Loop equivariant inference machine for channel estimation

Abstract

Methods, systems, and devices for wireless communication are described. A wireless device may receive an assignment of a set of resources associated with a channel, wherein the set of resources includes a first subset of resources allocated for data transmission and a second subset of resources allocated for reference signals. The wireless device may generate a first set of a plurality of channel estimates for each layer of the channel from the reference signal according to a minimum mean square estimation (MMSE) operation. The wireless device may generate a second set of a plurality of channel estimates for each layer of the channel from a non-linear two-dimensional interpolation of the channel, and may perform a refinement operation using the estimates to generate channel estimates associated with a plurality of layers.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This patent application claims priority to U.S. patent application No. 18 / 472,083, filed by PRATIK et al. on September 21, 2023, and entitled “RECURRENT EQUIVARIANT INFERENCE MACHINES FOR CHANNEL ESTIMATION,” and U.S. patent application No. 17 / 952,203, filed by PRATIK et al. on September 23, 2022, and entitled “RECURRENT EQUIVARIANT INFERENCE MACHINES FOR CHANNEL ESTIMATION,” each of which is assigned to the assignee of this application and is expressly incorporated herein by reference in its entirety. Background Art

[0003] The following relates to wireless communications, and more particularly to using machine learning models to estimate channels.

[0004] Wireless communication systems are widely deployed to provide various types of communication content, such as voice, video, packet data, messaging, broadcast, etc. These systems may be able to support communication with multiple users by sharing available system resources (e.g., time, frequency, and power). Examples of such multiple access systems include fourth generation (4G) systems (such as long term evolution (LTE) systems, advanced LTE (LTE-A) systems, or LTE-A Pro systems) and fifth generation (5G) systems (which may be referred to as new radio (NR) systems). These systems may employ techniques such as code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), or discrete Fourier transform spread orthogonal frequency division multiplexing (DFT-S-OFDM). A wireless multiple access communication system may include one or more base stations, each of which supports wireless communications for communication devices, which may be referred to as user equipment (UE). Summary of the invention

[0005] The technology relates to improved methods, systems, devices and apparatuses supporting a cyclic equivariant inference engine for channel estimation. For example, the technology allows channel estimates to be calculated using a cyclic equivariant inference engine. In some cases, a wireless communication system may place pilot symbols (e.g., demodulation reference signal (DMRS) symbols) in a transmission time slot according to a known pattern, thereby allowing a wireless device to estimate unknown resources of a channel based on known resources (e.g., DMRS symbols). For example, a wireless device may receive an assignment to a set of resources associated with a channel, wherein the set of resources includes a first subset of resources allocated for data transmission and a second subset of resources allocated for a reference signal (e.g., DMRS). The wireless device may generate multiple channel estimates (e.g., single-input and single-output (SISO) channel estimates) for each layer of the channel, and perform a refinement operation using the estimates to generate channel estimates associated with multiple layers (e.g., multiple-input and multiple-output (MIMO) channel estimates). In some cases, the refinement operation may include multiple iterations. For example, each iteration may include generating a corresponding gradient associated with each channel estimate at each layer based on the channel estimate at each layer and the observed resources of the channel (e.g., known DMRS resources); generating a current set of values ​​of the latent variable (e.g., inference variables based on observed variables) based on the previous set of values ​​of the latent variable, the corresponding gradient, and the channel estimate at each layer; and modifying (e.g., refining, updating, improving) the channel estimate based on the current set of values ​​of the latent variable, the channel estimate at each layer, and the corresponding gradient. In some cases, the refinement operation may be performed by a refinement network including a likelihood module, an encoder module, and a decoder module.

[0006] A method for wireless communication at a wireless communication device is described. The method may include: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to a minimum mean square estimation (MMSE) operation; generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates; and a refinement operation including one or more iterations on the second set of multiple channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. Executing each of the one or more iterations may include: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​of the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0007] An apparatus for wireless communication at a wireless communication device is described. The apparatus may include one or more memories and one or more processors coupled to the one or more memories. The one or more processors may be configured to cause the wireless communication device to: receive an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generate a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation; generate a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates; and perform a refinement operation including one or more iterations on the second set of multiple channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. To perform each of the one or more iterations, the one or more processors may be configured to cause the wireless device to: generate corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; generate a second set of values ​​of the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modify the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0008] Another apparatus for wireless communication is described. The apparatus may include components for: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation; generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates; and a refinement operation including one or more iterations on the second set of multiple channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. Executing each of the one or more iterations may include components for: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​of the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0009] A non-transitory computer-readable medium storing code for wireless communication at a wireless communication device is described. The code may include instructions that can be executed by one or more processors to cause the wireless communication device to perform the following operations: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation; generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates; and performing a refinement operation including one or more iterations on the second set of multiple channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. The instructions to perform each of the one or more iterations may be executable by the one or more processors to cause the wireless device to: generate corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; generate a second set of values ​​for the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modify the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​for the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0010] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters. Some examples of the methods, apparatus, and non-transitory computer-readable media described herein may include operations, features, components, or instructions for: performing a second refinement operation on the second set of multiple channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective attention calculations in a set of multiple attention calculations.

[0011] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the set of multiple attention computations includes intra-physical resource block (PRB) group computations, inter-PRB group computations, cross-multiple-input multiple-output (MIMO) computations, multi-layer perceptron (MLP) computations, or any combination thereof.

[0012] Some examples of the methods, apparatus, and non-transitory computer-readable media described herein may include operations, features, components, or instructions for performing the MMSE operation based on a resource configuration pattern of the second subset of resources allocated for the reference signal, the reference signal comprising a demodulation reference signal (DMRS).

[0013] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the corresponding gradient may include operations, features, components, or instructions for: generating a corresponding set of values ​​of residual variables based on a difference between the measured observation of the second resource subset and the second set of multiple channel estimates for the second resource subset; and combining the corresponding set of values ​​of the residual variables, the second resource subset, and the number of mask bits.

[0014] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for combining the second set of multiple channel estimates for the second subset of resources, the corresponding gradients, and corresponding values ​​in the first set of values ​​of the latent variable based on generating the corresponding gradients.

[0015] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for modeling dependencies between resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0016] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for modeling correlations between resources of each group in a set of multiple resource groups of the resource set and other groups in the set of multiple resource groups, wherein each group in the set of multiple resource groups includes a set of multiple resource blocks.

[0017] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for modeling dependencies between each layer in the set of multiple layers of the set of resources.

[0018] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, modifying the second set of multiple channel estimates may include operations, features, components, or instructions for combining the second set of values ​​of the latent variable, the second set of multiple channel estimates, and the corresponding gradients based on the set of machine learning parameters.

[0019] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the nonlinear two-dimensional interpolation of the channel is based on a machine learning model.

[0020] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the first set of multiple channel estimates and the second set of multiple channel estimates are associated with a set of multiple single-input single-output (SISO) antenna pairs.

[0021] In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, each of the one or more iterations is performed by a refinement network including a likelihood module, an encoder module, and a decoder module, the refinement network including a machine learning model. In some examples of the methods, apparatuses, and non-transitory computer-readable media described herein, each refinement network is performed according to the same set of machine learning parameters.

[0022] A method for wireless communication is described. The method may include: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources; and performing a refinement operation on the set of multiple channel estimates including one or more iterations, wherein each iteration of the one or more iterations may include operations, features, components, or instructions for: generating corresponding gradients associated with the set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generating a second set of values ​​of the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0023] An apparatus for wireless communication at a wireless communication device is described. The apparatus may include one or more memories and one or more processors coupled to the one or more memories. The one or more processors may be configured to cause the wireless communication device to: receive an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generate a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources; and perform a refinement operation including one or more iterations on the set of multiple channel estimates, wherein the instructions of each iteration of the one or more iterations are executable by the processor to cause the apparatus to: generate corresponding gradients associated with the set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generate a second set of values ​​of the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and modify the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0024] Another apparatus for wireless communication is described. The apparatus may include: a component for receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; a component for generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources; and a component for performing a refinement operation on the set of multiple channel estimates comprising one or more iterations, wherein the component for each iteration of the one or more iterations may include: a component for generating corresponding gradients associated with the set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; a component for generating a second set of values ​​of the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and a component for modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0025] A non-transitory computer-readable medium storing code for wireless communication is described. The code may include instructions executable by one or more processors to perform the following operations: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources; and performing a refinement operation on the set of multiple channel estimates comprising one or more iterations, wherein the instructions for each iteration of the one or more iterations are executable to: generate corresponding gradients associated with the set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generate a second set of values ​​of the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and modify the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0026] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the corresponding gradient may include operations, features, components, or instructions for: generating a corresponding set of values ​​of residual variables based on a difference between the measured observation of the second resource subset and the set of multiple channel estimates for the second resource subset; and combining the corresponding set of values ​​of the residual variables, the measured observation of the second resource subset, and the number of mask bits.

[0027] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for combining the set of multiple channel estimates for the second subset of resources, the corresponding gradients, and corresponding values ​​in the first set of values ​​of the latent variable based on generating the corresponding gradients.

[0028] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for modeling dependencies between resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0029] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for modeling correlations between resources of each group in a set of multiple resource groups of the resource set and other groups in the set of multiple resource groups, wherein each group in the set of multiple resource groups includes a set of multiple resource blocks.

[0030] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, generating the second set of values ​​for the latent variable may include operations, features, components, or instructions for modeling dependencies between each layer in the set of multiple layers of the set of resources.

[0031] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, modifying the set of multiple channel estimates may include operations, features, components, or instructions for combining the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0032] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, an initial value of the set of multiple channel estimates may be associated with a SISO antenna pair.

[0033] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the second subset of resources may be configured according to a resource configuration pattern in a set of resource configuration patterns.

[0034] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the set of resource configuration patterns may be a set of DMRS patterns.

[0035] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, each iteration may be performed by a refinement network including a likelihood module, an encoder module, and a decoder module, and each refinement network also includes corresponding parameters associated with the machine learning operation.

[0036] In some examples of the methods, apparatus, and non-transitory computer-readable media described herein, the resource set includes one or more resource groups, and each respective layer in the set of multiple layers may be associated with a respective antenna pair in the set of multiple SISO antenna pairs. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 An example of a wireless communication system supporting a cyclic equivariant inference engine for channel estimation in accordance with one or more aspects of the present disclosure is shown.

[0038] Figure 2An example of a network architecture supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown.

[0039] Figures 3 to 6 An example of a network supporting a cyclic equivariant inference engine for channel estimation in accordance with one or more aspects of the present disclosure is shown.

[0040] Figure 7 and Figure 8 A block diagram of a device supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown.

[0041] Fig. 9 A block diagram of a communication manager supporting a cyclic equivariant inference engine for channel estimation is shown in accordance with one or more aspects of the present disclosure.

[0042] Fig.10 A diagram of a system including a user equipment (UE) supporting a cyclic equivariant inference engine for channel estimation is shown in accordance with one or more aspects of the present disclosure.

[0043] Fig.11 A diagram of a system including network entities supporting a cyclic equivariant inference engine for channel estimation is shown in accordance with one or more aspects of the present disclosure.

[0044] Figures 12 to 15 A flow chart illustrating a method of supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. DETAILED DESCRIPTION

[0045] Some wireless communication systems may support channel estimation. For example, a wireless device may estimate the resources of a channel to maintain high data throughput. In order to achieve channel estimation, some communication systems may use a tracking reference signal (TRS) to calculate channel characteristics (e.g., Doppler, delay spread, signal-to-noise ratio (SNR), etc.) and estimate channel resources based on the channel characteristics. However, TRS may include relatively large overhead (e.g., memory, calculation), and not all wireless communication systems may send TRS continuously (e.g., periodically or relatively regularly over time), which may affect channel estimation. Additionally, some wireless communication systems may support multiple-input and multiple-output (MIMO) communications, in which multiple resource layers may interfere with each other, thereby further reducing the accuracy of channel estimation. A resource layer may refer to a spatial layer of a wireless channel. For MIMO communications, multiple antennas or antenna ports of a transmitting device may each be associated with a corresponding layer, which may be sent on the same time-frequency resource. In order to further improve TRS consistency and take into account cross-MIMO interference, the channel estimation process may be updated.

