bandwidth amplitude coefficients per tier for Type II CSI (see, for example, Clause 5.2.2.2.3 in 3GPP TS 38.214). The fields of Part 1—RI (if reported), CQI, and the indication of the number of non-zero bandwidth amplitude coefficients per tier—are encoded separately. Part 2 contains the PMI for Type II CSI. Parts 1 and 2 are encoded separately. For Rel-16 Type II CSI feedback, Part 1 contains an indication of the RI, CQI, and the total number of cross-layer non-zero amplitude coefficients for Rel-16 Type II CSI (see, for example, Clause 5.2.2.2.5 in 3GPP TS 38.214). The fields of Part 1—RI, CQI, and the indication of the total number of cross-layer non-zero amplitude coefficients—are encoded separately. Part 2 contains the PMI for Enhanced Type II CSI. Parts 1 and 2 are encoded separately.

[0072] Autoencoder for AI / ML Enhanced CSI Reports Recently, neural network (NN)-based autoencoders (AEs) have been used to compress downlink MIMO channel estimation for uplink feedback.

[0073] Furthermore, 3GPP has decided to launch a research project for Rel.18, which includes use cases for AI-based CSI reporting, with AE (Artificial Neural Network) being part of this research. Specifically, AE is an artificial neural network capable of compressing and decompressing data in an unsupervised manner, typically with high fidelity.

[0074] Figure 3 This demonstrates a low-complexity fully connected (dense) action equation (AE). The AE is divided into two parts: - Encoder (used to compress input data) X ),as well as - Decoder (used to decompress input data).

[0075] AEs can have different architectures. For example, AEs can be based on dense neural networks, multidimensional convolutional neural networks, variational neural networks, recurrent neural networks, transformer networks, or any combination thereof. However, all AE architectures possess... Figure 3 The encoder-bottleneck-decoder structure is shown in the figure.

[0076] AE's code (in) Figure 3 Chinese Y The size of (to represent) is usually much smaller than the input data ( Figure 3 In X The size of the input features. Therefore, the AE encoder will input the features. X The dimensions are reduced to Y The decoder part of AE attempts to invert the encoder and reconstruct the image with minimal error. X (Based on a predefined loss function).

[0077] Figure 4 This illustrates how AE in NR can be used for AI / ML-enhanced CSI reporting. The wireless device uses CSI-RS to measure the channel in the downlink. The wireless device estimates the channel for each subcarrier (SC) from the transmit (TX) antenna at each network node and the receive (RX) antenna at each wireless device. This estimation can be viewed as a three-dimensional (3D) channel matrix. This 3D channel matrix represents the estimated MIMO channel over several SCs and is input to the encoder.

[0078] An AE encoder is implemented in a wireless device, and an AE decoder is implemented in a network, such as a network node. The output of the AE encoder is transmitted from the wireless device to the network node via an uplink. The codeword can be viewed as a potential representation of the channel being learned. The architecture of the AE (e.g., number of layers, nodes per layer, activation function) typically needs to be numerically optimized via a process called hyperparameter tuning for CSI reporting. When optimizing the AE architecture, data properties (e.g., CSI-RS channel estimation), channel size, uplink feedback rate, and hardware limitations of the encoder and decoder may all need to be considered.

[0079] Train the weights and biases of the AE (with a fixed architecture) to minimize the reconstruction error (input) on a given training dataset. X and output (The error between them). For example, weights and biases can be trained to minimize the mean squared error (MSE). Model training is typically performed on a large training dataset using a variant of the gradient descent algorithm. To achieve good performance during live training, the training dataset should represent the actual data that the AE will encounter during live training.

[0080] In bilateral CSI compression, the output of the encoder on the wireless device side needs to be transmitted to the network node decoder via the air interface as an assigned CSI report payload, and therefore needs to be quantized to a limited number of bits (e.g., 1-4 bits per sample for UCI) for efficient transmission, such as... Figure 5 The diagram illustrates the quantization operation at the encoder output used to adapt the CSI payload to the air interface. Therefore, quantization layers are typically connected to the encoder output or are included directly within the encoder. In one example, the quantization layer implements scalar quantization, quantizing the output of each neuron in the encoder output layer (the bottleneck layer of the AE) to generate bits that match the CSI report payload in the UCI. Other quantization methods, such as vector quantization, can also be used.

[0081] Preprocessing of input data to AE Appropriate preprocessing of the input to the encoder can significantly reduce the size and complexity of designing and / or training AI / ML models, while simultaneously improving model scalability and transferability. In CSI compression, preprocessing methods can include transforming the channel from the antenna frequency domain to the beam delay domain, or from the antenna frequency time domain to the beam delay Doppler domain. Furthermore, preprocessing is used to reduce the need for multiple models, which depends on variations in bandwidth and the number of antenna ports at network nodes.

[0082] To further explain this, the channel representation in the antenna frequency domain is typically rich and difficult to compress; however, its equivalent in the beam delay domain is sparse and much easier to compress. This sparsity reflects, to some extent, the physical interpretation of the propagation channel. That is, it reflects how numerous sinusoidal signals propagate from the transmitter to the receiver along different paths. Essentially, each beam can be associated with a certain direction of the propagation path, and if the signals propagate along different paths, each delay can reflect a relative difference in distance. If there exists infinite spatial resolution and delay resolution, then each pair of beams and delays can be associated with a single propagation path.

[0083] In real-world propagation environments, when considering the entire 3D space, the dominant paths contributing to signal transmission are typically sparse because signals cannot reach the receiver from any direction. This is limited, among other reasons, by the number and directivity of antenna elements deployed at both the transmitter and receiver, and the number of objects in the propagation environment capable of reflecting signals without introducing significant loss. This sparsity can be leveraged to assist AI / ML models. For example, beam delay domain transformation can aid AI / ML models through initial feature extraction. Another advantage of this preprocessing is that beam delay transformation can be obtained using Fast Fourier Transform (FFT), for which fast hardware-supported implementations are readily available. Sparsity can be further utilized by removing multiple insignificant beams and delays, allowing for the reduction of input dimensionality with minimal loss, potentially leading to smaller AI / ML models. Beam delay transformation and feature extraction can be applied to both explicit channel feedback and feature vector-based feedback.

[0084] The following is a brief example of preprocessing feature vector-based feedback, which has received considerable attention in 3GPP. The first step is for the radio device to measure the channel on CSI-RS. For example, assume the radio device has 4 Rx-ports, the configured CSI format has 32 virtual Tx-ports, and the bandwidth is 52 resource blocks (RBs), corresponding to 10 MHz at a 15 kHz subcarrier spacing. Feature extraction for feature vector-based feedback is then performed... Figure 6 The steps are shown below: 1. The wireless device performs a spatial domain DFT on the 32x4 matrix of each RB and selects L of the strongest beams (for one polarization) from the 16 beams (box S10). This is done in a wideband manner, including spatial oversampling based on the spatial domain (SD), and the same beam is used for both polarizations. The covariance of the beam spatial channel is summed over, for example, 4 RBs to produce the covariance matrix for each subband. 2. For each covariance matrix (each subband), the device extracts multiple eigenvectors and can select the rank, i.e., the number of layers (box S12). 3. The device performs a frequency-domain DFT on each layer, transforms it to the delay domain, and then selects... M The strongest tap (box S12). The resulting dimension is 2L × number of layers × M The tensor is called the linear combination coefficient and can be used to reconstruct the precoding matrix by the proposed wireless device. 4. The tensor of the linear combination coefficients is used as input to the AI / ML model (box S14). This input can be further enhanced with information about the selected beam and taps, noise level, etc.

[0085] Implicit CSI feedback in AE model (for) RI >1) When RI > 1, i.e., the rank of the indication is greater than 1, several schemes of AI / ML models can exist for implicit CSI feedback. This disclosure focuses on two main classifications of AI / ML models when RI > 1: 1. Common Transport Layer Scheme: This scheme includes an AI / ML model that is trained based on estimated RI and deployed across all transport layers, such as... Figure 7 As shown, where H and H I These are the estimated transmission channel and the interference channel, respectively. 2. Transport Layer Specific Scheme: This scheme includes multiple AI / ML models that are trained and deployed for each transport layer based on estimated RI, such as... Figure 8 As shown, where H and H I These are the estimated transmission channel and the interference channel, respectively.

[0086] Note that in the example above, the model was trained for the transport layer and is independent of RI. Furthermore, the model used for the transport layer can also depend on RI. However, the CSI reporting mechanism in UCI can be applied to both cases where the model is independent of RI or depends on RI.

[0087] Some embodiments of this disclosure provide AI-based CSI compression and feedback for RRC signaling and CSI reporting.

[0088] Referring now to the drawn diagram, similar elements are identified by similar reference numerals. Figure 9The diagram illustrates a communication system 10 (such as a 3GPP-type cellular network supporting standards such as LTE and / or NR (5G)) according to an embodiment, the communication system including an access network 12 (such as a radio access network) and a core network 14. The access network 12 includes a plurality of network nodes 16a, 16b, 16c (collectively referred to as network nodes 16), such as NBs, eNBs, gNBs, or other types of wireless access points, each defining a corresponding coverage area 18a, 18b, 18c (collectively referred to as coverage area 18). Each network node 16a, 16b, 16c can be connected to the core network 14 via a wired or wireless connection 20. A first wireless device 22a located in coverage area 18a is configured to wirelessly connect to or be paged by the corresponding network node 16a. A second wireless device 22b in coverage area 18b can wirelessly connect to the corresponding network node 16b. Although multiple wireless devices 22a, 22b (collectively referred to as wireless devices 22) are shown in this example, the disclosed embodiments are equally applicable to situations where a single wireless device is located within a coverage area or where a single wireless device is connected to a corresponding network node 16. It should be noted that although only two wireless devices 22 and three network nodes 16 are shown for convenience, the communication system may include more wireless devices 22 and network nodes 16.