[0046] Some wireless communication devices support machine learning-based channel estimation operations. The techniques described herein allow the application of techniques for minimum mean square estimation (MMSE) operations for channel estimation to machine learning-based channel estimation techniques to improve channel estimation techniques. For example, performing a combination of machine learning-based and MMSE-based channel estimation can provide enhanced channel estimation through machine learning capabilities while maintaining existing channel estimation and demapping hardware for MMSE operations at wireless devices, which can reduce memory consumption and computational costs, among other possibilities. MMSE operations may include the wireless device estimating the channel by minimizing the mean square error of variables associated with the channel. The techniques described herein may allow the use of a cyclic equivariant inference engine to calculate channel estimates, which may be a machine learning model of a type that provides relatively reliable and accurate estimates based on a given set of inputs. In some cases, a wireless communication system may place pilot symbols (e.g., demodulation reference signal (DMRS) symbols, other types of reference symbols, or any other pilot symbols) in a transmit slot according to a known pattern, thereby allowing a wireless device to estimate unknown resources of a channel based on known resources (e.g., DMRS symbols). For example, the wireless device may receive an assignment of a first set of resources allocated for data transmission within a channel and a second set of resources allocated for a reference signal (e.g., DMRS) within the channel. The wireless device may perform an MMSE operation to generate a first set of multiple channel estimates for each layer of the channel (e.g., single-input and single-output (SISO) channel estimates). The MMSE operation may be an estimation technique that estimates the channel using linear equalization. MMSE may be supported by hardware within the wireless device, which may achieve reduced complexity and processing compared to performing an initial estimate based on machine learning. The wireless device may then modify the MMSE estimate using nonlinear two-dimensional interpolation and generate a second set of multiple channel estimates for each layer of the channel. For example, the wireless device may construct a first set of multiple channel estimates using interpolation in the time domain and the frequency domain according to a machine learning model. In some cases, the nonlinear two-dimensional interpolation may utilize a machine learning model to improve the accuracy of the channel estimate generated by the MMSE operation and apply the channel estimate to the time domain and the frequency domain.

[0047] The wireless device may perform a refinement operation using a second set of multiple channel estimates to generate channel estimates associated with multiple layers (e.g., MIMO channel estimates). That is, in one example, the wireless device may employ a channel estimate generated via an MMSE operation and subsequent machine learning interpolation, and the wireless device may refine the channel estimate and combine the estimates of each layer of the channel into a single MIMO channel estimate. In some cases, the refinement operation may include multiple iterations of refinement. Each refinement iteration may generate a corresponding gradient associated with each layer of the channel estimate based on each layer of the channel estimate and the observed resources of the channel (e.g., known DMRS resources). Each refinement iteration may further generate a current set of values ​​of a latent variable (e.g., an inference variable based on the observed variable). The current set of values ​​of the latent variable may be generated based on a previous set of values ​​of the latent variable, the corresponding gradient, and each layer of the channel estimate. Each refinement iteration may further modify (e.g., refine, update, improve) the channel estimate based on the current set of values ​​of the latent variable, the channel estimate of each layer, and the corresponding gradient. That is, in one example, each refinement iteration may further improve the channel estimate and generate a new or improved value of the inferred latent variable. In some cases, the refinement operation may be performed by a refinement network, which may be a machine learning model including a likelihood module, an encoder module, and a decoder module.

[0048] The techniques described herein may allow a wireless device to reliably and accurately estimate a channel with relatively low memory consumption and processing. For example, by first utilizing MMSE operations to generate channel estimates, the wireless device may reduce memory consumption and processing compared to machine learning based estimation or other techniques. Additionally or alternatively, by using MMSE for channel estimation, the wireless device may support reduced hardware cost and complexity compared to other machine learning based channel estimation techniques, at least because the MMSE operations may be supported by current hardware components of the wireless device.

[0049] Various aspects of the disclosure are first described in the context of a wireless communication system. Then, various aspects of the disclosure are described in the context of a network. Various aspects of the disclosure are further illustrated and described by and with reference to apparatus diagrams, system diagrams, and flow charts related to a cyclic equivariant inference engine for channel estimation.

[0050] Figure 1 An example of a wireless communication system 100 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The wireless communication system 100 may include one or more network entities 105, one or more UEs 115, and a core network 130. In some examples, the wireless communication system 100 may be an LTE network, an LTE-A network, an LTE-APro network, an NR network, or a network operating according to other systems and radio technologies (including future systems and radio technologies not explicitly mentioned herein).

[0051] The network entities 105 may be dispersed throughout a geographic area to form the wireless communication system 100, and may include devices in different forms or with different capabilities. In various examples, the network entities 105 may be referred to as network elements, mobility elements, radio access network (RAN) nodes, or network equipment, among other nomenclature. In some examples, the network entities 105 and the UE 115 may communicate wirelessly via one or more communication links 125 (e.g., radio frequency (RF) access links). For example, the network entity 105 may support a coverage area 110 (e.g., a geographic coverage area) within which the UE 115 and the network entity 105 may establish one or more communication links 125. The coverage area 110 may be an example of a geographic area within which the network entity 105 and the UE 115 may support signal communications according to one or more radio access technologies (RATs).

[0052] The UEs 115 may be dispersed throughout the coverage area 110 of the wireless communication system 100, and each UE 115 may be stationary or mobile, or stationary and mobile at different times. The UEs 115 may be devices in different forms or with different capabilities. Figure 1 Some example UEs 115 are illustrated in FIG. 1. The UEs 115 described herein may be capable of supporting communication with various types of devices, such as Figure 1 105) for communication. The UE 115 may include a communication manager 101 configured to send communications to the network entity 105 and receive communications from the network entity. In some examples, the communication manager 101 may be configured to receive an assignment to a resource set, the resource set including resources allocated for data signals and resources allocated for reference signals. Additionally or alternatively, the communication manager 101 may be configured to generate a channel estimate based on an MMSE operation, a nonlinear two-dimensional interpolation of a channel, a refinement operation, or any combination thereof.

[0053] As described herein, a node, which may be referred to as a node, a network node, a network entity, or a wireless node, may be a base station (e.g., any base station described herein), a UE (e.g., any UE described herein), a network controller, an apparatus, a device, a computing system, one or more components, and / or another suitable processing entity configured to perform any of the techniques described herein. For example, the network node may be a UE. As another example, the network node may be a base station. As another example, a first network node may be configured to communicate with a second network node or a third network node. In one aspect of this example, the first network node may be a UE, the second network node may be a base station, and the third network node may be a UE. In another aspect of this example, the first network node may be a UE, the second network node may be a base station, and the third network node may be a base station. In yet other aspects of this example, the first network node, the second network node, and the third network node may be different relative to these examples. Similarly, references to UE, base station, apparatus, device, computing system, etc. may include disclosure of UE, base station, apparatus, device, computing system, etc. as a network node. For example, a disclosure that a UE is configured to receive information from a base station also discloses that the first network node is configured to receive information from a second network node. Consistent with the present disclosure, once a specific example is expanded according to the present disclosure (e.g., a UE is configured to receive information from a base station and it is also disclosed that a first network node is configured to receive information from a second network node), a broader example of a narrower example can be interpreted inversely, but in a broad, open-ended manner. In the above example where a UE is configured to receive information from a base station and it is also disclosed that a first network node is configured to receive information from a second network node, the first network node may refer to a first UE configured to receive information, a first base station, a first device, a first equipment, a first computing system, a first one or more components, a first processing entity, etc.; and the second network node may refer to a second UE, a second base station, a second device, a second equipment, a second computing system, a second one or more components, a second processing entity, etc.

[0054] As described herein, different terms may be used in various aspects to describe the communication of information (e.g., any information, signal, etc.). Disclosure of one communication term includes disclosure of other communication terms. For example, a first network node may be described as being configured to send information to a second network node. In this example and consistent with the present disclosure, disclosure that the first network node is configured to send information to the second network node includes disclosure that the first network node is configured to provide, transmit, output, communicate, or send information to the second network node. Similarly, in this example and consistent with the present disclosure, disclosure that the first network node is configured to send information to the second network node includes disclosure that the second network node is configured to receive, obtain, or decode information provided, transmitted, output, communicated, or sent by the first network node.

[0055] In some examples, the network entities 105 may communicate with the core network 130, or with each other, or both. For example, the network entities 105 may communicate with the core network 130 via one or more backhaul communication links 120 (e.g., according to S1, N2, N3, or other interface protocols). In some examples, the network entities 105 may communicate with each other directly (e.g., directly between the network entities 105) or indirectly (e.g., via the core network 130) via the backhaul communication links 120 (e.g., according to X2, Xn, or other interface protocols). In some examples, the network entities 105 may communicate with each other via midhaul communication links 162 (e.g., according to a midhaul interface protocol) or fronthaul communication links 168 (e.g., according to a fronthaul interface protocol) or any combination thereof. The backhaul communication links 120, midhaul communication links 162, or fronthaul communication links 168 may be or include one or more wired links (e.g., electrical links, optical fiber links), one or more wireless links (e.g., radio links, wireless optical links), etc. or various combinations thereof. UE 115 may communicate with core network 130 via communication link 155 .

[0056] One or more of the network entities 105 described herein may include or may be referred to as a base station 140 (e.g., a transceiver base station, a radio base station, an NR base station, an access point, a radio transceiver, a Node B, an evolved Node B (eNB), a next generation Node B, or a Gigabit Node B (any of which may be referred to as a gNB), a 5G NB, a next generation eNB (ng-eNB), a home Node B, a home evolved Node B, or other suitable terms). In some examples, the network entity 105 (e.g., a base station 140) may be implemented in an aggregated (e.g., monolithic, stand-alone) base station architecture, which may be configured to utilize a protocol stack physically or logically integrated within a single network entity 105 (e.g., a single RAN node, such as a base station 140). The network entity 105 may include a communication manager 102 configured to send communications to and receive communications from a UE 115. In some examples, the communication manager 102 may be configured to send an assignment to a set of resources, the set of resources comprising resources allocated for data signals and resources allocated for reference signals. Additionally or alternatively, the communication manager 102 may be configured to generate a channel estimate based on an MMSE operation, a nonlinear two-dimensional interpolation of a channel, a refinement operation, or any combination thereof.

[0057] In addition to or instead of being performed between the UE 115 and the network entity 105, the techniques described herein may also be implemented via additional or alternative wireless devices (including IAB nodes 104, distributed units (DUs) 165, centralized units (CUs) 160, radio units (RUs) 170, etc.). For example, in some implementations, the various aspects described herein may be implemented in the context of a decomposed radio access network (RAN) architecture (e.g., an open RAN architecture). In a decomposed architecture, the RAN may be split into three functional areas corresponding to the CU 160, DU 165, and RU 170. The functional split between the CU 160, DU 165, and RU 175 is flexible, and therefore many permutations of different functionality are generated depending on which functions (e.g., MAC functions, baseband functions, radio frequency functions, and any combination thereof) are performed at the CU 160, DU 165, and RU 175. For example, a functional split of the protocol stack may be employed between the DU 165 and the RU 170 such that the DU 165 may support one or more layers of the protocol stack and the RU 170 may support one or more different layers of the protocol stack.

[0058] Some wireless communication systems (e.g., wireless communication system 100), infrastructure for NR access, and spectrum resources may additionally support wireless backhaul link capabilities to supplement wired backhaul connections, thereby providing an IAB network architecture. One or more network entities 105 may include CU 160, DU 165, and RU 170, and may be referred to as a donor network entity 105 or an IAB donor. One or more DUs 165 (e.g., and / or RU 170) associated with the donor network entity 105 may be controlled in part by the CU 160 associated with the donor network entity 105. One or more donor network entities 105 (e.g., IAB donors) may communicate with one or more additional network entities 105 (e.g., IAB nodes 104) via supported access and backhaul links. The IAB node 104 may support mobile terminal (MT) functionality controlled and / or scheduled by the DU 165 of the coupled IAB donor. Additionally, the IAB node 104 may include a DU 165 that supports communication links with additional entities (e.g., IAB node 104, UE 115, etc.) within a relay chain or configuration (e.g., downstream) of the access network. In such cases, one or more components of the decomposed RAN architecture (e.g., one or more IAB nodes 104 or components of the IAB node 104) may be configured to operate in accordance with the techniques described herein.