[0089] It is also anticipated that the wireless device 22 can simultaneously communicate with more than one network node 16 and more than one type of network node 16, and / or be configured to communicate individually with more than one network node 16 and more than one type of network node 16. For example, the wireless device 22 can have dual connectivity with LTE-enabled network nodes 16 and the same or different NR-enabled network nodes 16. As an example, the wireless device 22 can communicate with an eNB of LTE / E-UTRAN and a gNB of NR / NG-RAN.

[0090] The communication system 10 itself may be connected to a host computer 24, which may be implemented as hardware and / or software of a standalone server, a cloud-based server, a distributed server, or as a processing resource in a server farm. The host computer 24 may be owned or controlled by a service provider, or may be operated by or on behalf of the service provider. Connections 26, 28 between the communication system 10 and the host computer 24 may extend directly from the core network 14 to the host computer 24, or may extend via an optional intermediate network 30. The intermediate network 30 may be one or more of a public, private, or hosted network. The intermediate network 30 (if any) may be a backbone network or the Internet. In some embodiments, the intermediate network 30 may include two or more subnetworks (not shown).

[0091] Figure 9The communication system as a whole enables connectivity between one of the connected wireless devices 22a and 22b and the host computer 24. This connectivity can be described as an over-the-top (OTT) connection. The host computer 24 and the connected wireless devices 22a and 22b are configured to transmit data and / or signaling via the OTT connection using access network 12, core network 14, any intermediate network 30, and other possible infrastructure (not shown) acting as intermediaries. The OTT connection can be transparent in the sense that at least some of the participating communication devices are unaware of the routing of uplink and downlink communications. For example, it may not be necessary to inform network node 16 of past routing of incoming downlink communications with data originating from host computer 24 and to be forwarded (e.g., handed over) to the connected wireless device 22a. Similarly, network node 16 does not need to know the future routing of outgoing uplink communications originating from wireless device 22a toward host computer 24.

[0092] Network node 16 is configured to include a configuration unit 32, which is configured to perform one or more of the network node 16 functions described herein, including functions related to RRC signaling and CSI reporting for AI-based CSI compression and feedback. Wireless device 22 is configured to include an implementation unit 34, which is configured to perform one or more of the wireless device 22 functions described herein, including functions related to RRC signaling and CSI reporting for AI-based CSI compression and feedback.

[0093] Now refer to Figure 10The following describes an exemplary implementation of the wireless device 22, network node 16, and host computer 24 described in the preceding paragraphs, according to an embodiment. In the communication system 10, the host computer 24 includes hardware (HW) 38, which includes a communication interface 40 configured to establish and maintain wired or wireless connections with interfaces of different communication devices of the communication system 10. The host computer 24 also includes a processing circuitry module 42, which may have storage and / or processing capabilities. The processing circuitry module 42 may include a processor 44 and a memory 46. In particular, as an addition to or alternative to the processor (such as a central processing unit) and memory, the processing circuitry module 42 may include integrated circuit modules for processing and / or control, such as one or more processors and / or processor cores suitable for executing instructions and / or FPGAs (Field-Programmable Gate Arrays) and / or ASICs (Application-Specific Integrated Circuit Modules). The processor 44 may be configured to access (e.g., write and / or read) memory 46, which may include any kind of volatile and / or non-volatile memory, such as cache memory and / or buffer memory and / or RAM (random access memory) and / or ROM (read-only memory) and / or optical memory and / or EPROM (erasable programmable read-only memory).

[0094] Processing circuit module 42 may be configured to control any methods and / or processes described and / or performed herein, and / or to cause such methods and / or processes to be performed, for example, by host computer 24. Processor 44 corresponds to one or more processors 44 for performing the functions of host computer 24 as described herein. Host computer 24 includes memory 46 configured to store data, programming software code, and / or other information as described herein. In some embodiments, software 48 and / or host application 50 may include instructions that, when executed by processor 44 and / or processing circuit module 42, cause processor 44 and / or processing circuit module 42 to perform the processes described herein for host computer 24. The instructions may be software associated with host computer 24.

[0095] Software 48 may be executable by processing circuitry module 42. Software 48 includes host application 50. Host application 50 may be operable to provide services to remote users, such as wireless device 22 connected via an OTT connection 52 terminated between wireless device 22 and host computer 24. In providing services to remote users, host application 50 may provide user data, which is transmitted using OTT connection 52. “User data” may be data and information described herein for implementing the aforementioned functionality. In one embodiment, host computer 24 may be configured to provide control and functionality to a service provider and may be operated by or on behalf of the service provider. Processing circuitry module 42 of host computer 24 enables host computer 24 to observe, monitor, control network node 16 and / or wireless device 22, and to transmit and / or receive data to and / or from network node 16 and / or wireless device 22. The processing circuit module 42 of the host computer 24 may include a control unit 54 configured to enable the service provider to observe / monitor / control network node 16 and / or wireless device 22 / to transmit and / or receive data from network node 16 and / or wireless device 22.

[0096] The communication system 10 also includes a network node 16, which is provided within the communication system 10 and includes hardware 58 enabling it to communicate with the host computer 24 and with the wireless device 22. Hardware 58 may include: a communication interface 60 for establishing and maintaining wired or wireless connections with different communication devices of the communication system 10; and a radio interface 62 for establishing and maintaining at least a wireless connection 64 with the wireless device 22 located within the coverage area 18 served by the network node 16. The radio interface 62 may be configured as or may include, for example, one or more RF transmitters, one or more RF receivers, and / or one or more RF transceivers. The communication interface 60 may be configured to facilitate a connection 66 to the host computer 24. The connection 66 may be direct, or it may traverse the core network 14 of the communication system 10 and / or one or more intermediate networks 30 outside the communication system 10.

[0097] In the illustrated embodiment, the hardware 58 of network node 16 further includes a processing circuitry module 68. The processing circuitry module 68 may include a processor 70 and memory 72. Specifically, as an addition to or alternative to the processor (such as a central processing unit) and memory, the processing circuitry module 68 may include integrated circuit modules for processing and / or control, such as one or more processors and / or processor cores suitable for executing instructions and / or FPGAs (Field-Programmable Gate Arrays) and / or ASICs (Application-Specific Integrated Circuit Modules). The processor 70 may be configured to access (e.g., write to and / or read from) memory 72, which may include any kind of volatile and / or non-volatile memory, such as cache memory and / or buffer memory and / or RAM (Random Access Memory) and / or ROM (Read-Only Memory) and / or optical memory and / or EPROM (Erasable Programmable Read-Only Memory).

[0098] Therefore, network node 16 further includes software 74, which is internally stored, for example, in memory 72, or stored in external memory (e.g., a database, storage array, network storage device, etc.) accessible by network node 16 via an external connection. Software 74 may be executable by processing circuitry module 68. Processing circuitry module 68 may be configured to control any methods and / or processes described herein, and / or to cause such methods and / or processes to be executed, for example, by network node 16. Processor 70 corresponds to one or more processors 70 for performing the functions of network node 16 as described herein. Memory 72 is configured to store data, programming software code, and / or other information as described herein. In some embodiments, software 74 may include instructions that, when executed by processor 70 and / or processing circuitry module 68, cause processor 70 and / or processing circuitry module 68 to perform the processes described herein for network node 16. For example, the processing circuitry module 68 of network node 16 may include a configuration unit 32 configured to perform one or more network node 16 functions as described herein, including functions related to RRC signaling and CSI reporting for AI-based CSI compression and feedback.

[0099] The communication system 10 also includes the previously mentioned wireless device 22. The wireless device 22 may have hardware 80, which may include a radio interface 82 configured to establish and maintain a wireless connection 64 with a network node 16 serving the coverage area 18 where the wireless device 22 is currently located. The radio interface 82 may be configured as or may include, for example, one or more RF transmitters, one or more RF receivers, and / or one or more RF transceivers.

[0100] The hardware 80 of the wireless device 22 also includes a processing circuit module 84. The processing circuit module 84 may include a processor 86 and a memory 88. In particular, as an addition to or alternative to the processor (such as a central processing unit) and memory, the processing circuit module 84 may include integrated circuit modules for processing and / or control, such as one or more processors and / or processor cores suitable for executing instructions and / or FPGAs (Field-Programmable Gate Arrays) and / or ASICs (Application-Specific Integrated Circuit Modules). The processor 86 may be configured to access (e.g., write to and / or read from) the memory 88, which may include any kind of volatile and / or non-volatile memory, such as cache memory and / or buffer memory and / or RAM (Random Access Memory) and / or ROM (Read-Only Memory) and / or optical memory and / or EPROM (Erasable Programmable Read-Only Memory).

[0101] Therefore, the wireless device 22 may also include software 90, which is stored, for example, in memory 88 at the location of the wireless device 22 or in external memory (e.g., a database, storage array, network storage device, etc.) accessible by the wireless device 22. The software 90 may be executable by the processing circuitry module 84. The software 90 may include a client application 92. The client application 92 may be operable to provide services to human or non-human users via the wireless device 22 with the support of the host computer 24. In the host computer 24, the executing host application 50 may communicate with the executing client application 92 via an OTT connection 52 terminated between the wireless device 22 and the host computer 24. When providing services to a user, the client application 92 may receive request data from the host application 50 and provide user data in response to the request data. The OTT connection 52 may transmit both the request data and the user data. The client application 92 may interact with the user to generate the user data it provides.

[0102] Processing circuitry module 84 may be configured to control any of the methods and / or processes described herein, and / or to cause such methods and / or processes to be performed, for example, by wireless device 22. Processor 86 corresponds to one or more processors 86 for performing the functions of wireless device 22 described herein. Wireless device 22 includes memory 88 configured to store the data described herein, programming software code, and / or other information. In some embodiments, software 90 and / or client application 92 may include instructions that, when executed by processor 86 and / or processing circuitry module 84, cause processor 86 and / or processing circuitry module 84 to perform the processes described relative to wireless device 22. For example, processing circuitry 84 of wireless device 22 may include implementation unit 34 configured to perform one or more functions of wireless device 22 described herein, including functions related to RRC signaling and CSI reporting for AI-based CSI compression and feedback.

[0103] In some embodiments, the internal operations of network node 16, wireless device 22, and host computer 24 may be as follows: Figure 10 As shown, and independently, the surrounding network topology can be Figure 9 The network topology.

[0104] Figure 10 In the diagram, OTT connection 52 is abstractly depicted to illustrate communication between host computer 24 and wireless device 22 via network node 16, without explicitly mentioning any intermediate devices or the exact routing of messages through these devices. The network infrastructure can determine the routing, which can be configured to be hidden from wireless device 22, the service provider operating host computer 24, or both. While OTT connection 52 is active, the network infrastructure can further make decisions, dynamically altering the routing (e.g., based on network load balancing considerations or reconfiguration).

[0105] The wireless connection 64 between wireless device 22 and network node 16 is based on the teachings of the embodiments described throughout this disclosure. One or more of the various embodiments utilize an OTT connection 52 to improve the performance of the OTT services provided to wireless device 22, in which the wireless connection 64 may form a final segment. More precisely, the teachings of some embodiments in these embodiments can improve data rates, latency, and / or power consumption, and thereby provide beneficial effects such as reduced user wait times, relaxed file size limits, better responsiveness, and extended battery life.

[0106] In some embodiments, a measurement process may be provided to facilitate monitoring of data rates, latency, and other factors improved in one or more embodiments. Optional network functionality may also be available for reconfiguring the OTT connection 52 between the host computer 24 and the wireless device 22 in response to changes in measurement results. The measurement process and / or the network functionality for reconfiguring the OTT connection 52 may be implemented via software 48 of the host computer 24 or software 90 of the wireless device 22, or both. In embodiments, sensors (not shown) may be deployed in or associated with a communication device within the OTT connection 52; the sensors may participate in the measurement process by providing values ​​of the monitored quantities illustrated above or by providing values ​​of other physical quantities from which the software 48, 90 can calculate or estimate the monitored quantities. Reconfiguration of the OTT connection 52 may include message formats, retransmission settings, preferred routing, etc.; reconfiguration does not affect network node 16, and it may be unknown or undetectable to network node 16. Some such processes and functionalities may be known and implemented in the art. In some embodiments, the measurement may involve proprietary wireless signaling that facilitates the host computer 24 in measuring throughput, propagation time, latency, etc. In some embodiments, the measurement can be implemented because the software 48, 90 enables messages to be transmitted using the OTT connection 52, particularly empty or 'fake' messages, while monitoring propagation time, errors, etc.

[0107] Therefore, in some embodiments, the host computer 24 includes: a processing circuitry module 42 configured to provide user data; and a communication interface 40 configured to forward the user data to a cellular network for transmission to the wireless device 22. In some embodiments, the cellular network further includes a network node 16 having a radio interface 62. In some embodiments, the network node 16 is configured and / or its processing circuitry module 68 is configured to perform the functions and / or methods described herein for: preparing / initiating / maintaining / supporting / terminating transmissions to the wireless device 22 and / or preparing / terminating / maintaining / supporting / terminating reception of transmissions from the wireless device 22.

[0108] In some embodiments, the host computer 24 includes a processing circuitry module 42 and a communication interface 40 configured to receive user data from transmissions from the wireless device 22 to the network node 16. In some embodiments, the wireless device 22 is configured to perform the functions and / or methods described herein for the following operations and / or includes a radio interface 82 and / or a processing circuitry module 84 configured to perform the functions and / or methods described herein for the following operations: preparing / initiating / maintaining / supporting / terminating transmissions to the network node 16 and / or preparing / terminating / maintaining / supporting / terminating reception of transmissions from the network node 16.

[0109] Although Figure 9 and Figure 10 Various "units" (such as configuration unit 32 and implementation unit 34) are shown as residing within their respective processors, but it is contemplated that these units can be implemented such that a portion of the unit is stored in a corresponding memory within the processing circuitry module. In other words, the units can be implemented in hardware or through a combination of hardware and software within the processing circuitry module.

[0110] Figure 11 This illustrates, according to one embodiment, in a communication system (e.g., such as...) Figure 9 and Figure 10 The flowchart illustrates an example method implemented in a communication system. The communication system may include a host computer 24, a network node 16, and a wireless device 22, which may be referenced... Figure 10 The aforementioned host computers, network nodes, and wireless devices. In the first step of the method, host computer 24 provides user data (block S100). In an optional sub-step of the first step, host computer 24 provides user data by executing a host application (e.g., such as host application 50) (block S102). In the second step, host computer 24 initiates a transmission carrying user data to wireless device 22 (block S104). In an optional third step, according to the teachings of the embodiments described throughout this disclosure, network node 16 transmits user data to wireless device 22 (block S106), the user data being carried in the transmission initiated by host computer 24. In an optional fourth step, wireless device 22 executes a client application (e.g., such as client application 92) associated with host application 50 executed by host computer 24 (block S108).

[0111] Figure 12 This illustrates, according to one embodiment, in a communication system (e.g., such as...) Figure 9 The flowchart illustrates an example method implemented in a communication system. The communication system may include a host computer 24, a network node 16, and a wireless device 22, which may be referenced... Figure 9 and Figure 10 The aforementioned host computer, network node, and wireless device. In the first step of the method, host computer 24 provides user data (block S110). In an optional sub-step (not shown), host computer 24 provides user data by executing a host application (e.g., host application 50). In the second step, host computer 24 initiates a transmission carrying user data to wireless device 22 (block S112). According to the teachings of the embodiments described throughout this disclosure, the transmission may be carried out via network node 16. In an optional third step, wireless device 22 receives the user data carried in the transmission (block S114).

[0112] Figure 13 This illustrates, according to one embodiment, in a communication system (e.g., such as...) Figure 9 The flowchart illustrates an example method implemented in a communication system. The communication system may include a host computer 24, a network node 16, and a wireless device 22, which may be referenced... Figure 9 and Figure 10 The aforementioned host computers, network nodes, and wireless devices. In an optional first step of the method, wireless device 22 receives input data provided by host computer 24 (block S116). In an optional sub-step of the first step, wireless device 22 executes client application 92, which responds to the received input data provided by host computer 24 to provide user data (block S118). Additionally or alternatively, in an optional second step, wireless device 22 provides user data (block S120). In an optional sub-step of the second step, the wireless device provides user data by executing a client application (e.g., such as client application 92) (block S122). In providing user data, the executed client application 92 may further consider user input received from the user. Regardless of the specific manner in which user data is provided, wireless device 22 initiates the transmission of user data to host computer 24 in an optional third sub-step (block S124). In accordance with the teachings of the embodiments described throughout this disclosure, in the fourth step of the method, the host computer 24 receives user data transmitted from the wireless device 22 (block S126).

[0113] Figure 14 This illustrates, according to one embodiment, in a communication system (e.g., such as...) Figure 9 The flowchart illustrates an example method implemented in a communication system. The communication system may include a host computer 24, a network node 16, and a wireless device 22, which may be referenced... Figure 9 and Figure 10 The aforementioned host computer, network node, and wireless device. In an optional first step of the method, according to the teachings of the embodiments described throughout this disclosure, network node 16 receives user data from wireless device 22 (block S128). In an optional second step, network node 16 initiates a transmission of the received user data to the host computer (block S130). In a third step, host computer 24 receives the user data carried in the transmission initiated by network node 16 (block S132).

[0114] Figure 15AThis is a flowchart of an example process in network node 16. One or more blocks described herein can be performed by one or more elements of network node 16, such as one or more of processing circuitry module 68 (including configuration unit 32), processor 70, radio interface 62, and / or communication interface 60. Network node 16 is configured to transmit an instruction to a wireless device that results in the generation of an AI-based Channel State Information (CSI) report (block S134). Network node 16 is configured to receive the AI-based CSI report (block S136). Network node 16 is configured to perform at least one action based on the received CSI report (block S138).

[0115] In at least one embodiment, the indication includes at least one of the following: a limitation on the rank that the wireless device can report; at least one configuration of at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; at least one configuration of quantization bits that the wireless device can use to quantize the feature vector; at least one configuration of the number of potential spatial coefficients at the output of the AI ​​model; and an indication for the wireless device to use at least one of transport layer common processing and transport layer specific processing to generate a CSI report.

[0116] In at least one embodiment, network node 16 is configured to indicate at least one parameter to be used in generating a CSI report, the at least one parameter being indicated on at least one of downlink control information (DCI) and media access control element (MAC-CE).

[0117] Figure 15B This is a flowchart of an example process in network node 16. One or more blocks described herein can be executed by one or more elements of network node 16, such as one or more of processing circuitry module 68 (including configuration unit 32), processor 70, radio interface 62, and / or communication interface 60. Network node 16 has one or more decoders of one or more available autoencoders. Network node 16 is configured to instruct (block S135) a wireless device to transmit a CSI report compressed by the autoencoder via RRC signaling. Network node 16 is configured to receive a first CSI report and a second CSI report at different parts of the UCI (block S137). Network node 16 is configured to use one of the decoders (block S139) to decode the first CSI report and the second CSI report.

[0118] In at least one embodiment, the indication includes at least one of the following: a limitation on the rank that the wireless device can report; at least one configuration of at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; at least one configuration of quantization bits that the wireless device can use to quantize the feature vector; at least one configuration of the number of potential spatial coefficients at the output of the AI ​​model; and an indication for the wireless device to use at least one of transport layer common processing and transport layer specific processing to generate a CSI report.