[0059] In some examples, wireless communication system 100 may include core network 130 (e.g., Next Generation Core Network (NGC)), one or more IAB donors, IAB nodes 104, and UE 115, where IAB nodes 104 may be controlled in part by each other and / or the IAB donor. IAB donor nodes and IAB nodes 104 may be examples of aspects of network entity 105. The IAB donor and one or more IAB nodes 104 may be configured as a relay chain (e.g., or communicate according to a relay chain).

[0060] For example, an access network (AN) or RAN may refer to an access node (e.g., an IAB donor), communications between an IAB node 104, and one or more UEs 115. An IAB donor may facilitate a connection between a core network 130 and an AN (e.g., via a wired or wireless connection to the core network 130). That is, an IAB donor may refer to a RAN node having a wired or wireless connection to the core network 130. An IAB donor may include a CU 160 and at least one DU 165 (e.g., and RU 170), wherein the CU 160 may communicate with the core network 130 via an NG interface (e.g., some backhaul links). The CU 160 may be responsible for layer 3 (L3) (e.g., radio resource control (RRC), service data adaptation protocol (SDAP), PDCP, etc.) functionality and signaling. At least one DU 165 and / or RU 170 may be responsible for lower layers, such as layer 1 (L1) and layer 2 (L2) (e.g., RLC, MAC, physical (PHY), etc.) functionality and signaling, and may each be controlled at least in part by CU 160. DU 165 may support one or more different cells. The IAB donor and IAB node 104 may communicate over the F1 interface according to a certain protocol (e.g., F1 AP protocol) that defines signaling messages. Additionally, CU 160 may communicate with the core network over an NG interface (which may be an example of a portion of a backhaul link) and may communicate with other CUs 160 (e.g., CUs 160 associated with alternative IAB donors) over an Xn-C interface (which may be an example of a portion of a backhaul link).

[0061] An IAB node 104 may refer to a RAN node that provides IAB functionality (e.g., access for UE 115, wireless self-backhaul capabilities, etc.). An IAB node 104 may include a DU 165 and a MT. The DU 165 may act as a distributed scheduling node toward a child node associated with the IAB node 104, and the MT may act as a scheduled node toward a parent node associated with the IAB node 104. That is, an IAB donor may be referred to as a parent node that communicates with one or more child nodes (e.g., an IAB donor may relay a UE's transmission through one or more other IAB nodes 104). Additionally, depending on the relay chain or configuration of the AN, an IAB node 104 may also be referred to as a parent node or a child node of other IAB nodes 104. Thus, the MT entity (eg, MT) of IAB node 104 may provide a Uu interface for the child node to receive signaling from parent IAB node 104, and the DU interface (eg, DU 165) may provide a Uu interface for the parent node to signal to child IAB node 104 or UE 115.

[0062] For example, the IAB node 104 may be referred to as a parent node associated with the IAB node and a child node associated with the IAB donor. The IAB donor may include a CU 160 having a wired (e.g., optical fiber) or wireless connection to the core network, and may act as a parent node of the IAB node 104. For example, the DU 165 of the IAB donor may relay transmissions to the UE 115 through the IAB node 104, and may directly signal the transmissions to the UE 115. The CU 160 of the IAB donor may signal the IAB node 104 that a communication link is established via an F1 interface, and the IAB node 104 may schedule transmissions (e.g., transmissions relayed from the IAB donor to the UE 115) via the DU 165. That is, data may be relayed to and from the IAB node 104 via signaling via the NR Uu interface of the MT to the IAB node 104. Communications with the IAB node 104 may be scheduled by the DU 165 of the IAB donor, and communications with the IAB node 104 may be scheduled by the DU 165 of the IAB node 104 .

[0063] Where the techniques described herein are applied in the context of a decomposed RAN architecture, one or more components of the decomposed RAN architecture (e.g., one or more IAB nodes 104 or components of the IAB node 104) may be configured to support techniques for large round trip times in random access channel procedures, as described herein. For example, some operations described as being performed by the UE 115 or the network entity 105 may additionally or alternatively be performed by components of the decomposed RAN architecture (e.g., IAB nodes, DUs, CUs, etc.).

[0064] In some examples, the network entity 105 may be implemented in a disaggregated architecture (e.g., a disaggregated base station architecture, a disaggregated RAN architecture) that may be configured to utilize a protocol stack that is physically or logically distributed between two or more network entities 105, such as an IAB network, an open RAN (O-RAN) (e.g., a network configuration sponsored by the O-RAN Alliance), or a virtualized RAN (vRAN) (e.g., a cloud RAN (C-RAN)). For example, the network entity 105 may include one or more of the following: a CU 160, a DU 165, a RU 170, a RAN intelligent controller (RIC) 175 (e.g., a near real-time RIC (near RT RIC), a non-real-time RIC (non-RT RIC)), a service management and orchestration (SMO) system 180, or any combination thereof. The RU 170 may also be referred to as a radio head, a smart radio head, a remote radio head (RRH), a remote radio unit (RRU), or a transmit receive point (TRP). One or more components of the network entity 105 in the decomposed RAN architecture may be co-located, or one or more components of the network entity 105 may be located in distributed locations (e.g., separate physical locations). In some examples, one or more network entities 105 of the decomposed RAN architecture may be implemented as virtual units (e.g., virtual CU (VCU), virtual DU (VDU), virtual RU (VRU)).

[0065] The functional split between CU 160, DU 165, and RU 170 is flexible and may support different functionality depending on which functions (e.g., network layer functions, protocol layer functions, baseband functions, RF functions, and any combination thereof) are performed at CU 160, DU 165, or RU 170. For example, a functional split of a protocol stack may be employed between CU 160 and DU 165 such that CU 160 may support one or more layers of a protocol stack and DU 165 may support one or more different layers of a protocol stack. In some examples, CU 160 may host higher protocol layer (e.g., layer 3 (L3), layer 2 (L2)) functionality and signaling (e.g., RRC, SDAP, PDCP). The CU 160 may be connected to one or more DUs 165 or RUs 170, and the one or more DUs 165 or RUs 170 may host lower protocol layers, such as layer 1 (L1) (e.g., PHY layer) or L2 (e.g., RLC layer, medium access control (MAC) layer) functionality and signaling, and may each be at least partially controlled by the CU 160. Additionally or alternatively, a functional split of the protocol stack may be employed between the DU 165 and the RU 170, such that the DU 165 may support one or more layers of the protocol stack, and the RU 170 may support one or more different layers of the protocol stack. The DU 165 may support one or more different cells (e.g., via one or more RUs 170). In some cases, the functional split between CU 160 and DU 165 or between DU 165 and RU 170 may be within a protocol layer (e.g., some functions of a protocol layer may be performed by one of CU 160, DU 165, or RU 170, while other functions of the protocol layer are performed by different ones of CU 160, DU 165, or RU 170). CU 160 may be further functionally split into CU control plane (CU-CP) and CU user plane (CU-UP) functions. CU 160 may be connected to one or more DUs 165 via midhaul communication links 162 (e.g., F1, F1-c, F1-u), and DU 165 may be connected to one or more RUs 170 via fronthaul communication links 168 (e.g., an open fronthaul (FH) interface). In some examples, midhaul communication link 162 or fronthaul communication link 168 may be implemented based on interfaces (eg, channels) between layers of a protocol stack supported by respective network entities 105 that communicate via those communication links.

[0066] In some wireless communication systems (e.g., wireless communication system 100), infrastructure and spectrum resources for radio access may support wireless backhaul link capabilities to supplement wired backhaul connections, thereby providing an IAB network architecture (e.g., to core network 130). In some cases, in an IAB network, one or more network entities 105 (e.g., IAB nodes 104) may be partially controlled by each other. One or more IAB nodes 104 may be referred to as donor entities or IAB donors. One or more DUs 165 or one or more RUs 170 may be controlled in part by one or more CUs 160 associated with a donor network entity 105 (e.g., donor base station 140). One or more donor network entities 105 (e.g., IAB donors) may communicate with one or more additional network entities 105 (e.g., IAB nodes 104) via supported access and backhaul links (e.g., backhaul communication links 120). The IAB node 104 may include an IAB mobile terminal (IAB-MT) controlled (e.g., scheduled) by a coupled IAB donor's DU 165. The IAB-MT may include an independent set of antennas for relaying communications with the UE 115, or may share the same antennas of the IAB node 104 (e.g., of the RU 170) for access via the DU 165 of the IAB node 104 (e.g., referred to as a virtual IAB-MT (vIAB-MT)). In some examples, the IAB node 104 may include a DU 165 that supports communication links with additional entities (e.g., IAB node 104, UE 115) within a relay chain or configuration (e.g., downstream) of the access network. In such cases, one or more components of the decomposed RAN architecture (e.g., one or more IAB nodes 104 or components of the IAB node 104) may be configured to operate according to the techniques described herein.

[0067] For example, an access network (AN) or RAN may include an access node (e.g., an IAB donor), communications between an IAB node 104, and one or more UEs 115. The IAB donor may facilitate connectivity between the core network 130 and the AN (e.g., via a wired or wireless connection to the core network 130). That is, the IAB donor may refer to a RAN node having a wired or wireless connection to the core network 130. The IAB donor may include a CU 160 and at least one DU 165 (e.g., and RU 170), in which case the CU 160 may communicate with the core network 130 via an interface (e.g., a backhaul link). The IAB donor and the IAB node 104 may communicate via an F1 interface according to a protocol defining signaling messages (e.g., an F1 AP protocol). Additionally or alternatively, CU 160 may communicate with the core network via an interface (which may be an example of a portion of a backhaul link) and may communicate with other CUs 160 (e.g., CUs 160 associated with alternative IAB donors) via an Xn-C interface (which may be an example of a portion of a backhaul link).

[0068] The IAB node 104 may refer to a RAN node that provides IAB functions (e.g., access for UE 115, wireless self-backhaul capabilities, etc.). The DU 165 may act as a distributed scheduling node toward a child node associated with the IAB node 104, and the IAB-MT may act as a scheduled node toward a parent node associated with the IAB node 104. That is, the IAB donor may be referred to as a parent node that communicates with one or more child nodes (e.g., the IAB donor may relay the UE's transmission through one or more other IAB nodes 104). Additionally or alternatively, the IAB node 104 may also be referred to as a parent node or child node of other IAB nodes 104, depending on the relay chain or configuration of the AN. Therefore, the IAB-MT entity of the IAB node 104 may provide a Uu interface for the child IAB node 104 to receive signaling from the parent IAB node 104, and the DU interface (e.g., DU 165) may provide a Uu interface for the parent IAB node 104 to signal to the child IAB node 104 or the UE 115.

[0069] For example, the IAB node 104 may be referred to as a parent node supporting communications for a child IAB node or as a child IAB node associated with an IAB donor, or both. The IAB donor may include a CU 160 having a wired or wireless connection (e.g., backhaul communication link 120) to the core network 130, and may act as a parent node of the IAB node 104. For example, the DU 165 of the IAB donor may relay the transmission to the UE 115 through the IAB node 104, or may directly signal the transmission to the UE 115, or both. The CU 160 of the IAB donor may signal the IAB node 104 via the F1 interface that the communication link is established, and the IAB node 104 may schedule the transmission (e.g., the transmission relayed from the IAB donor to the UE 115) through the DU 165. That is, data may be relayed to and from the IAB node 104 via the NR Uu interface of the MT to the IAB node 104 via signaling. Communications with the IAB node 104 may be scheduled by the DU 165 of the IAB donor, and communications with the IAB node 104 may be scheduled by the DU 165 of the IAB node 104 .

[0070] In the case where the techniques described herein are applied to the context of a decomposed RAN architecture, one or more components of the decomposed RAN architecture may be configured to support a cyclic equivariant inference engine for channel estimation as described herein. For example, some operations described as being performed by a UE 115 or a network entity 105 (e.g., a base station 140) may additionally or alternatively be performed by one or more components of the decomposed RAN architecture (e.g., an IAB node 104, a DU 165, a CU 160, a RU 170, a RIC 175, a SMO 180).

[0071] UE 115 may include or may be referred to as a mobile device, a wireless device, a remote device, a handheld device, or a subscriber device, or some other suitable term, where a "device" may also be referred to as a unit, a station, a terminal, or a client, etc. UE 115 may also include or may be referred to as a personal electronic device, such as a cellular phone, a personal digital assistant (PDA), a tablet computer, a laptop computer, or a personal computer. In some examples, UE 115 may include or may be referred to as a wireless local loop (WLL) station, an Internet of Things (IoT) device, an Internet of Everything (IoE) device, or a machine type communication (MTC) device, etc., which may be implemented in various objects such as appliances or vehicles, meters, etc.

[0072] The UE 115 described herein may be capable of communicating with various types of devices such as other UEs 115 which may sometimes act as relays, as well as network entities 105 and network equipment including macro eNBs or gNBs, small cell eNBs or gNBs, or relay base stations, etc. Figure 1 shown.

[0073] The UE 115 and the network entity 105 may wirelessly communicate with each other via one or more communication links 125 (e.g., access links) using resources associated with one or more carriers. The term "carrier" may refer to a collection of RF spectrum resources having a physical layer structure defined for supporting the communication link 125. For example, a carrier for the communication link 125 may include a portion of an RF spectrum band (e.g., a bandwidth portion (BWP)) that operates according to one or more physical layer channels for a given radio access technology (e.g., LTE, LTE-A, LTE-A Pro, NR). Each physical layer channel may carry acquisition signaling (e.g., synchronization signals, system information), control signaling for coordinating carrier operations, user data, or other signaling. The wireless communication system 100 may support communications with the UE 115 using carrier aggregation or multi-carrier operation. Depending on the carrier aggregation configuration, the UE 115 may be configured to have multiple downlink component carriers and one or more uplink component carriers. Carrier aggregation may be used for both frequency division duplex (FDD) and time division duplex (TDD) component carriers. Communication between the network entity 105 and other devices may refer to communication between these devices and any portion (e.g., entity, sub-entity) of the network entity 105. For example, the terms "send," "receive," or "communicate" when referring to the network entity 105 may refer to any portion of a network entity 105 (e.g., base station 140, CU 160, DU 165, RU 170) of a RAN communicating with another device (e.g., directly or via one or more other network entities 105).

[0074] The signal waveform transmitted via the carrier may include multiple subcarriers (e.g., using multi-carrier modulation (MCM) techniques, such as orthogonal frequency division multiplexing (OFDM) or discrete Fourier transform spread OFDM (DFT-S-OFDM)). In a system using MCM techniques, a resource element may refer to a resource of one symbol period (e.g., the duration of one modulation symbol) and one subcarrier, in which case the symbol period and the subcarrier spacing may be inversely related. The number of bits carried by each resource element may depend on the modulation scheme (e.g., the order of the modulation scheme, the coding rate of the modulation scheme, or both), so that a relatively high number of resource elements (e.g., in the transmission duration) and a relatively high order modulation scheme may correspond to a relatively high communication rate. Wireless communication resources may refer to a combination of RF spectrum resources, time resources, and spatial resources (e.g., spatial layers or beams), and the use of multiple spatial resources may increase the data rate or data integrity used for communication with UE 115.

[0075] The time interval for the network entity 105 or the UE 115 may be expressed in multiples of a basic time unit, which may be, for example, a sampling period T s =1 / (Δf max ·N f ) seconds, where Δf max It can represent the supported subcarrier spacing, and N f The supported discrete Fourier transform (DFT) size may be indicated. The time intervals of the communication resources may be organized according to radio frames each having a specified duration (e.g., 10 milliseconds (ms)). Each radio frame may be identified by a system frame number (SFN) (e.g., ranging from 0 to 1023).

[0076] Each frame may include multiple consecutively numbered subframes or time slots, and each subframe or time slot may have the same duration. In some examples, the frame may be divided into subframes (e.g., in the time domain), and each subframe may be further divided into a certain number of time slots. Alternatively, each frame may include a variable number of time slots, and the number of time slots may depend on the subcarrier spacing. Each time slot may include a certain number of symbol periods (e.g., depending on the length of the cyclic prefix appended in front of each symbol period). In some wireless communication systems 100, the time slot may be further divided into a plurality of micro time slots associated with one or more symbols. Excluding the cyclic prefix, each symbol period may be associated with one or more (e.g., N f The duration of a symbol period may depend on the subcarrier spacing or the operating frequency band.

[0077] A subframe, a time slot, a mini-time slot, or a symbol may be the smallest scheduling unit (e.g., in the time domain) of the wireless communication system 100 and may be referred to as a transmission time interval (TTI). In some examples, the TTI duration (e.g., the number of symbol periods in a TTI) may be variable. Additionally or alternatively, the smallest scheduling unit of the wireless communication system 100 may be dynamically selected (e.g., in a burst of a shortened TTI (sTTI)).

[0078] Physical channels may be multiplexed according to various techniques in order to communicate using a carrier. For example, physical control channels and physical data channels may be multiplexed using one or more of a time division multiplexing (TDM) technique, a frequency division multiplexing (FDM) technique, or a hybrid TDM-FDM technique to signal via a downlink carrier. A control region (e.g., a control resource set (CORESET)) of a physical control channel may be defined by a set of symbol periods and may extend across a system bandwidth of a carrier or a subset of that system bandwidth. One or more control regions (e.g., CORESETs) may be configured for a set of UEs 115. For example, one or more UEs in UE 115 may monitor or search a control region to obtain control information according to one or more search space sets, and each search space set may include one or more control channel candidates in one or more aggregation levels arranged in a cascaded manner. The aggregation level of a control channel candidate may refer to the amount of control channel resources (e.g., control channel elements (CCEs)) associated with coded information for a control information format having a given payload size. The search space sets may include a common search space set configured for transmitting control information to multiple UEs 115 , and a UE-specific search space set for transmitting control information to a specific UE 115 .

[0079] In some examples, the network entities 105 (e.g., base stations 140, RUs 170) may be mobile and thus provide communication coverage for mobile coverage areas 110. In some examples, different coverage areas 110 associated with different technologies may overlap, but the different coverage areas 110 may be supported by the same network entity 105. In some other examples, overlapping coverage areas 110 associated with different technologies may be supported by different network entities 105. The wireless communication system 100 may include, for example, a heterogeneous network in which different types of network entities 105 provide coverage for various coverage areas 110 using the same or different radio access technologies.

[0080] The wireless communication system 100 may be configured to support ultra-reliable communication or low-latency communication or various combinations thereof. For example, the wireless communication system 100 may be configured to support ultra-reliable low-latency communication (URLLC). UE 115 may be designed to support ultra-reliable or low-latency or critical functions. Ultra-reliable communication may include private communication or group communication, and may be supported by one or more services (such as push-to-talk, video, or data). Support for ultra-reliable, low-latency functions may include prioritization of services, and such services may be used for public safety or general commercial applications. The terms ultra-reliable, low-latency, and ultra-reliable low-latency may be used interchangeably herein.

[0081] In some examples, the UE 115 may be configured to support communication directly with other UEs 115 via a device-to-device (D2D) communication link 135 (e.g., according to a peer-to-peer (P2P), D2D, or sidelink protocol). In some examples, one or more UEs 115 in a group that are performing D2D communication may be within a coverage area 110 of a network entity 105 (e.g., a base station 140, a RU 170), which may support aspects of such D2D communication configured (e.g., scheduled) by the network entity 105. In some examples, one or more UEs 115 in such a group may be outside of the coverage area 110 of the network entity 105, or may otherwise be unable or not configured to receive transmissions from the network entity 105. In some examples, a group of UEs 115 communicating via D2D communication may support a one-to-many (1:M) system, in which each UE 115 transmits to each of the other UEs 115 in the group. In some examples, network entity 105 may facilitate scheduling of resources for D2D communications. In some other examples, D2D communications may be performed between UEs 115 without involving network entity 105.

[0082] The core network 130 may provide user authentication, access authorization, tracking, Internet Protocol (IP) connectivity, and other access, routing, or mobility functions. The core network 130 may be an evolved packet core (EPC) or a 5G core (5GC), which may include at least one control plane entity (e.g., a mobility management entity (MME), an access and mobility management function (AMF)) for managing access and mobility and at least one user plane entity (e.g., a serving gateway (S-GW), a packet data network (PDN) gateway (P-GW), or a user plane function (UPF)) for routing packets or interconnecting to an external network. The control plane entity may manage non-access stratum (NAS) functions, such as mobility, authentication, and bearer management of UE 115 served by a network entity 105 (e.g., a base station 140) associated with the core network 130. User IP packets may be delivered through the user plane entity, which may provide IP address allocation and other functions. The user plane entity may be connected to the IP service 150 of one or more network operators. IP services 150 may include access to the Internet, an intranet, an IP Multimedia Subsystem (IMS), or packet-switched streaming services.

[0083] The wireless communication system 100 may operate using one or more frequency bands that may be in the range of 300 megahertz (MHz) to 300 gigahertz (GHz). Generally speaking, the region from 300 MHz to 3 GHz is referred to as the ultra-high frequency (UHF) region or decimeter band because the wavelength ranges from about one decimeter to one meter in length. UHF waves may be blocked or redirected by buildings and environmental features (which may be referred to as clusters), but these waves may be sufficient to penetrate structures so that macro cells provide services to UEs 115 located indoors. Communications using UHF waves may be associated with smaller antennas and shorter ranges (e.g., less than 100 kilometers) than communications using the lower frequencies and longer wavelengths of the high frequency (HF) or very high frequency (VHF) portions of the spectrum below 300 MHz.

[0084] The wireless communication system 100 may utilize licensed and unlicensed RF spectrum bands. For example, the wireless communication system 100 may use unlicensed bands (such as the 5 GHz Industrial, Scientific, and Medical (ISM) band) to employ license assisted access (LAA), LTE unlicensed (LTE-U) radio access technology, or NR technology. When operating using unlicensed RF spectrum bands, devices such as network entity 105 and UE 115 may employ carrier sensing for conflict detection and avoidance. In some examples, operations performed using unlicensed bands may be based on carrier aggregation configuration (e.g., LAA) in conjunction with component carriers operating using licensed bands. Operations performed using unlicensed spectrum may include downlink transmissions, uplink transmissions, P2P transmissions, or D2D transmissions, among others.

[0085] The electromagnetic spectrum is typically subdivided into various categories, bands, channels, etc. based on frequency / wavelength. In 5G NR, two initial operating bands have been identified as frequency range designations FR1 (410 MHz–7.125 GHz) and FR2 (24.25 GHz–52.6 GHz). It should be understood that, although a portion of FR1 is greater than 6 GHz, FR1 is often (interchangeably) referred to as the “sub-6 GHz” band in various documents and articles. A similar naming issue sometimes occurs with respect to FR2, which is often (interchangeably) referred to as the “millimeter wave” band in documents and articles, although different from the extremely high frequency (EHF) band (30 GHz–300 GHz) identified as the “millimeter wave” band by the International Telecommunication Union (ITU).

[0086] Frequencies between FR1 and FR2 are generally referred to as mid-band frequencies. Recent 5G NR research has identified the operating bands for these mid-band frequencies as frequency range designation FR3 (7.125 GHz–24.25 GHz). The bands falling within FR3 may inherit FR1 characteristics and / or FR2 characteristics, and thus may effectively extend the features of FR1 and / or FR2 to mid-band frequencies. In addition, higher frequency bands are currently being explored to extend 5G NR operations to more than 52.6 GHz. For example, three higher operating bands have been identified as frequency range designations FR4a or FR4–1 (52.6 GHz–71 GHz), FR4 (52.6 GHz–114.25 GHz), and FR5 (114.25 GHz–300 GHz). Each of these higher frequency bands falls within the EHF band.

[0087] In view of the above aspects, unless otherwise specifically stated, it should be understood that if used herein, the term "below 6 GHz" and the like can broadly refer to frequencies that may be less than 6 GHz, may be within FR1, or may include mid-band frequencies. In addition, unless otherwise specifically stated, it should be understood that if used herein, the term "millimeter wave" and the like can broadly refer to frequencies that may include mid-band frequencies, may be within FR2, FR4, FR4-a or FR4-1 and / or FR5, or may be within the EHF band.