[0119] In at least one embodiment, network node 16 is configured to indicate at least one parameter to be used in generating a CSI report, the at least one parameter being indicated on at least one of downlink control information (DCI) and media access control element (MAC-CE).

[0120] Figure 16A This is a flowchart of an example process in a wireless device 22 according to some embodiments of the present disclosure. One or more blocks described herein may be performed by one or more elements of the wireless device 22, such as one or more of the processing circuitry module 84 (including implementation unit 34), processor 86, radio interface 82, and / or communication interface 60. The wireless device 22 is configured to receive an instruction from a network node to generate an artificial intelligence-based (AI-based) Channel State Information (CSI) report (block S140). The wireless device 22 is configured to generate the CSI report (block S142). The wireless device 22 is configured to transmit the CSI report to the network node (block S144).

[0121] In at least one embodiment, the indication includes at least one of the following: a limitation on the rank that the wireless device can report; at least one configuration of at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; at least one configuration of quantization bits that the wireless device can use to quantize the feature vector; at least one configuration of the number of potential spatial coefficients at the output of the AI ​​model; and an indication for the wireless device to use at least one of transport layer common processing and transport layer specific processing to generate a CSI report.

[0122] In at least one embodiment, the wireless device 22 is configured to receive at least one parameter to be used in generating a CSI report, the at least one parameter being received on at least one of downlink control information (DCI) and medium access control element (MAC-CE).

[0123] Figure 16BThis is a flowchart of an example process in a wireless device 22 according to some embodiments of the present disclosure. One or more blocks described herein may be performed by one or more elements of the wireless device 22, such as one or more of the processing circuitry module 84 (including implementation unit 34), processor 86, radio interface 82, and / or communication interface 60. The wireless device 22 has one or more encoders of one or more available autoencoders. The wireless device 22 is configured to receive, via RRC signaling, an indication from a network node to use the encoder to generate a Channel State Information (CSI) report (block S141). The wireless device 22 is configured to use the encoder to generate the CSI report (block S143). The wireless device 22 is configured to segment the encoder output into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different parts of the UCI (block S145).

[0124] In at least one embodiment, the indication includes at least one of the following: a limitation on the rank that the wireless device can report; at least one configuration of at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; at least one configuration of quantization bits that the wireless device can use to quantize the feature vector; at least one configuration of the number of potential spatial coefficients at the output of the AI ​​model; and an indication for the wireless device to use at least one of transport layer common processing and transport layer specific processing to generate a CSI report.

[0125] In at least one embodiment, the wireless device 22 is configured to receive at least one parameter to be used in generating a CSI report, the at least one parameter being received on at least one of downlink control information (DCI) and medium access control element (MAC-CE).

[0126] The overall process flow of the arrangements of this disclosure has been described, and examples of hardware and software arrangements for implementing the processes and functions of this disclosure have been provided. The following sections provide details and examples of arrangements for RRC signaling and CSI reporting for AI-based CSI compression and feedback. The functions of one or more wireless devices 22 described below may be executed by one or more of processing circuitry module 84, processor 86, implementation unit 34, etc. The functions of one or more network nodes 16 described below may be executed by one or more of processing circuitry module 68, processor 70, configuration unit 32, etc.

[0127] In at least one embodiment, the wireless device 22 estimates the DL channel based on a configured DL reference signal (e.g., CSI-RS, demodulation reference signal (DMRS)) and generates a channel estimate H (e.g., in the antenna frequency domain). The original channel H can be represented by the CSI-RS port (TX side), by the receiving antenna (RX side), by the frequency subband, and measured at one or more time points. Therefore, in the most general case, the channel H is a four-dimensional matrix or tensor.

[0128] The original channel estimate H (possibly along with interfering channels) is used to estimate the appropriate rank for downlink transmission and is further processed to determine the rank based on the estimated rank. r Extract the feature vector corresponding to each layer. Based on r The feature vector of each transport layer is represented as ,in For measurements taken via CSI-RS in a single time instance, It is a tensor, whose dimension is equal to the number of CSI-RS ports × the number of layers × the number of frequency subbands. Extracted It is compressed and quantized into bits at the encoder, making Indicates the use of quantization of the first l r Each transport layer's bits. Then, the concatenated bits across all transport layers (represented as...) The Rank Indicator (RI) and Channel Quality Indicator (CQI) are reported back to network node 16 as part of the uplink CSI report. b AE The CSI report and the conventional parameters (i.e., RI and CQI) calculated from the estimated channel H are fed into the decoder deployed at network node 16 to reconstruct the feature vector for each layer, which is represented as follows: Network node 16 can further process the feature vectors to obtain the pre-encoder for each layer (represented as...). () for use in the transmission of PDSCH.

[0129] Depending on the number of CSI-RS ports and sub-bands, the original The dimensionality can be very high, which complicates the AE model and training. Therefore, H can be further preprocessed to achieve a reduced dimensionality for feature extraction based on feature vectors compared to the original feature vectors at each layer. As described in this paper, it is used to... L SD basics and M A delayed tap (through) M The channel preprocessing (based on FD) extracts the features of each layer of feature vectors in the beam delay domain, generating a linear combination coefficient tensor with a dimension of 2. L ×Number of layers× M This is represented as W2. In a more specific implementation,L and M The value can be selected from Rel-16 type-II preprocessing, for example as specified in 3GPP TS 38.214 ( M Depending on, for example, 3GPP TS 38.214 p v The value of , where v (This is a layer index). Through the above preprocessing, the encoder at wireless device 22 compresses and quantizes W2, where the reduction in dimensionality of W2 leads to a reduction in the AE model size and a decrease in training complexity. As described in this paper, the NZC feedback of W2 is the main overhead contributing to Type-II, which can be reduced by utilizing the feedback through the AE. The above preprocessing of the feature vectors may require wireless device 22 to... L SD and M Each FD basis is explicitly fed back to network node 16 (as part of the uplink CSI report in the UCI). Therefore, L SD and M The index of each FD base is encoded into bits, represented as b model It was reported to network node 16 as part of the uplink CSI report.

[0130] Based on the above, the channel processing used to generate CSI reports in UCI to generate precoders for each transport layer via automatic encoders (AEs) is as follows: Figure 17 The figure shown illustrates the architecture of a channel feature vector feedback method for generating CSI reports in the UCI for each transport layer of the precoder at network node 16. The input to each layer of the encoder is called a feature vector. However, the term "feature vector" can also be used in a broader sense to encompass the different ways in which the wireless device 22 extracts precoding information for different layers.

[0131] Based on the above, the standardized format of CSI reports (e.g., what to report and how to report) enables wireless device 22 to efficiently compress and report CSI, and then network node 16 can correctly retrieve CSI based on the reported CSI. Therefore, the detailed reporting mechanism for AI-based implicit CSI feedback (what quantities to report and how to report them) is based on the eigenvector decomposition of the channel estimated at each transport layer (RI-based).

[0132] AI-based CSI report RRC configuration Network node 16 can be configured with various parameters via RRC, which determine the size of the CSI report and help network node 16 correctly decode the CSI report received from wireless device 22. Specifically, network node 16 can explicitly configure parameters such as the model ID used to identify the AI ​​model to be used, the number of quantization bits to be used at each transport layer at wireless device 22, any limitations on the rank sum and / or precoder of the report, etc.

[0133] Table 1 shows the AI-based fields for RRC configuration used in CSI reporting in the CodebookConfig Information Element (IE). • In at least one embodiment, network node 16 can configure the use of AI-based PMI reports by setting reportQuantity in CSI-reportConfig IE to a new value “cri-RI-aiPMI-CQI”.

[0134] • In at least one embodiment, the new codebook type “typeAI” may be included under “codebookType” in the CodebookConfig IE, as shown in Table 1. Network node 16 may be RRC configured with parameters similar to those of a regular “type1” or “type2” codebook type CSI report, which may include additional parameters depending on the AI ​​model deployed by network node 16 and / or wireless device 22. Note that “typeAI” may also include regular parameters such as “n1-n2-codebookSubsetRestriction-ai” and “typeAI-RI-Restriction”, which respectively restrict the beams available for precoding and the rank that wireless device 22 can report, as specified in 3GPP TS 38.331 and 38.212. The following additional parameters specific to AI-based CSI reporting may be configured under codebook type “typeAI”: i. Quantization bit limit typeAI-QB-Restriction The field restricts the number of quantization bits that can be used to quantize the potential space coefficients. In at least one embodiment, if the maximum quantization bits that the AI ​​model at network node 16 can process are Q, then a string of length 4 is specified, for example... ,in s 0 is an LSB (corresponding to a quantized bit of 1), and s Q-1(The quantization bits corresponding to Q) are the MSB. Network node 16 can further restrict the values ​​of the quantization bits usable at wireless device 22 to a subset of the allowed number of quantization bits in the bit string. For example, if Q=4 and the allowed number of quantization bits is restricted to 2 and 4, the bit string can be set to... . In at least one embodiment, this limitation can be achieved through One bit is configured, of which N QB This indicates the number of possible quantization bits. For example, if the encoder / decoder supports 2-bit and 4-bit quantization of the latent spatial variables, then... N QB =2, and '0' can be used to configure 2-bit quantization, while '1' can be used to configure 4-bit quantization. In at least one embodiment, the number of quantization bits is rank-dependent. In this case, one way to configure the number of quantization bits is to predefine multiple configurations. For example, two configurations can be predefined, one being... And the other is {2,2,2,2}, where for r = 1,2,3,4 is the number of bits used for quantization layer r. Then, network node 16 can configure one of the predefined configurations to wireless device 22 for AI CSI reporting. In at least one embodiment, there are multiple predefined assumptions regarding the number of quantized bits. In a dependent claim, the wireless device 22 is free to determine the assumptions, and the wireless device 22 reports the assumptions used as part of the CSI report. In at least one embodiment, the network node 16 configures a subset of the assumptions, and the wireless device 22 reports the assumptions within the configured subset for use in the CSI report as part of the CSI report. Note that scalar quantization is used as an example in the above text. Other embodiments may use other types of quantization methods, such as vector quantization, where, instead of the number of quantized bits, the wireless device 22 may be configured with (optionally or additionally) a quantization codebook size. Furthermore, the wireless device 22 may also be additionally configured with more detailed aspects of the quantization method. For example, in scalar quantization, a mechanism for determining the quantization point may also be configured, such as tanh-based quantization, uniform (or non-uniform) quantization, etc.