[0088] The network entity 105 (e.g., base station 140, RU 170) or UE 115 may be equipped with multiple antennas, which can be used to employ technologies such as transmit diversity, receive diversity, MIMO communication or beamforming. The antennas of the network entity 105 or UE 115 may be located in one or more antenna arrays or antenna panels, which can support MIMO operation or transmit or receive beamforming. For example, one or more base station antennas or antenna arrays may be co-located at an antenna assembly, such as an antenna tower. In some examples, antennas or antenna arrays associated with the network entity 105 may be located at different geographical locations. The network entity 105 may include an antenna array having a collection of multiple rows and columns of antenna ports that the network entity 105 can use to support beamforming for communication with the UE 115. Similarly, the UE 115 may include one or more antenna arrays that can support various MIMO or beamforming operations. Additionally or alternatively, the antenna panel may support RF beamforming for signals sent via the antenna ports.

[0089] The network entity 105 or UE 115 may use MIMO communication to utilize multipath signal propagation and improve spectral efficiency by sending or receiving multiple signals via different spatial layers. Such a technology may be referred to as spatial multiplexing. The multiple signals may be sent, for example, by a transmitting device via different antennas or different combinations of antennas. Similarly, the multiple signals may be received by a receiving device via different antennas or different combinations of antennas. Each of the multiple signals may be referred to as a separate spatial stream and may carry information associated with the same data stream (e.g., the same codeword) or different data streams (e.g., different codewords). Different spatial layers may be associated with different antenna ports for channel measurement and reporting. MIMO technologies include: single-user MIMO (SU-MIMO), for which multiple spatial layers are sent to the same receiving device; and multi-user MIMO (MU-MIMO), for which multiple spatial layers are sent to multiple devices.

[0090] Beamforming (which may also be referred to as spatial filtering, directional transmission, or directional reception) is a signal processing technique that may be used at a transmitting device or a receiving device (e.g., a network entity 105, a UE 115) to shape or direct an antenna beam (e.g., a transmit beam, a receive beam) along a spatial path between the transmitting device and the receiving device. Beamforming may be achieved by combining signals conveyed via antenna elements of an antenna array so that some signals propagating in a particular direction relative to the antenna array experience constructive interference, while other signals experience destructive interference. Adjustments to signals conveyed via antenna elements may include the transmitting device or the receiving device applying an amplitude offset, a phase offset, or both to signals carried via antenna elements associated with the device. Adjustments associated with each of these antenna elements may be defined by a set of beamforming weights associated with a particular direction (e.g., relative to the antenna array of the transmitting device or the receiving device or relative to some other direction).

[0091] The network entity 105 or UE 115 may use beam scanning techniques as part of a beamforming operation. For example, the network entity 105 (e.g., base station 140, RU 170) may use multiple antennas or antenna arrays (e.g., antenna panels) to perform beamforming operations for directional communication with UE 115. Some signals (e.g., synchronization signals, reference signals, beam selection signals, or other control signals) may be sent multiple times by the network entity 105 along different directions. For example, the network entity 105 may send signals according to different sets of beamforming weights associated with different transmission directions. Transmissions along different beam directions may be used to identify (e.g., by a transmitting device (such as network entity 105), or by a receiving device (such as UE 115)) beam directions for later transmission or reception by the network entity 105.

[0092] Some signals, such as data signals associated with a particular receiving device, may be sent by a transmitting device (e.g., transmitting network entity 105, transmitting UE 115) along a single beam direction (e.g., a direction associated with a receiving device (such as receiving network entity 105 or receiving UE 115)). In some examples, the beam direction associated with the transmission along the single beam direction may be determined based on signals sent along one or more beam directions. For example, UE 115 may receive one or more of the signals sent by network entity 105 along different directions, and may report to network entity 105 an indication of the signal received by UE 115 with the highest signal quality or other acceptable signal quality.

[0093] In some examples, transmission by a device (e.g., by network entity 105 or UE 115) may be performed using multiple beam directions, and the device may use a combination of digital precoding or beamforming to generate a combined beam for transmission (e.g., from network entity 105 to UE 115). UE 115 may report feedback indicating precoding weights for one or more beam directions, and the feedback may correspond to a configured set of beams across a system bandwidth or one or more subbands. Network entity 105 may send a reference signal (e.g., a cell-specific reference signal (CRS), a channel state information reference signal (CSI-RS)), which may or may not be precoded. UE 115 may provide feedback for beam selection, which may be a precoding matrix indicator (PMI) or codebook-based feedback (e.g., a multi-panel codebook, a linear combination codebook, a port selection codebook). Although these techniques are described with reference to signals sent along one or more directions by a network entity 105 (e.g., base station 140, RU 170), UE 115 may use similar techniques to send signals multiple times along different directions (e.g., to identify a beam direction for subsequent transmission or reception by UE 115), or to send signals along a single direction (e.g., to send data to a receiving device).

[0094] A receiving device (e.g., UE 115) may perform receiving operations according to multiple receiving configurations (e.g., directional listening) when receiving various signals (such as synchronization signals, reference signals, beam selection signals, or other control signals) from a transmitting device (e.g., network entity 105). For example, the receiving device may perform reception according to multiple receiving directions by: receiving via different antenna subarrays, processing the received signals according to different antenna subarrays, receiving according to different receiving beamforming weight sets (e.g., different directional listening weight sets) applied to signals received at multiple antenna elements of an antenna array, or processing the received signals according to different receiving beamforming weight sets applied to signals received at multiple antenna elements of an antenna array, any of which may be referred to as "listening" according to different receiving configurations or receiving directions. In some examples, the receiving device may use a single receiving configuration to receive along a single beam direction (e.g., when receiving a data signal). The single receive configuration may be aligned along a beam direction determined based on listening according to different receive configuration directions (e.g., a beam direction determined to have the highest signal strength, highest SNR, or other acceptable signal quality based on listening according to multiple beam directions).

[0095] The wireless communication system 100 may be a packet-based network that operates according to a layered protocol stack. In the user plane, communications at the bearer or PDCP layer may be IP-based. The RLC layer may perform packet segmentation and reassembly to communicate via logical channels. The MAC layer may perform priority processing and multiplexing of logical channels into transport channels. The MAC layer may also implement error detection techniques, error correction techniques, or both to support retransmission to improve link efficiency. In the control plane, the RRC layer may provide the establishment, configuration, and maintenance of an RRC connection that supports a radio bearer for user plane data between the UE 115 and the network entity 105 or the core network 130. The PHY layer may map a transport channel to a physical channel.

[0096] In some cases, the wireless communication system 100 may support SISO communication, MIMO communication, or both. For example, SISO communication may include communication between a single transmitter and a single receiver, while MIMO communication may include communication between multiple transmitters and multiple receivers. In some cases, MIMO communication may include multiple SISO communications via a transmitter and receiver pair (e.g., a transmit antenna and a receive antenna pair). For example, the network entity 105 may include a first transmit antenna and a second transmit antenna, and the UE 115 may include a first receive antenna and a second receive antenna. The antenna pair may include a first transmission using a first receive antenna, a first transmission using a second receive antenna, a second transmission using a first receive antenna, and a second transmission using a second receive antenna. Each antenna pair may modify the signal using a beamforming matrix (e.g., a precoding matrix, an orthogonal matrix) before the signal is transmitted to minimize interference (e.g., a linear precoder, a beamformer may create a beam that focuses energy for each receive antenna by weighting the phase and magnitude of the transmit antenna). Because the beamforming matrix may be unknown to the receiver (e.g., UE 115, network entity 105), the receiver may estimate the precoding channel.

[0097] In some examples, the wireless communication system 100 may support channel estimation. For example, channel estimation may be used for a wireless MIMO-OFDM system based on a resource grid (e.g., based on a time slot). The channel estimation may be a 5G NR channel estimation with a varying DMRS pattern, number of resource blocks, etc. The channel estimation may be used for super-resolution or signal recovery based on sparse observations.

[0098] In MIMO communication, multiple layers of information (e.g., communications) may interfere with each other, making channel estimation more complex. In some cases, orthogonal cover codes may be used to remove interference through a despreading step. However, the despreading step may be insufficient for frequency selective or fast fading channels (e.g., high delay spread and high Doppler response). Additionally, narrowband MIMO communications (e.g., communications using blocks of relatively small bandwidth portions) may use different pre-coding matrices for each resource group (e.g., physical resource groups (PRGs)) that may be unknown to UE 115. Unknown pre-coding matrices may add higher complexity to channel estimation, so that some wireless communication systems may not use correlations between non-continuous PRGs in channel estimation. For example, some estimation techniques (e.g., MMSE) may utilize TRS or another continuous reference signal to calculate channel characteristics (e.g., Doppler, delay spread, SNR, etc.) and estimate channel resources based on the channel characteristics.

[0099] In some cases, some estimation techniques may include least squares and linear MMSE (LMMSE). Least squares may not use information about channel statistics or noise variance, and may not model correlations across different PRGs and MIMO layers, making it relatively simple to implement with low computational overhead. However, the estimation accuracy may be insufficient for most use cases (e.g., practical applications). LMMSE may utilize second-order channel statistics and noise variance (e.g., a binning-based strategy based on estimated channel parameters (such as Doppler, delay spread, etc.)). Under some conditions, LMMSE may have high computational cost and have relatively low estimation error. However, LMMSE may not model correlations across different PRGs and MIMO layers.

[0100] In some cases, some estimation techniques may include deep learning based techniques. Deep learning based techniques may not utilize explicit information of channel statistics and may utilize nonlinear interpolation. Unlike LMMSE, deep learning techniques may not utilize large matrix inversion operations. Deep learning techniques may utilize separate networks for each DMRS pattern. In some cases, deep learning techniques may not consider correlations across different PRGs.

[0101] The techniques described herein allow channel estimates to be computed using a cyclic equivariant inference engine, which can result in channel estimates based on DMRS symbols that exploit correlations between non-adjacent PRGs (e.g., pre-decoder agnostic). In some cases, a wireless communication system 100 (e.g., an OFDM system) can deploy pilot-based channel estimation techniques to obtain CSI relatively accurately (e.g., obtaining accurate CSI can help maintain high data throughput, for example, in a fast fading environment). The pilot signal can be referred to as a DMRS symbol. DMRS symbols can be inserted into a transmit slot according to a known DMRS pattern, allowing a wireless device to estimate unknown resources of a channel (e.g., non-DMRS locations, resources allocated for data signals) based on known resources (e.g., DMRS symbols). In some cases, the DMRS pattern can be preconfigured (e.g., a fixed set of possible DMRS patterns). The DMRS pattern can be used based on channel characteristics.

[0102] In some examples, a wireless device may receive an assignment of a set of resources associated with a channel, wherein the set of resources includes a first subset of resources allocated for data transmission (e.g., non-DMRS) and a second subset of resources allocated for reference signals (e.g., DMRS). The wireless device may generate multiple channel estimates (e.g., SISO channel estimates) for each layer of the channel, and perform a refinement operation using the estimates to generate channel estimates associated with multiple layers (e.g., MIMO channel estimates). In some cases, the refinement operation may include multiple iterations. For example, each iteration may include generating a corresponding gradient associated with each channel estimate based on the channel estimate for each layer and observed resources (e.g., known DMRS resources) of the channel; generating a current set of values ​​of the latent variable (e.g., inference variables based on observed variables) based on a previous set of values ​​of the latent variable, the corresponding gradient, and the channel estimate for each layer; and modifying (e.g., refining, updating, improving) the channel estimate based on the current set of values ​​of the latent variable, the channel estimate for each layer, and the corresponding gradient. In some cases, the refinement operation may be performed by a refinement network including a likelihood module, an encoder module, and a decoder module.

[0103] In some cases, the wireless communication system 100 may incorporate end-to-end (E2E) use of a neural network for channel state feedback. Such a neural network structure may be used to provide channel state information feedback (CSF) by providing an intermediate channel representation to the wireless communication network, and the wireless device (e.g., receiving wireless device, network entity 105) may reconstruct the channel (e.g., separate implementation of the proposed method at the UE and network sides), so that this type of model architecture may have some degree of interoperability specifications.

[0104] Figure 2An example of a network architecture 300 (e.g., a disaggregated base station architecture, a disaggregated RAN architecture) supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is illustrated. The network architecture 200 may illustrate an example for implementing one or more aspects of the wireless communication system 100. The network architecture 200 may include one or more CUs 160-a, which may communicate directly with the core network 130-a via a backhaul communication link 120-a, or indirectly with the core network 130-a through one or more disaggregated network entities 105 (e.g., a near RT RIC 175-b via an E2 link or a non-RT RIC 175-a associated with an SMO 180-a (e.g., an SMO framework), or both. The CU 160-a may communicate with one or more DUs 165-a via corresponding midhaul communication links 162-a (e.g., an F1 interface). The DU 165-a may communicate with one or more RUs 170-a via corresponding fronthaul communication links 168-a. The RUs 170-a may be associated with respective coverage areas 110-a and may communicate with the UEs 115-a via one or more communication links 125-a. In some implementations, a UE 115-a may be simultaneously served by multiple RUs 170-a.

[0105] Each of the network entities 105 of the network architecture 200 (e.g., CU 160-a, DU 165-a, RU 170-a, non-RT RIC 175-a, near-RT RIC 175-b, SMO 180-a, open cloud (O-Cloud) 205, open eNB (O-eNB) 210) may include one or more interfaces or may be coupled to one or more interfaces configured to receive or send signals (e.g., data, information) via a wired or wireless transmission medium. Each network entity 105 or an associated processor (e.g., a controller) that provides instructions to an interface of the network entity 105 may be configured to communicate with one or more of the other network entities 105 via a transmission medium. For example, these network entities 105 may include a wired interface that is configured to receive signals on a wired transmission medium or to send signals to one or more of the other network entities 105 on a wired transmission medium. Additionally or alternatively, the network entity 105 may include a wireless interface, which may include a receiver, transmitter, or transceiver (e.g., an RF transceiver) configured to receive signals on a wireless transmission medium, or to send signals to one or more of the other network entities 105 on a wireless transmission medium, or both.

[0106] In some examples, CU 160-a may host one or more higher layer control functions. Such control functions may include RRC, PDCP, SDAP, etc. Each control function may be implemented using an interface configured to communicate signals with other control functions hosted by CU 160-a. CU 160-a may be configured to process user plane functions (e.g., CU-UP), control plane functions (e.g., CU-CP), or a combination thereof. In some examples, CU 160-a may be logically split into one or more CU-UP units and one or more CU-CP units. When implemented in an O-RAN configuration, the CU-UP unit may communicate bidirectionally with the CU-CP unit via an interface (such as an E1 interface). As needed, CU 160-a may be implemented to communicate with DU 165-a for network control and signaling.

[0107] DU 165-a may correspond to a logical unit that includes one or more functions (e.g., base station functions, RAN functions) for controlling the operation of one or more RUs 170-a. In some examples, DU 165-a may at least partially host one or more of the RLC layer, the MAC layer, and one or more aspects of the PHY layer (e.g., a high PHY layer, such as modules for FEC encoding and decoding, scrambling, modulation and demodulation, etc.), which depends at least in part on functional divisions, such as those defined by the Third Generation Partnership Project (3GPP). In some examples, DU 165-a may also host one or more low PHY layers. Each layer may be implemented using an interface that is configured to communicate signals with other layers hosted by DU 165-a or with control functions hosted by CU160-a.

[0108] In some examples, lower layer functions may be implemented by one or more RUs 170-a. For example, a RU 170-a controlled by a DU 165-a may correspond to a logical node that hosts RF processing functions or low PHY layer functions (e.g., performing fast Fourier transform (FFT), inverse FFT (iFFT), digital beamforming, physical random access channel (PRACH) extraction and filtering, etc.) or both based at least in part on functional splitting (such as lower layer functional splitting). In this architecture, the RU 170-a may be implemented to handle over-the-air (OTA) communications with one or more UEs 115-a. In some specific implementations, real-time and non-real-time aspects of control plane and user plane communications with the RU 170-a may be controlled by the corresponding DU 165-a. In some examples, this configuration may enable the DU 165-a and CU 160-a to be implemented in a cloud-based RAN architecture (such as a vRAN architecture).

[0109] The SMO 180-a may be configured to support RAN deployment and provisioning of non-virtualized and virtualized network entities 105. For non-virtualized network entities 105, the SMO 180-a may be configured to support the deployment of dedicated physical resources for RAN coverage requirements, which may be managed via an operation and maintenance interface (e.g., an O1 interface). For virtualized network entities 105, the SMO 180-a may be configured to interact with a cloud computing platform (e.g., O-Cloud 205) via a cloud computing platform interface (e.g., an O2 interface) to perform network entity lifecycle management (e.g., to instantiate virtualized network entities 105). Such virtualized network entities 105 may include, but are not limited to, CU 160-a, DU 165-a, RU 170-a, and near-RT RIC 175-b. In some specific implementations, the SMO 180-a may communicate with components configured according to a 4G RAN (e.g., via an O1 interface). Additionally or alternatively, in some implementations, the SMO 180-a may communicate directly with one or more RUs 170-a via an O1 interface. The SMO 180-a may also include a non-RT RIC 175-a configured to support the functionality of the SMO 180-a.

[0110] The non-RT RIC 175-a may be configured to include logic functions that implement non-real-time control and optimization of RAN elements and resources, artificial intelligence (AI) or machine learning (ML) workflows (including model training and updating, or policy-based guidance of applications / features in the near-RT RIC 175-b). The non-RT RIC 175-a may be coupled to or communicate with the near-RT RIC 175-b (e.g., via an A1 interface). The near-RT RIC 175-b may be configured to include logic functions that implement near real-time control and optimization of RAN elements and resources via data collection and actions on an interface connecting one or more CUs 160-a, one or more DUs 165-a, or both, and the O-eNB 210 to the near-RT RIC 175-b (e.g., via an E2 interface).

[0111] In some examples, in order to generate an AI / ML model to be deployed in the near-RT RIC 175-b, the non-RT RIC 175-a may receive parameters or external enrichment information from an external server. Such information may be utilized by the near-RT RIC 175-b and may be received from a non-network data source or from a network function at the SMO 180-a or the non-RT RIC 175-a. In some examples, the non-RT RIC 175-a or the near-RT RIC 175-b may be configured to tune RAN behavior or performance. For example, the non-RT RIC 175-a may monitor long-term trends and patterns of performance and employ an AI model or ML model to perform corrective actions through the SMO 180-a (e.g., via reconfiguration of O1) or via the generation of a RAN management policy (such as an A1 policy).

[0112] Figure 3 An example of a network 300 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. In some examples, the network 300 can be implemented by aspects of the wireless communication system 100. For example, the network 300 can be implemented by the UE 115, the network entity 105, or both, as described herein with reference to Figure 1 and Figure 2 described.

[0113] In some examples, a wireless device (e.g., UE 115, network entity 105) may use multiple resources (e.g., time resources, frequency resources, etc.) to send signals via one or more time slots (e.g., frequency and time grids). Resource elements (e.g., symbols 315, DMRS 320, single subcarriers for a single OFDM symbol) may be grouped with multiple resource elements to form a physical resource block (PRB) (e.g., PRB 310). Each column of resources (e.g., resource elements of the same time resources and different frequency resources) of PRB 310 may be considered as a single resource block (e.g., OFDM symbol). In some cases, PRB 310 may include twelve subcarrier frequencies (in the frequency domain) and fourteen OFDM symbols (in the time domain). Multiple PRBs 310 may be bundled (e.g., grouped, combined) to form a single PRG, such as PRG 305. In some cases, the PRG 305 may include a number of PRBs 310 determined by a bundle size parameter (e.g., bundleSize, the number of consecutive PRBs stacked together in a single PRG). For example, the bundle size parameter may indicate two PRBs 310 or four PRBs 310 (e.g., four consecutive resource blocks) for narrowband precoding operation or zero PRBs 310 (e.g., no stacked PRBs) for wideband precoding operation and form part of the bandwidth portion.

[0114] In some cases, a wireless communication system (e.g., wireless communication system 100) may utilize pilot-based channel estimation techniques. Pilot symbols may be referred to as DMRS symbols (e.g., DMRS symbols 320). For example, a wireless device may transmit a signal including one or more PRBs 310 via a physical downlink shared channel (PDSCH) (e.g., a channel for user data). PRBs 310 may include various symbols 315 (e.g., resource elements allocated for data signals), various DMRS symbols 320 (e.g., resource elements configured for demodulation reference signals), and in some cases null symbols (e.g., resource elements to which no data is allocated).

[0115] DMRS symbols 320 may be inserted in various resource elements of PRB 310 according to a resource configuration pattern (e.g., a DMRS pattern). For example, a wireless device may be configured with various DMRS patterns and select a DMRS pattern 345 for the signal. In some cases, the various DMRS patterns may include DMRS symbols inserted in adjacent resource elements, in every other resource element, in a single resource block of a PRB, in multiple resource blocks of a PRB, and in other potential configurations.

[0116] In some examples, the DMRS pattern 345 may be known by the receiving wireless device. For example, the receiving wireless device may receive the signal and determine which symbols of the PRG 305 include various DMRS 320 based on the DMRS pattern 345. To extract (e.g., process, determine, decode, estimate) data at the symbol 315, the receiving wireless device may perform a channel estimation process. For example, according to Equation 1, the received DMRS 320 (e.g., y i ) can be equal to noise (e.g., interference, n i ) and channels (e.g., PDSCH, h i ) and the original data (e.g., x i ) is a combination of the products of .

[0117] Equation 1: y i =h i ⊙x i +n i

[0118] Because DMRS 320 is a pilot symbol known to both the transmitting wireless device (e.g., UE 115, network entity 105) and the receiving wireless device (e.g., UE 115, network entity 105), the receiving wireless device can use the known DMRS 320 to determine the pilot symbol. i and x i (given some noise) to estimate h i(e.g., extract the channel at the DMRS location). The receiving wireless device may then interpolate the estimated channel across various symbols 315 (e.g., remaining resource elements) (e.g., for image restoration) to extract (e.g., calculate) the data at the symbol 315. In some cases, n i This may include cross-MIMO interference, inter-PRG interference, intra-PRG interference, etc. Some channel estimation techniques may not consider (eg, calculate) these types of interference (eg, noise), which may result in inaccurate channel estimates and inaccurate data estimates.

[0119] In some implementations, the signal may include multiple PRGs 305, where each PRG 305 may be configured according to a unique precoding matrix 305, each PRG is precoded according to a unique precoding matrix. The unique precoding matrix can be found by performing a singular value decomposition (SVD) on the resource blocks within the PRG 305 or as a random (e.g., orthogonal) precoder. In some cases, according to Equation 2, the effective channel of the first PRG 305 (e.g., ) may be equal to the unique precoding matrix (v prg ) is the product of .

[0120] Equation 2: In some cases, a receiving wireless device may not be aware of which precoding matrix is ​​applied to which PRG 305 , and therefore, unique precoding of channels for each PRG 305 may prohibit smooth interpolation of channels between PRGs 305 .

[0121] In some implementations, MIMO communications may introduce additional complexity to the channel estimation formula (e.g., SISO channel estimation formula). For example, at the DMRS tone inference (e.g., extraction) step, the formula may include multiple unknown equations (e.g., multiple equations with multiple unknown variables), which may result in solving an undetermined inverse problem. The complexity may include determining the alignment of channels across different PRG beams and then exploiting the correlation across them. Additionally, these multiple layers may add additional multiplexing information (e.g., correlation across receiver and transmitter antenna pairs) that may be used to enhance channel estimation performance.

[0122] The techniques described herein allow channel estimates to be calculated using a cyclic equivariant inference engine. For example, a single neural network estimator is used for a variety of use cases of channel estimation including modular and interpretable model design (e.g., channel profile estimation, various DMRS pattern configurations, SNR, cross-MIMO estimation, inter-PRG estimation, intra-PRG estimation, etc.). In some specific implementations, channel estimation may include multiple stages (e.g., steps). For example, the first stage may include solving SISO channel estimates (e.g., for each transmitter and receiver antenna pair). The second stage may include using SISO channel estimates to solve MIMO channel estimates (e.g., learning correlations between antenna pairs). The third stage may include solving MIMO channel estimates within a PRG beam (e.g., within a PRG, learning correlations between resource elements within each PRG). The fourth stage may include solving MIMO channel estimates across PRG beams (e.g., between PRGs, learning correlations between resource elements across PRGs).