[0135] ii. Similar to a regular Rel-16 type-II codebook, parameters It may need to be configured at wireless device 22, where L andp v The number of spatial domain bases and the number of frequency domain bases are given to preprocess the feature vectors of each transport layer. In at least one embodiment, The value is implicitly determined by the AI ​​model deployed at network node 16 and is sent to wireless device 22 via a signal using the model ID. In related embodiments, network node 16 and wireless device 22 are configured with multiple models to process a certain range. At that time, it is possible CSI-reportConfig IE contains ' paramCombination-ai 'Fields, to explicitly include The vectors are sent to wireless device 22 for preprocessing of the feature vectors for each transport layer, where: In at least one embodiment, The value is comparable to that of regular Rel-16 Type-II preprocessing, where p v The value of is constant across the report's rank, as shown in Table 2. Additionally, higher resolution parameters may be included, i.e. L and p v Higher values, for example, in Table 2 paramCombination Index 8. Furthermore, network node 16 can be configured by containing... paramCombination Indices 9 and 10 are used to configure either spatial domain preprocessing only or frequency domain preprocessing only. Finally, in another embodiment, network node 16 can be configured by including... paramCombination Index 11 is used to configure the wireless device 22 to compress and feed back the original feature vector of each layer. - Table 2 - Codebook parameter configuration (for L ,β, p v and n b,v ) In at least one embodiment, wherein The value is implicitly sent by network node 16 using a signal. paramCombination-ai The field may contain the number of active potential space coefficients for each transport layer (this number is given by the ratio of active potential space coefficients and is expressed as...). The number of quantization bits for the active potential space coefficients per layer and / or the number of quantization bits. The configuration is as shown in Table 3. Note that... The value can be explicitly specified, or it can be a function of the allowed quantization bits, which is determined by ' typeAI-QB-Restriction Configure it using fields. Table 3 – Codebook Parameter Configuration (for and ) In at least one embodiment, the 'paramCombination-ai' field may contain a subset of the parameter configurations mentioned above, wherein the inclusion of additional parameter configurations is not excluded. In at least one embodiment, the 'paramCombination-ai' field can be excluded from the RRC configuration, and the parameter The network node 16 can implicitly send the model ID to the wireless device 22 via a signal, or the wireless device 22 can configure and send the model ID to the network node 16 via a signal (explicitly or implicitly) (discussed further below).

[0136] iii. Whether to use layer-wide processing or layer-specific processing can be determined by network node 16 via the boolean field ' typeAI- Layer-Common 'To configure. Network node 16 can send a signal to wireless device 22 by activating the above parameters to compress all transport layers using the same parameters through the layer common model.

[0137] • In at least one embodiment, a subset of the above parameters is included for use with codebook type ' typeAI '. It is not excluded that additional parameters may be included to further enhance the configuration of network node 16 for AI-based CSI reporting. In another embodiment, the codebook type' typeAI 'Based on network node 16, CSI reports can be configured to be layer-common or layer-specific, and further divided into subtypes, which may be used to define codebook types.' typeAI Any other criteria for the subtype of '.

[0138] • In at least one embodiment, a subset of the parameters related to the AI ​​model described in this section may be dynamically transmitted by network node 16 to wireless device 22 via DCI and / or MAC-CE, or may be transmitted by wireless device 22 to network node 16 via UCI.

[0139] • In at least one embodiment, as shown in Table 1, network node 16 may also RRC configure multiple AI-based CSI reporting settings. Network node 16 can then instruct wireless device 22 which CSI reporting setting should be used to compute the CSI report. This can be performed in a variety of ways. For example, each of the multiple configured CSI reporting settings may be associated with one CSI- SemiPersistentOnPUSCH-TriggerState or CSI-AperiodicTriggerState Then, network node 16 instructs wireless device 22 via MAC-CE and / or DCI which CSI report setting can be used to calculate the CSI report.

[0140] • For each of the configurations described above, a value (or a combination of values ​​or an index) can be used as a default value. For example, the default value could be the first of several values ​​for the parameters configured for wireless device 22. This default value can be used, for example, in cases where wireless device 22 cannot receive DCI.

[0141] UCI Configuration of AI-Based CSI Reporting Framework This paper describes a further CSI reporting method based on AI-based implicit CSI feedback, which focuses on enhancing the traditional UCI framework. Following the traditional structure, the CSI report is segmented into Part 1 CSI and Part 2 CSI, where Part 2 CSI can be further subdivided to meet the feedback mechanism requirements of the deployed AI model. The CSI reports carried on Part 1 and Part 2 are part of the uplink control information (UCI), which can be carried on either PUCCH or PUSCH. Part 1 CSI reports common reporting quantities of both AI-based CSI and traditional CSI (e.g., Type I / II), such as CSI-RS resource indicators (CRI), rank indicators (RI), and channel quality indicators (CQI). As described in this paper, preprocessing is performed to extract... L SD basics and M The characteristics of the feature vector of each transmission layer in the beam delay domain of an FD-based beam, which are obtained through b model The bits are fed back. Bit sequence. b model Having a predefined order allows network node 16 to know how to... b model The corresponding part is mapped to a certain extracted feature. Therefore, a signal is sent to { L , M The number of bits for each SD and FD base can be carried in part 1 of the CSI report of the UCI.

[0142] • In at least one embodiment, a signal is used to send { L , MThe number of bits for each SD and FD basis is implicitly mapped to the AI ​​model. Such models can be configured by network node 16 or wireless device 22. When the model is configured by wireless device 22, the selected model needs to be reported to network node 16. The selected model (identified by the model ID) can be reported to network node 16 in CSI section 1.

[0143] Extracting transport layer information b model and b AE These bits form part 2 of the CSI report in UCI. To prioritize the processing of wideband, large-scale features of the feature vectors from each transport layer, preprocessed information is transmitted via signaling. b model Bits compared to those extracted from AE-based processing steps b AE Bits are reported with higher priority. Note that since the preprocessing step is optional (as described in this article), the reporting... b model (and sending signals) b model The number of bits is unnecessary (i.e., when the wireless device 22 compresses and reports the raw feature vector for each transmission layer).

[0144] With b AE Example embodiments related to bit width and segmentation The number of bits generated from the AI ​​model at wireless device 22 (i.e., b AE The bit width of the model can be implicitly mapped to the AI ​​model. Such a model can be configured by network node 16 or wireless device 22. When the model is configured by wireless device 22, the selected model (via model ID) is subsequently reported to network node 16 in CSI section 1. In various embodiments, methods for providing implicit CSI feedback based on AI are discussed. b AE The bit width is further measured by sending a signal to network node 16. Specifically, b AE The bit width can implicitly depend on auxiliary information for each eigenvector of each transport layer, such as the number of active latent space coefficients and / or the number of quantization bits per latent space coefficient used to process a particular transport layer.

[0145] In at least one embodiment, the auxiliary information may be bound to a model ID, which may be configured by network node 16 or signaled to network node 16 by wireless device 22. Using the auxiliary information and bit width, network node 16 can determine the parameters used to decode all transport layers.b AE The correct sequence of bits.

[0146] In at least one embodiment, the auxiliary information can be dynamically configured by the wireless device 22 based on the CSI report payload and explicitly signaled to the network node 16 in the UCI. Using the auxiliary information and the bit width, the network node 16 can determine the parameters used to decode all transport layers. b AE The correct sequence of bits. In at least one embodiment, the aforementioned auxiliary information for each transport layer is reported to network node 16 in CSI section 1, because it can be used by network node 16 to determine. b AE The bit width. In at least one embodiment, the wireless device 22 may explicitly transmit signals in part 1 of the CSI report. b AE The bit width is used to move the signaling of auxiliary information for each transport layer to part 2 of the CSI report. In at least one embodiment, the wireless device 22 may transmit auxiliary information for each transport layer in part 1 of the CSI report using signals. b AE The size of both.

[0147] Furthermore, due to the bit sequence b AE It may have a large payload, bit sequence b AE It can be divided into multiple segments, which are transmitted in CSI Part 2. When UCI resources (such as PUSCH allocations) for carrying such CSI reports are insufficient, it allows the discarding of one (or more) segments. b AE This is consistent with the traditional CSI reporting framework. Therefore, for AI-based implicit CSI feedback, b AE It can be segmented into multiple non-overlapping parts, with each segment corresponding to a transport layer.

[0148] • In at least one embodiment, the AI-based CSI report follows implicit CSI feedback. b AE Each segment can be further divided into sub-segments to carry transport layer-specific auxiliary information in CSI section 2, such as the number of active potential spatial coefficients and / or the number of quantization bits per potential spatial coefficient, as well as the output of the AI ​​model at the wireless device 22 of each transport layer.

[0149] Implementation examples related to the mapping order of CSI reports to UCI bit sequences In CSI reports, the number of CQI, RI, CQI, model ID, SD, and FD bases is determined. b model The parameters (size of the CSI report), the number of active potential space coefficients, and / or the number of quantization bits per potential space coefficient (used for processing transport layer information) provide auxiliary information so that network node 16 can estimate the payload of the received CSI report and decode the transport layer information. The following embodiments describe... b AUX The mapping strategy from the bits corresponding to these parameters to the CSI report based on AI-based implicit CSI feedback.