[0123] Additionally or alternatively, in some examples as described herein, a cyclic equivariant inference engine may be used and channel estimates may be calculated based on MMSE operations. MMSE operations may be estimation techniques that utilize linear equalization to estimate channels. MMSE may be supported by hardware within a wireless device, which may achieve reduced complexity and processing compared to performing initial estimates based on machine learning. The cyclic equivariant inference engine may represent an example of a machine learning model that provides a relatively reliable and accurate estimate of the type based on a given input set. For example, the first stage of channel estimation may include solving a SISO channel estimate (e.g., for each transmitter and receiver antenna pair) based on an MMSE operation 350 (e.g., which may also be referred to as an average MMSE (AMMSE) operation). A first set of multiple channel estimates per layer of a channel may be generated based on an MMSE operation 350. In this case, the remaining segments may be built on estimates generated by MMSE. In some cases, various stages of channel estimation may be iteratively performed (e.g., including multiple iterations of various stages). In some cases, one or more of the stages may utilize SNR estimates (e.g., genie values). Although four stages are described, the channel estimation process utilizing the described techniques may include more or fewer stages, stages including various other steps, stages without one or more of the described steps, or any combination thereof. Although these stages are described as four separate stages, they may be considered to be a continuous process.

[0124] In some cases, the network may perform various stages. For example, the network may include Figure 3 The rough network described and as referenced herein Figures 4 to 6The request network of the refinement network described. In some cases, the network may include u-net type (e.g., u-net 525) encoder (e.g., encoder 430) and decoder (e.g., decoder 435) convolution blocks, followed by an attention-based (e.g., attention 605) refinement network for longer range correlations. In some examples, the coarse network 325 may provide an initial channel estimate (e.g., a coarse estimate) per PRG per antenna pair (e.g., transmit and receiver antenna pair) by learned interpolation (e.g., smoothing). Additionally or alternatively, the MMSE operation 350 may provide an initial channel estimate per PRG per antenna pair, and the coarse network 325 may modify or update the channel estimate by learned interpolation.

[0125] Network 300 may represent a first stage. A wireless device (eg, as described herein with reference to Figure 1 The UE 115, the network entity 105, or both) may receive an assignment of a set of resources associated with a channel (e.g., a PDSCH channel). The set of resources may include a first subset of resources allocated for data signals (e.g., symbols 315) and a second subset of resources allocated for reference signals (e.g., DMRS symbols 320).

[0126] In some cases, the wireless device may perform MMSE operation 350 to generate a first set 355 of multiple channel estimates associated with a corresponding layer 360 of a channel for a set of resources (e.g., LS channel estimates at DMRS symbols 320). For example, the set of resources may include one or more PRGs 305, and the second subset of resources may include respective DMRS symbols 320 inserted into corresponding resource elements based on one resource configuration pattern (e.g., pattern 345) in a set of resource configuration patterns (e.g., a set of DMRS patterns). In some examples, one or more resources in the first set of resources including non-DMRS symbols 315 may be initialized with zero entries. Each corresponding layer 360 may be associated with a corresponding antenna pair (e.g., a transmitter and receiver pair) of a plurality of SISO antenna pairs. In some implementations, the initial values ​​of the first set 355 of multiple channel estimates may be associated with a SISO antenna pair. Compared to other channel estimation techniques, MMSE operation 350 may be associated with relatively low computational cost and processing. As described herein, network 300 may utilize MMSE channel estimates as input and may be built on MMSE estimates. For example, the MMSE operation 350 may be followed by one or more attention-based refinement operations (eg, machine learning-based channel estimation) for relatively long-range correlations, which may reduce complexity and processing compared to other channel estimation techniques.

[0127] After performing MMSE operation 350, the wireless device may use the first set of multiple channel estimates 355 to generate a second set of multiple channel estimates 335 using a network 325 (e.g., a coarse network). In some examples, network 325 may include or be based on a machine learning model for channel estimation. In some examples, network 325 may perform nonlinear two-dimensional interpolation of the channel based on the first set of multiple channel estimates 335. For example, network 325 may interpolate the channel in the time domain and the frequency domain (e.g., two-dimensional).

[0128] The network 325 may involve (e.g., input) a first set 355 of channel estimates generated by the MMSE operation 350 based on the PRG 305. The network 325 may perform various iterations 330 on the first set 355 of channel estimates. For example, the network 325 may include a u-net encoder-decoder fully convolutional network, where each iteration may include gated and gated dilated convolutional units. In some cases, for one or more iterations 330, the network 325 may copy and cascade the results of the previous iteration 330 to one or more iterations 330. In some examples, the first set 355 of multiple channel estimates and the second set 335 of multiple channel estimates may be channel estimates for corresponding SISO antenna pairs per PRG beam (e.g., Where h is the channel estimate, m is the PRG index, k is the PRB index, and i,j are the MIMO indices for a given antenna pair. and (f, t) is a resource element index within the two-dimensional grid of PRBs). For example, the wireless device may receive a corresponding PRG beam for each antenna pair. If the MIMO communication includes two transmit antennas and two receive antennas, there may be four antenna pairs, with multiple PRG beams for each antenna pair. The wireless device may utilize MMSE operation 350 to generate a first channel estimate 355 for each antenna pair and each PRG beam 305, and utilize network 325 to further generate a second channel estimate 335 for each antenna pair and each PRG beam 305. In some cases, network 325 may output The network 325 may additionally generate latent variables. For example, the latent variable may be an estimate 340 (e.g., a z estimate) that may not be directly observed but rather inferred from other observed parameters. The latent variable may be an abstract representation of the underlying channel characteristics (e.g., Doppler shift, delay spread). The network 325 (e.g., an embedding network) may be used as a feature extractor to generate initial latent values ​​(e.g., z τ＝0 ), which can be further used by subsequent refinement modules. In some examples, the output of network 325 can be an input to network 400 (eg, input 415).

[0129] In some cases, the techniques described herein can result in various advantages over other channel estimation techniques. For example, channel estimation via a cyclic equivariant inference engine (e.g., networks 300, 400, 500, and 600) can provide various signal processing and deep learning advantages. For example, signal processing can be based on DMRS (e.g., eliminating dependence on TRS, except for SNR), eliminating explicit parameter estimation (e.g., Doppler shift, delay spread, etc.), avoiding old binning strategies, utilizing relatively less memory and computational overhead (e.g., reducing maintenance of parameter libraries), modeling additional interactions (e.g., interference, cross-MIMO, intra-PRG, inter-PRG processing gain), abstracting an orthogonal cover code (OCC) despreading step (as described in reference to Figure 1 The invention relates to a method for implementing a deep learning channel estimation method for a wireless device 320 that can be used to estimate the performance of a wireless device 320. The method may include ...

[0130] Figure 4 An example of a network 400 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. In some examples, the network 400 can be implemented by aspects of the wireless communication system 100. For example, the network 400 can be implemented by the UE 115, the network entity 105, or both, as described herein with reference to Figure 1 and Figure 2 described.

[0131] In some cases, network 400 may support channel estimation. For example, the channel estimation problem may be defined as maximizing the posterior (e.g., In some cases, the conditional probability distribution (e.g., P(H|y,x)) can be parameterized by channel characteristics (e.g., delay-Doppler distribution). Figure 3 As described above, the latent variable (eg, z) may be an abstract representation of the underlying channel characteristics (eg, Doppler, delay spread, etc.).

[0132] In some cases, the output of network 300 may be an input to network 400. For example, input 415 may include one or more channel estimates (e.g., channel estimates 335 per SISO antenna pair per PRG 305) and corresponding estimates 340 (e.g., corresponding latent variables per channel estimate). Input 415 may be an input to refinement network 405, which may include various iterations 410. For example, a wireless device (e.g., network entity 105, UE 115) may perform a refinement operation on the channel estimates (e.g., via refinement network 405). The refinement operation may include multiple iterations of generating corresponding gradients (e.g., gradient 480) based on observations of channel estimates for a subset of resources (e.g., DMRS symbols) and measurements for the subset of resources (e.g., DMRS symbols 320), as described herein with reference to Figure 4 The method (e.g., module 425) generates a second set of latent variables based on the channel estimates and the corresponding gradients, as described herein with reference to Figure 4 and Figure 5 The (e.g., module 430); and modifying channel estimates associated with a plurality of layers based on a second set of latent variables, channel estimates, and corresponding gradients, as described herein with reference to Figure 5 Said (e.g., module 435).

[0133] In some cases, refinement network 405 may include iterative refinement performed by various refinement units. For example, each iteration 410 of refinement network 405 may be performed by various modules (e.g., three or four unique refinement modules). For example, iteration 410 may include module 425 (e.g., likelihood module), module 430 (e.g., encoder module), and module 435 (e.g., decoder module). Additionally or alternatively, iteration 410 may include a processor for performing Figure 4 In some examples, the refinement network 405 may include a corresponding set of machine learning parameters 485 associated with the machine learning operation. For example, the set of machine learning parameters 485 may be represented as θ. The set of machine learning parameters 485 may be set during a machine learning simulation (e.g., a pre-field operation). In some cases, the set of machine learning parameters 485 may include corresponding parameters unique to each iteration 410 (e.g., for T iterations, θ 1 ,θ 2 ,...θ T , where θ 1 ≠θ 2 ≠...≠θ T), because the input to each iteration may vary (e.g., unlike other conventional encoder and decoder machine learning implementations). In some other cases, the set of machine learning parameters 485 may be common in each iteration 410 to reduce the complexity or size of the implementation of the refinement network 405. That is, each iteration 410 may use the same parameters (e.g., θ for T iterations). 1 =θ 2 ,=...=θ T ).

[0134] In some cases, module 425 may output to module 430 and module 435, module 430 may output to module 435, and module 435 may output to the next iteration 410 or to the final (e.g., last, final) output (e.g., output 420) of refinement network 405. In some examples, the output of module 435 (e.g., output 420) may include channel estimates (e.g., a set of MIMO channel estimates for each of the PRGs, PRBs, and antenna pairs at the respective iterations 410) of the second module 425, the second module 430, and the second module 435 that are output to the next iteration 410. For example, the set of channel estimates (e.g., H) for iteration 410 (e.g., τ) may be τ ) can be equal to In some cases, the output of module 435 may include latent variables (e.g., a set of latent variables for each channel estimate) that are output to second module 425 and second module 430 of the next iteration 410. For example, for the set of latent variables for iteration 410 (e.g., z τ ) can be equal to

[0135] In some examples, at least a portion of input 415 can be an input to module 425. For example, as described herein with reference to Figure 3 The second set 335 of the plurality of channel estimates may be input to module 425. Additionally or alternatively, the observed DMRS symbols (eg, y dmrs ) and known DMRS symbols (e.g., x dmrs ) may be input to module 425 (e.g., (y dmrs ,x dmrs ,σ)). In some cases, module 425 may coordinate the decrease z and H and use circular reasoning as the gradient. For example, according to Equation 3, module 425 may be based on the measured observations of the resource subset (e.g., DMRS symbol 320) and the second set 335 (H) of multiple channel estimates associated with the resource subset. τ x) generates a corresponding set of values ​​for residual variable 465 (e.g., ). According to Equation 3, the modulus is the number of mask bits 475 (e.g., a binary mask).

[0136] Block 425 may combine a corresponding set of values ​​of residual variables 465, known observations 470 associated with a subset of resources (e.g., )by

[0137] Equation 3:

[0138] Thus, module 425 may generate corresponding gradients 480 (e.g., ). For example, each antenna pair may have a respective gradient based on a channel component associated with the respective antenna pair. In some cases, since the observations are sparse, the feedback from module 425 (eg, feedback module) may also be sparse.

[0139] In some examples, module 430 may include various steps. For example, module 430 (e.g., an encoder module) may receive inputs including a first set of values ​​of latent variables, channel estimates, and gradients 480 at 440 (e.g., ); At 445, perform input fusion, as described in reference to this article Figure 5 At 450, various attention calculations are performed (e.g., intra-PRG calculations at 450-a, inter-PRG calculations at 450-b, and cross-MIMO calculations at 450-c), as described herein with reference to Figure 6 At 455, a multilayer perceptron (MLP) process is performed; and at 460, a second set of values ​​of the latent variables is output (eg, ). In some cases, the output of each step may be concatenated with the output of the next step. Module 430 may generate a second set of values ​​of the latent variables according to Equation 4 via various steps:

[0140] Equation 4:

[0141] In some cases, module 435 may receive a second set of values ​​for the latent variables and modify channel estimates associated with the plurality of layers based on the second set of values ​​for the latent variables, channel estimates (e.g., channel estimates for a previous iteration), and corresponding gradients. For example, the channel estimates for this iteration 410 may be generated according to Equation 5:

[0142] Equation 5:

[0143] As referenced in this article Figure 5 described.