[0150] • In at least one embodiment, bits associated with auxiliary information common across all transport layers (such as the number of CQI, RI, CQI, model ID, SD, and FD bases) and layer-specific auxiliary information (such as the number of active latent space coefficients and / or the number of quantization bits per latent space coefficient) are included. b AUX The bits in CSI Part 1 are then transmitted. Subsequently, the bits in CSI Part 1 across multiple CSI reports are mapped onto the UCI bit sequence.

[0151] • In at least one embodiment, bits associated with auxiliary information common across all transport layers (such as CQI, RI, CQI, model ID, and the number of SD and FD bases) are included. b AUX The bits in CSI Part 1 are transmitted in CSI Part 2, while the bits related to layer-specific auxiliary information (such as the number of active latent space coefficients and / or the number of quantization bits per latent space coefficient) are transmitted in CSI Part 2. Subsequently, the bits in CSI Part 1 across multiple CSI reports are mapped onto the UCI bit sequence.

[0152] As discussed in this article, b AE and b model The bits are allocated to CSI section 2 of the CSI report, and then the multiple CSI reports are mapped to the UCI bit sequence. Next, in the following embodiments, The bit-to-CSI report mapping corresponds to different AI-based CSI feedback models.

[0153] Example implementation of a shared CSI report across layers: • In at least one embodiment, a common CSI report for its layers is employed, corresponding to bAUX , b model and b AE The parameters can be defined as: - b AUX : Corresponding to traditional quantities (such as CRI, RI, and CQI), AI model ID, number of selected SD and FD bases, and b AE Total number of bits in the middle (if reported) - b model : The bits corresponding to the selected SD and FD bases. - b AE These correspond to the bits of compressed transport layer information generated at the AI ​​model output of wireless device 22. These bits are segmented, with each segment associated with a transport layer, such that an equal number of bits are allocated to each transport layer. Therefore, assuming The output of the AI ​​model in wireless device 22 is compressed and quantized. i The bits generated by each transport layer, then ,in v This is the number of transport layers based on the reported RI.

[0154] pass b AUX , b model and b AE The above description, b AUX It was transmitted in CSI Part 1, and The bits are transmitted in CSI Part 2. Furthermore, They are segmented into two distinct groups, similar to a traditional CSI reporting framework, with further segmentation within each group, as shown in Table 4. Table 4 shows... The segmentation can be specified by 3GPP. Therefore, network node 16 can know... The network node 16 is segmented and decoded sequentially, starting with CSI Part 1. CSI Part 1 always has a fixed payload size and carries information for calculating the payload size of CSI Part 2. Note that network node 16 can access the payload reported in CSI Part 1. b AE The total number of bits and RI in the data are used to infer or determine the corresponding data for each transport layer. b AE The number of bits in the middle.

[0155] In at least one embodiment, b AUX The bits may include the number of active latent space coefficients at the output of the AI ​​model at the wireless device's 22 locations, and the number of quantization bits for each latent space coefficient, rather than... b AE The total number of bits in the middle. Therefore, network node 16 can infer (e.g., determine) the number of quantized bits for each potential spatial coefficient and the number of active potential spatial coefficients, RI, as reported in CSI Part 1. b AE The number of bits in the middle.

[0156] Table 4 - Used for Layer Co-processing b AUX , b model and b AE The mapping order of different parts / subparts of a CSI report: In at least one embodiment, if the feature vectors of each layer are reported at a certain time interval and preprocessed in the Doppler domain, the CSI report may also include a time-domain (TD) basis.

[0157] Example of a layer-specific CSI report: • In at least one embodiment, a layer-specific CSI report is employed, corresponding to b AUX , b model and b AE The parameters can be defined as: - b AUX : Corresponding to traditional quantities (such as CRI, RI, and CQI), AI model ID, number of selected SD and FD bases, and b AE The number of bits in - b model : The bits corresponding to the selected SD base and FD base. - b AE These bits correspond to the compressed transport layer information generated at the AI ​​model output of wireless device 22. These bits are segmented such that each segment is associated with a transport layer, where each segment consists of... Bits are the components, which can be functions of the transport layer. Therefore, ,in v This is the number of transport layers based on the reported RI. The total number of bits corresponding to each layer. It can be further divided into three parts, also known as subgroups (per level): • Bit, corresponding to the first i The number of active potential space coefficients at the encoder output of each transport layer. • Bit, corresponding to the first i The number of quantization bits used for each potential spatial coefficient of each transport layer. • Bit, corresponding to the first i The quantized potential space coefficients at the output of the AI ​​model of the wireless device 22 in the transmission layer.

[0158] pass b AUX , b model and b AE The above description, b AUX It was transmitted in CSI Part 1, and It is transmitted in CSI Part 2. Furthermore, They are divided into two distinct groups, each of which is further subdivided within each group, as shown in Table 5. Additionally, The segment can be divided into three sub-segments, corresponding to Table 5 shows The segmentation can be specified by 3GPP. Therefore, network node 16 can know... The segments are divided and decoded sequentially, starting with CSI Part 1. CSI Part 1 always has a fixed payload size and carries information for calculating the payload size of CSI Part 2.

[0159] Table 5 - Used for Layer-Specific Processing b AUX , b model and b AE The mapping order of different parts / subparts of a CSI report: • In at least one embodiment, where the number of potential spatial coefficients is not explicitly signaled, then It is divided into only two parts, also called subgroups (per layer): Bit, corresponding to the first i Each teleportation layer A description of the total number of bits in the group. Group 1.i.1. Bit, corresponding to the first i AI models at 22 wireless devices in the transmission layer. Output The latent space coefficients at the quantization point. Group 1.i.2.

[0160] • In at least one embodiment, when layer-specific CSI reporting is used, corresponding to b AUX , b model and b AE The parameters can be defined as: - b AUX Compared to traditional metrics (such as CRI, RI, and CQI), AI model ID, the number of selected SD and FD bases, b AE The report includes the number of bits, the number of active potential space coefficients at the encoder of each transport layer, and the corresponding number of quantization bits used per potential space coefficient. - b model : Bits corresponding to the selected SD base and FD base. - b AE These bits correspond to the compressed transport layer information generated at the output of the AI ​​model at wireless device 22. These bits are segmented such that each segment is associated with a transport layer, where each segment consists of... Bits are the components, which can be functions of the transport layer. Therefore, ,in v This is the number of transport layers based on the reported RI. It can be further divided into two parts: • Bits correspond to the step size used to sample the active latent space coefficients at the output of the AI ​​model at wireless device 22. This makes the number of people from each k i The output of the first coefficient is sent to network node 16 for use in the second... i A transport layer. • Bit, corresponding to the first i The quantized potential space coefficients at the output of the AI ​​model of the wireless device 22 in the transmission layer.

[0161] pass b AUX , b model andb AE As described above, bits are segmented, similar to the method described in the above embodiments, as shown in Table 6. However, unlike the above embodiments, the number of active latent space coefficients common to the layers and the number of quantization bits for each latent space coefficient are reported in CSI section 1, where layer-specific CSI processing is enforced by including outputs only from specific latent space coefficients acquired at regular intervals, where the interval is a function of the transport layer, and explicitly through... It is sent to network node 16 using a signal. Therefore, for each segment (or sub-segment) corresponding to the transport layer. The bits are first decoded by network node 16 for processing. (Information related to the number of quantization bits for each latent space coefficient and the number of active latent space coefficients reported in CSI Part 1).

[0162] Table 6 - For PMI-based reporting b AUX , b model and b AE The mapping order of different parts / subparts of a CSI report: • In at least one embodiment, when the AI ​​model does not preprocess the feature vectors of each layer, i.e., the AI ​​model compresses and quantizes the raw feature vectors of each layer, the CSI report generated by the wireless device 22 does not include any bits corresponding to the SD and FD bases. • In at least one embodiment, it is not excluded that Each bit and its corresponding segment contains a subset of parameters or additional parameters.

[0163] Examples of explicit CSI reporting: • In at least one embodiment, the explicit CSI estimated by the wireless device 22 (e.g., Figure 10 (As shown) can be directly preprocessed, compressed, and quantized by an AI model for use in such CSI reports, corresponding to b AUX , b model and b AE The parameters can be defined as: - b AUX : Corresponding to traditional quantities (such as CRI, RI, and CQI), AI model ID, number of selected SD and FD bases, and b AE The total number of bits in the middle. - b model : The bits corresponding to the selected SD and FD bases. - b AE : The compressed, preprocessed CSI bits generated at the output of the AI ​​model at wireless device 22 (in Figure 10 (As shown in the diagram). These bits are divided into segments such that each segment contains the bits generated by each latent spatial coefficient. Therefore, the bits in each segment can be generated by... To indicate, making , where N is the number of potential space coefficients.

[0164] pass b AUX , b model and b AE The above description, b AUX It was transmitted in CSI Part 1, and It is transmitted in CSI Part 2. Furthermore, They are segmented into two distinct groups, similar to a traditional CSI reporting framework, with further segmentation within each group, as shown in Table 7. (See Table 7 for details.) The segmentation can be specified by 3GPP. Therefore, Figure 10 The network node 16 shown in the figure can be known The segments are divided and decoded sequentially, starting with CSI Part 1. CSI Part 1 always has a fixed payload size and carries information for calculating the payload size of CSI Part 2. In at least one embodiment, b AUX Bits may include Figure 10 The number of active latent space coefficients at the output of the AI ​​model at the wireless device 22 shown, and the number of quantization bits for each latent space coefficient, are not... b AE The total number of bits in the middle. Therefore, Figure 10 The network node 16 shown can be determined by the number of quantization bits for each latent space coefficient reported in CSI Part 1 and the number of active latent space coefficients. b AE The number of bits in the middle.

[0165] Table 7 - Used for Explicit CSI Processing b AUX , b model andb AE The mapping order of different parts / subparts of a CSI report: • In at least one embodiment, for example, the amounts of RI and / or CQI are after decoding the explicit CSI, Figure 10 The network node 16 shown is calculated, and it can then be sent in the DCI as a signal. Figure 10 The wireless device 22 shown. In this case, RI and / or CQI can be excluded from the CSI report in the UCI.

[0166] Implementation examples related to the mapping of CSI reports to UCI bit sequences once Bits are mapped to CSI reports, and one or more CSI reports are mapped to a UCI bit sequence, which is then signaled to network node 16, for example, as specified in 3GPP TS 38.214. The UCI bit sequence can be transmitted on PUCCH or PUSCH. According to NR 3GPP Rel-17, two bit sequences can be created, one for CSI part 1. And for CSI Part 2 ,in and These represent the number of bits in CSI Part 1 and Part 2, respectively. CSI Part 1 to UCI bit sequence. The mapping order can be performed in the same manner as defined in, for example, Tables 6.3.2.1.2-6 of 3GPP NR Rel-17 TS 38.212 V17.2.0 (and is therefore omitted here). In the following embodiments, the mapping order of the corresponding bit sequences of multiple CSI reports to CSI Part 2 is discussed, which is enhanced for use in AI-based CSI reporting as described herein. • In at least one embodiment, for the UCI bit sequence Partial 2CSI mapping order can be accomplished by prioritizing report numbers, group numbers, or subgroup numbers within a group. When prioritizing group numbers as shown in the tables below (which further requires prioritizing subgroup numbers for each report, if configured), it ensures that all delivery layers of the report are delivered first. Any remaining resources can then be used to deliver additional reports. When prioritizing report numbers (as in Tables 8 and 9), it ensures that all lower delivery layers of the report are delivered first, which further requires prioritizing subgroup numbers of the report, if configured.

[0167] Table 8 - CSI Report to UCI Bit Sequence The mapping order, where groups have higher priority: Table 9 - CSI Report to UCI Bit Sequence The mapping order, where the report number has higher priority. Based on some examples of this disclosure: Example 1 One method in which network node 16 can use RRC signaling to send AI-based CSI reports.

[0168] Example 2 Example 1, where wireless device 22 reports an AI-based CSI report on the UCI, wherein the generation of the CSI report includes one or more of the following: Extract the feature vector of each transport layer from the estimated channel; The dimensionality of the feature vectors for each transport layer is reduced by applying preprocessing in the spatial, frequency, and / or temporal domains, where the information on the corresponding SD, FD, and / or TD bases forms part of the CSI report on the UCI. The preprocessed (or raw) feature vectors of each transport layer are compressed and quantized by the deployed AI model, with the quantized bits forming part of the CSI report on UCI. The AI-based CSI report is segmented into Part 1 CSI and Part 2 CSI, which are transmitted across different parts of the UCI. Each of Part 1 and Part 2 of the CSI can be further segmented into multiple sub-segments.

[0169] Example 3 : Either Example 1 or 2, wherein network node 16 RRC configures the parameters to determine the content and size of the AI-based CSI report on the UCI, which may include one or more of the following: Limitations on the rank of reports that wireless device 22 can receive; Configuration of SD, FD, and / or TD bases for preprocessing feature vectors of each transport layer; Wireless device 22 can be used to configure the quantization bits of the preprocessed (or raw) feature vectors of each transmission layer; The configuration of the number of potential spatial coefficients at the output of the AI ​​model at wireless device 22 in each transport layer; Configuration for generating CSI reports at wireless device 22 using transport layer common or transport layer specific processing.

[0170] Example 4Example 3, wherein one or more of the parameters are implicitly associated with (and configured with) the model deployed at network node 16 and wireless device 22. The model ID representing the model can be explicitly sent by signaling from network node 16 or wireless device 22.

[0171] Example 5 Example 3: One or more of the parameters are dynamically configured by network node 16 on DCI and / or MAC-CE.

[0172] Example 6 Example 3, wherein one or more of the parameters are explicitly configured by the wireless device 22 and reported as part of the CSI report on the UCI.

[0173] Example 7 Any of Examples 2-6, wherein, if reported, Part 1 of the CSI contains one or more of the following: Any traditional CSI reporting volume, such as CRI, RI, CQI, etc.; Any information about the AI ​​model used to generate the CSI report, such as the AI ​​model ID, as described in the host's notes, such as RAN1 110bis-e version 17; AI-based quantized bits, along with any auxiliary information based on the corresponding preprocessing, are used to decode the feature vectors of each transport layer transmitted in Part 2 CSI.

[0174] Example 8 Any of Examples 2-6, wherein, if reported, Part 2 of the CSI includes one or more of the following: Indexes of the SD, FD, and / or TD preprocessing basis of the feature vectors for each transport layer; Network node 16 decodes the transport layer-specific information required by the quantized bits of each transport layer; The quantized bits of the AI ​​model output from wireless device 22 per transmission layer.

[0175] As those skilled in the art will appreciate, the concepts described herein can be implemented as methods, data processing systems, computer program products, and / or computer storage media storing executable computer programs. Therefore, the concepts described herein can take the form of purely hardware embodiments, purely software embodiments, or embodiments combining software and hardware aspects, all of which are collectively referred to herein as “circuit” or “module.” Any process, step, action, and / or functionality described herein can be performed and / or associated with a corresponding module, which can be implemented in software and / or firmware and / or hardware. Furthermore, this disclosure can take the form of a computer program product stored on a tangible computer-readable storage medium having computer program code implemented in a computer-executable medium. Any suitable tangible computer-readable medium can be utilized, including hard disks, CD-ROMs, electronic storage devices, optical storage devices, or magnetic storage devices.

[0176] Some embodiments are described herein with reference to flowchart illustrations and / or block diagrams of methods, systems, and computer program products. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer (thereby creating a special-purpose computer), a special-purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create components for implementing the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams.

[0177] These computer program instructions may also be stored in a computer-readable storage medium or storage medium, which are capable of directing a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce manufactured articles, the manufactured articles including instruction components that implement functions / actions specified in one or more blocks of a flowchart and / or block diagram.

[0178] Computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0179] It is important to understand that the functions / actions shown in the boxes may not be performed in the order shown in the operation diagram. For example, two boxes shown consecutively may actually be executed substantially concurrently, or the boxes may sometimes be executed in reverse order, depending on the functions / actions involved. Although some diagrams include arrows on the communication path to indicate the main direction of communication, it is important to understand that communication may proceed in the opposite direction to the arrows shown.

[0180] Computer program code used to perform the operations of the concepts described herein can be written in an object-oriented programming language such as Python, Java®, or C++. However, computer program code used to perform the operations of this disclosure can also be written in a conventional procedural programming language such as the "C" programming language. The program code can be executed entirely on the user's computer, partially on the user's computer as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer. In the latter case, the remote computer can be connected to the user's computer via a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., through the Internet provided by an Internet service provider).

[0181] Many different embodiments have been disclosed herein in conjunction with the foregoing description and accompanying drawings. It will be understood that literally describing and illustrating every combination and sub-combination of these embodiments would be excessively repetitive and confusing. Accordingly, all embodiments can be combined in any manner and / or combination, and this specification, including the accompanying drawings, should be understood as a complete written description of all combinations and sub-combinations constituting the embodiments described herein and the ways and processes of making and using them, and should support the claims for any such combinations or sub-combinations.

[0182] The abbreviations that may be used in the preceding description include: Explanation of abbreviations 3GPP Third Generation Partnership Project AE automatic encoder AI (Artificial Intelligence) CQI Channel Quality Indicator CSI Channel State Information CSI-RS Channel State Information Reference Signal DCI Downlink Control Information FD frequency domain Radio base stations in gNB NR LSB (Least Significant Bit) ML Machine Learning MSB (Most Significant Bit) MU-MIMO (Multi-User Multiple Input Multiple Output) NR New Radio PMI Precoder Matrix Indicator PUCCH (Physical Uplink Control Channel) PUSCH Physical Uplink Shared Channel RI rank indicator RRC Radio Resource Control SD space domain SRS Detection Reference Signal TD time domain UCI uplink control information UE User Equipment Those skilled in the art will appreciate that the embodiments described herein are not limited to the specific examples and descriptions above. Furthermore, unless stated to the contrary above, it should be noted that all figures are not to scale. Various modifications and variations are possible based on the above teachings.

[0183] As illustrated in some of the examples presented in this article : Example A1. A network node configured to communicate with a wireless device, the network node being configured and / or including a radio interface and / or including a processing circuitry module, configured to: Transmit an instruction to the wireless device that results in the generation of an AI-based Channel State Information (CSI) report. Receive the AI-based CSI report; and Perform at least one action based on the received CSI report.

[0184] Example A2. The network node of Example A1, wherein the indication includes at least one of the following: Limitations on the rank that the wireless device can report; At least one configuration for at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; The wireless device can be used for at least one configuration of quantization bits of the feature vector; At least one configuration of the number of latent space coefficients at the output of the AI ​​model; and The wireless device uses at least one of transport layer common processing and transport layer specific processing to generate the indication of the CSI report.

[0185] Example A3. The network node of Example A1, wherein the processing circuit module is further configured to indicate at least one parameter to be used in generating the CSI report, the at least one parameter being indicated on at least one of Downlink Control Information (DCI) and Media Access Control-Control Element (MAC-CE).

[0186] Example B1. A method implemented in a network node, the method comprising: Transmit instructions to the wireless device that result in the generation of an AI-based Channel State Information (CSI) report; Receive the AI-based CSI report; and Perform at least one action based on the received CSI report.

[0187] Example B2. The method of Example B1, wherein the instructions include at least one of the following: Limitations on the rank that the wireless device can report; At least one configuration for at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; The wireless device can be used for at least one configuration of quantization bits of the feature vector; At least one configuration of the number of latent space coefficients at the output of the AI ​​model; and The wireless device uses at least one of transport layer common processing and transport layer specific processing to generate the indication of the CSI report.

[0188] Example B3. The method of Example B1 further includes indicating at least one parameter to be used in generating the CSI report, the at least one parameter being indicated on at least one of Downlink Control Information (DCI) and Media Access Control-Control Element (MAC-CE).

[0189] Example C1. A wireless device configured to communicate with a network node, the wireless device being configured and / or including a radio interface and / or a processing circuit module, configured to: Receive an instruction from the network node to generate an AI-based Channel State Information (CSI) report; Generate the CSI report; and The CSI report is transmitted to the network node.

[0190] Example C2. The wireless device of Example C1, wherein the indication includes at least one of the following: Limitations on the rank that the wireless device can report; At least one configuration for at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; The wireless device can be used for at least one configuration of quantization bits of the feature vector; At least one configuration of the number of latent space coefficients at the output of the AI ​​model; and The wireless device uses at least one of transport layer common processing and transport layer specific processing to generate the indication of the CSI report.

[0191] Example C3. The wireless device of Example C1, wherein the processing circuit module is further configured to receive at least one parameter to be used in generating the CSI report, the at least one parameter being received on at least one of downlink control information (DCI) and medium access control-control element (MAC-CE).

[0192] Example D1. A method implemented in a wireless device, the method comprising: Receive instructions from network nodes to generate an AI-based Channel State Information (CSI) report; Generate the CSI report; and The CSI report is transmitted to the network node.

[0193] Example D2. The method of Example D1, wherein the instructions include at least one of the following: Limitations on the rank that the wireless device can report; At least one configuration for at least one of the spatial domain SD, frequency domain FD, and time domain TD, the configuration being used to preprocess the feature vector; The wireless device can be used for at least one configuration of quantization bits of the feature vector; At least one configuration of the number of latent space coefficients at the output of the AI ​​model; and The wireless device uses at least one of transport layer common processing and transport layer specific processing to generate the indication of the CSI report.

[0194] Example D3. The method of Example D1 further includes receiving at least one parameter to be used in generating the CSI report, the at least one parameter being received on at least one of downlink control information (DCI) and media access control-control element (MAC-CE).

Claims

1. A method for reporting Channel State Information (CSI), performed by a wireless device having one or more encoders of one or more available autoencoders, the method comprising: Use the encoder to generate the (S143) CSI report; as well as The encoder output is segmented (S145) into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different parts of the uplink control information (UCI).

2. The method of claim 1, further comprising using an encoder to generate an indication for a CSI report via receiving Radio Resource Control (RRC) signaling (S141).

3. The method according to any one of claims 1-2, wherein, The encoder is used to generate (S142) the CSI report, which includes preprocessed estimated channels.

4. The method according to claim 3, wherein, The preprocessed estimated channels include: Extract the feature vector of each transport layer from the estimated channel; and The dimensionality of the feature vectors extracted from each transport layer is reduced by applying preprocessing in the spatial, frequency, and / or temporal domains.

5. The method according to claim 4, wherein, Using an encoder to generate the (S14) CSI report involves compressing and quantizing the preprocessed feature vectors of each transport layer using an autoencoder.

6. The method according to any one of claims 1-5, wherein, The spatial, frequency, and / or time domain information forms part of the CSI report mentioned on the UCI website.

7. The method according to any one of claims 1-6, wherein, The quantized bits form part of the CSI report mentioned in the UCI.

8. The method according to any one of claims 1-7, further comprising receiving (S141) parameters associated with the content and / or size of the encoded CSI via RRC signaling, wherein, The parameters include one or more of the following; Restrictions on the rank of the encoded CSI; Indicators for the spatial, frequency, and / or temporal domain basis of the feature vectors used for preprocessing each transport layer; The wireless device can be used to indicate the quantization bits of the preprocessed feature vector for each transport layer. An indication of the number of potential spatial coefficients for the coded CSI per transport layer; An indication of whether the CSI report is generated at the wireless device using transport layer common processing or transport layer specific processing.

9. The method according to claim 8, wherein, One or more of the parameters are explicitly associated with the encoder deployed on the wireless device.

10. The method according to claim 9, wherein, A model identifier uniquely associated with one of the encoders available to the wireless device is sent to the wireless device via a signal.

11. The method according to claim 8, wherein, One or more of the parameters are dynamically configured on the downlink control information (DCI) and / or the media access control (MAC) control element (CE).

12. The method according to claim 8, wherein, One or more of the parameters are configured by the wireless device and reported as part of the CSI report on the UCI.

13. The method according to any one of claims 1-12, wherein, The first CSI report includes one or more of the following: Regular CSI reporting volume; Information about the automatic encoder used to generate the CSI report; The AI-based auxiliary information for quantized bits; The basis for preprocessing.

14. The method according to any one of claims 1-13, wherein, The second CSI report includes one or more of the following: Indexes of the spatial, frequency, and / or temporal dimensions of the feature vectors for each transport layer, preprocessed based on the underlying indexes. The network node decodes the quantized bits of each transport layer to obtain the transport layer-specific information required. The quantized bits per transmission layer from the output of the autoencoder.

15. A method for reporting Channel State Information (CSI), performed by a network node (16) having one or more decoders of one or more available autoencoders, the method comprising: The radio device is instructed (S135) to transmit a CSI report compressed by an automatic encoder via Radio Resource Control (RRC) signaling. Receive (S137) first CSI report and second CSI report on different parts of uplink control information UCI from the wireless device (22); The first CSI report and the second CSI report are decoded using a decoder from one or more of the decoders (S139).

16. The method of claim 15, further comprising transmitting parameters associated with the content and / or size of the encoded CSI via RRC signaling, wherein the parameters include one or more of the following; Restrictions on the rank of the encoded CSI; Indicators for the spatial, frequency, and / or temporal domain basis of the feature vectors used for preprocessing each transport layer; The wireless device can be used to indicate the quantization bits of the preprocessed feature vector for each transmission layer. An indication of the number of potential spatial coefficients for the coded CSI per transport layer; An indication of whether the CSI report is generated at the wireless device using transport layer common processing or transport layer specific processing.

17. The method according to claim 16, wherein, One or more of the parameters are explicitly associated with the autoencoder deployed at the network node.

18. The method according to claim 17, wherein, A model identifier uniquely associated with one of the encoders available to the wireless device is sent to the wireless device via a signal.

19. The method of claim 16, wherein, One or more of the parameters are dynamically configured on the downlink control information (DCI) and / or the media access control (MAC) control element (CE).

20. The method of claim 16, wherein, One or more of the parameters are configured by the wireless device and received by the network node as part of the CSI report on the UCI.

21. The method according to any one of claims 15-20, wherein, The first CSI report includes one or more of the following: Regular CSI reporting volume; Information about the automatic encoder used to generate the CSI report; The AI-based auxiliary information for quantized bits; The basis for preprocessing.

22. The method according to any one of claims 15-21, wherein, The second CSI report includes one or more of the following: Indexes of the spatial, frequency, and / or temporal dimensions of the feature vectors for each transport layer, preprocessed based on the underlying indexes. The network node decodes the quantized bits of each transport layer to obtain the transport layer-specific information required. The quantized bits per transmission layer from the output of the autoencoder.

23. A wireless device (22) configured to perform a method for reporting Channel State Information (CSI), the wireless device having one or more encoders of one or more available autoencoders, the method comprising: Use the encoder to generate the (S143) CSI report; as well as The encoder output is segmented (S145) into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different parts of the uplink control information (UCI).

24. The wireless device (22) according to claim 23 is further configured to perform the method according to any one of claims 2-14.

25. A wireless device (22) configured to perform a method for reporting Channel State Information (CSI), the wireless device including a processing circuit module and a memory, the wireless device having one or more encoders of one or more available autoencoders, the method comprising: Use the encoder to generate the (S143) CSI report; as well as The encoder output is segmented (S145) into a first CSI report and a second CSI report, wherein the first CSI report and the second CSI report are transmitted on different parts of the uplink control information (UCI).

26. The wireless device (22) according to claim 25 is further configured to perform the method according to any one of claims 2-14.

27. A radio access node (16) in a communication network, configured to perform a method for reporting channel state information (CSI), the network node having one or more decoders of one or more available autoencoders, the method comprising: The radio device is instructed (S135) to transmit a CSI report compressed by an automatic encoder via Radio Resource Control (RRC) signaling. Receive (S137) first CSI report and second CSI report on different parts of uplink control information UCI from the wireless device (22); The first CSI report and the second CSI report are decoded using a decoder from one or more of the decoders (S139).

28. The radio access node (16) according to claim 27 is further configured to perform the method according to any one of claims 15-22.

29. A radio access node (16) in a communication network, configured to perform a method for reporting Channel State Information (CSI), the network node including a processing circuit module and a memory, the network node having one or more decoders of one or more available autoencoders, the method comprising: The radio device is instructed (S135) to transmit a CSI report compressed by an automatic encoder via Radio Resource Control (RRC) signaling. Receive (S137) first CSI report and second CSI report on different parts of uplink control information UCI from the wireless device (22); The first CSI report and the second CSI report are decoded using a decoder from one or more of the decoders (S139).

30. The radio access node (16) according to claim 29 is further configured to perform the method according to any one of claims 15-22.

31. A computer program comprising machine-readable instructions, which, when executed by a processor of a wireless device, cause the wireless device to perform the method according to any one of claims 1-14.

32. A computer program product comprising a non-transitory computer-readable storage medium on which the computer program of claim 31 is stored.

33. A computer program comprising machine-readable instructions, which, when executed by a processor of a radio access node, cause the radio access node to perform the method according to any one of claims 15-22.

34. A computer program product comprising a non-transitory computer-readable storage medium on which the computer program of claim 33 is stored.

Images