[0144] In some examples, each iteration 410 of the refinement network 405 can utilize the same set 485 of one or more machine learning parameters. For example, the first iteration 410 can operate according to the set 485 of machine learning parameters. Each of the modules 425, 430, and 435 can operate according to a respective machine learning parameter from the set 485 of machine learning parameters. The remaining iterations 410 of the refinement network 405 can utilize the same set 485 of machine learning parameters to estimate a given attention computation. The set 485 of machine learning parameters can be generated based on training the refinement network 405 according to one or more sets of training parameters. If weights (e.g., θ) are shared during the training operation, the weights (e.g., θ) can be shared according to the training parameter set 485. 1 ,θ 2 ,...θ T ), the same set of machine learning parameters 485 may be generated for each iteration 410. The set of machine learning parameters 485 may be stored on the refinement network 405 and may be utilized during operation of the refinement network 405. In some examples, different refinement operations may be performed for different attention calculations, and all iterations 410 of a given refinement operation may share the corresponding same set of machine learning parameters 485, where the set of machine learning parameters for different refinement operations may be different. Attention calculations may include, for example, intra-PRB calculations, inter-PRB calculations, cross-MIMO calculations, MLP calculations, or any combination thereof. The size of the shared parameter set may be independent of the number of resources and may be based on the number of machine learning model parameters used for a given attention calculation.

[0145] Sharing the same set of machine learning parameters 485 across all iterations 410 can achieve a reduction in the size of the refinement network 405 and a reduction in the complexity associated with the refinement operation, compared to a refinement operation that utilizes different parameters for each iteration 410. The reduction in size and complexity can be achieved while the reliability and accuracy of the refinement operation are relatively unchanged. For example, the number of parameters in the parameter set for iteration 410 of the refinement network can range from 100K to 2M parameters, and thus sharing the same parameter set across iterations 410 can provide substantial benefits in terms of storage space for the parameter set. Therefore, the number of parameters that can be generated when training the refinement network 405 can be less than if there are different parameters for each iteration 410, and the refinement operation can still produce a reliable and accurate output 420 associated with the refined channel estimates and latent variables.

[0146] Figure 5An example of a network 500 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. In some examples, the network 500 can be implemented by aspects of the wireless communication system 100, the network 400, or both. For example, the network 500 can be implemented by the UE 115, the network entity 105, the encoder 430, the decoder 435, or any combination thereof, as described herein with reference to Figure 1 and 4 described.

[0147] In some cases, network 500 can perform fusion associated with a neural network. For example, network 500 can be an example of a convolutional neural network (CNN) (e.g., a neural network that uses convolutions instead of more general matrix multiplications for multiple layers). In some cases, fusion associated with a CNN can fuse (e.g., combine, compress) two or more convolutional layers (e.g., weights associated with each layer) together.

[0148] In some examples, such as those described herein, Figure 4 As described, one or more modules (e.g., module 430, module 435) of a refinement network (e.g., refinement network 405) may utilize network 500. For example, an encoder module (e.g., module 430) may perform a fusion operation (e.g., at 445). The encoder module may generate a second set of values ​​of latent variables associated with a plurality of channel estimates based on combining (e.g., fusing) channel estimates, corresponding gradients (e.g., gradient 480), and corresponding values ​​of a first set of values ​​of latent variables. For example, the encoder module may generate output 530 (e.g., based on Equation 6) ):

[0149] Equation 6: Among them, FuCNN is a fusion operation. represents a first set of values ​​of the latent variable (e.g., a first portion of the estimate 505), represents a channel estimate (eg, the second part of estimate 505), and , represents the corresponding gradient (e.g., gradient 510). For example, the encoder module can fuse the gradient 510 (e.g., from the likelihood module 425) into the hidden state variable (e.g., the first set of values ​​of the latent variable), thereby modeling MIMO multiplexing (e.g., the multiplexed MIMO phenomenon). The fusion operation can act independently on each PRG of the channel (e.g., PRG 305), and incorporate the gradient 510 (e.g., gradient information) into the latent state (e.g., the fusion operation is performed for each PRG of each antenna pair).

[0150] In some cases, the fusion operation may include various steps. For example, the encoder module may receive as input estimates 505 (e.g., channel estimates and latent variable values ​​from a previous iteration 410) and gradients 510 (e.g., gradients 480 from the same iteration 410), as described herein with reference to Figure 4 The encoder module may combine (e.g., fuse, concatenate) the estimate 505 and the gradient 510 to generate a combination 520 (e.g., ). In some cases, combination 520 can be input to a u-net 525 (e.g., a tiny u-net). In some examples, u-net 525 can be an example of a CNN type that utilizes an upsampling operator (e.g., an upsampling operator with a relatively large number of feature channels). u-net 525 can perform various computations (e.g., combinations, operations) associated with a neural network to generate output 530 (e.g., ).

[0151] In some examples, the decoder module (e.g., module 435) may perform a fusion operation. The decoder module may modify the channel estimate (e.g., generate a third iteration of the channel estimate) based on combining (e.g., fusion) the second set of values ​​of the latent variable (e.g., output from step 460), the second set of multiple channel estimates, and the corresponding gradients (e.g., gradient 480). For example, the decoder module may generate output 530 (e.g., based on Equation 7). ):

[0152] Equation 7: Among them, FuCNN is a fusion operation. represents a second set of values ​​of the latent variable (e.g., a first portion of the estimate 505), represents a channel estimate (eg, the second part of estimate 505), and Denotes a corresponding gradient (e.g., gradient 510). For example, a decoder module may receive as input estimates 505 (e.g., channel estimates and latent variable values ​​for the same iteration 410) and gradients 510 (e.g., gradients 480 for the same iteration 410), as described herein with reference to Figure 4 The decoder module may combine (e.g., fuse, concatenate) estimates 505 and gradients 510 to generate a combination 520 (e.g., In some cases, combination 520 may be input to a u-net 525 (e.g., a tiny u-net). u-net 525 may perform various computations (e.g., combinations, operations) associated with a neural network to generate output 530 (e.g., ). The decoder module may utilize information from the likelihood module and the various submodules of the encoder module (steps 440 to 460) to update the latent variables, the channel estimate, or both. The decoder module may act on each PRG independently to utilize the updated latent variables (e.g., the second set of values) along with the gradient information to improve the channel estimate (e.g., to make the channel estimate closer to the actual channel).

[0153] Figure 6 An example of a network 600 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. In some examples, the network 600 can be implemented by aspects of the wireless communication system 100, the network 400, or both. For example, the network 600 can be implemented by the UE 115, the network entity 105, the encoder 430, or any combination thereof, as described herein with reference to Figure 1 and 4 described.

[0154] In some cases, network 600 may perform attention operations (e.g., attention 605) associated with a neural network. For example, attention 605 may include weighting portions of input data (e.g., input 610) differently than other portions of the input data (e.g., enhancing portions of the data while reducing other portions of the data). In some cases, applying attention 605 may aggregate (e.g., modify, align) input data (e.g., observed DMRS symbols) with known data (e.g., known DMRS symbols). In some examples, attention 605 may model interactions (e.g., correlations) within data elements 620. For example, data element 620-a, data element 620-b, data element 620-c, and data element 620-d may interact with data element 620-e, data element 620-f, data element 620-g, and data element 620-h, and vice versa.

[0155] In some cases, an encoder module (e.g., module 430) may perform attention 605 (e.g., self-attention), for example, at various steps (e.g., step 450) of the channel estimation process, as described herein with reference to Figure 4 For example, the encoder module may (e.g., at 450-a) perform an intra-PRG attention operation to model the correlation between resources (e.g., resource elements) of each PRB (e.g., PRB 310) of a PRG (e.g., PRG 305) and other PRBs of the PRG (e.g., PRBs belonging to a single PRG bundle). The encoder module may determine a set of values ​​of a latent variable (e.g., output 530) from the fusion operation, as described herein with reference to Figure 5 The encoder module may flatten each subset of values ​​associated with each PRB of the PRG. For example, a set of values ​​(e.g., ) may include four subsets of values ​​(e.g., ). The encoder module may flatten (e.g., combine, compress to a single frequency row) the four subsets into data element 620-a, data element 620-b, data element 620-c, and data element 620-d (e.g., input 610). In some cases, data element 620-e, data element 620-f, data element 620-g, and data element 620-h may be copies (copies) of data element 620-a, data element 620-b, data element 620-c, and data element 620-d, respectively (e.g., ). In some examples, intra-PRG attention can be used to model long-range dependencies (e.g., PRBs separated across a frequency axis).

[0156] In some cases, the encoder module may (e.g., at 450-b) perform an inter-PRG attention operation to model the correlation between resources (e.g., resource elements) of each PRG (e.g., PRG 305) of the MIMO communication. The encoder module may determine a flattened subset of values ​​associated with each PRB of each PRG (e.g., ), and combining (e.g., concatenating, averaging, mean pulling) the flattened subsets (e.g., embedded subsets) into a single set of values ​​(e.g., ) as data element 620-a, data element 620-b, data element 620-c, and data element 620-d, respectively (e.g., input 610). In some cases, data element 620-e, data element 620-f, data element 620-g, and data element 620-h may be represented by The encoder module may combine the corresponding output of attention 605 (e.g., output 615, residual) with the flattened subset of values ​​(e.g., the subset before averaging) ) are combined (e.g., added) as the output of the inter-PRG attention 615. In some examples, the inter-PRG attention can facilitate information exchange across different PRG bundles.

[0157] In some cases, the encoder module may (e.g., at 450-c) perform a cross-MIMO attention operation to model the correlation between each of the multiple layers associated with the MIMO communication (e.g., the interaction between the antenna pairs per PRB per PRG beam). In some cases, the MIMO communication may be a resource set including the DMRS symbol 320 and the data symbol 315, as described herein with reference to Figure 3 The encoder module may determine the source blocks (eg, ). For example, in a two-dimensional grid, data element 620-a, data element 620-b, data element 620-c, and data element 620-d (eg, input 610 for a cross-MIMO attention operation) may be represented by and In some cases, data element 620-e, data element 620-f, data element 620-g, and data element 620-h may be represented by and Represented. The encoder module may output 615 for cross-MIMO attention. In some examples, cross-MIMO attention may model interactions (e.g., correlations) between different transmit links per PRG per PRB (e.g., between different MIMO links in an equivariant manner). In some examples, the encoder module may utilize the same set of machine learning parameters for all iterations of a given attention calculation. That is, in multiple attention calculations, each separate attention calculation may utilize a corresponding set of machine learning parameters, and the set of machine learning parameters may be different across attention calculations.

[0158] Figure 7 A block diagram 700 of a device 705 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The device 705 may be an example of aspects of a UE 115 or a network entity 105 as described herein. The device 705 may include a receiver 710, a transmitter 715, and a communication manager 720. The device 705 or one or more components of the device 705 (e.g., the receiver 710, the transmitter 715, and the communication manager 720) may include at least one processor (or processing circuit), which may be coupled to at least one memory (or memory circuit) to support or implement the described technology individually or collectively. Each of these components may communicate with each other (e.g., via one or more buses).

[0159] The receiver 710 may provide means for receiving information (such as packets, user data, control information, or any combination thereof) associated with various information channels (e.g., control channels, data channels, information channels associated with a cyclic equivariant inference engine for channel estimation). The information may be communicated to other components of the device 705. The receiver 710 may utilize a single antenna or a collection of multiple antennas.

[0160] The transmitter 715 may provide means for transmitting signals generated by other components of the device 705. For example, the transmitter 715 may transmit information associated with various information channels (e.g., control channels, data channels, information channels associated with a cyclic equivariant inference engine for channel estimation), such as packets, user data, control information, or any combination thereof. In some examples, the transmitter 715 may be co-located with the receiver 710 in a transceiver module. The transmitter 715 may utilize a single antenna or a collection of multiple antennas.

[0161] The communication manager 720, the receiver 710, the transmitter 715, or various combinations thereof or various components thereof may be examples of means for performing various aspects of a cyclic equivariant inference engine for channel estimation as described herein. For example, the communication manager 720, the receiver 710, the transmitter 715, or various combinations thereof or components thereof may be capable of performing one or more of the functions described herein.

[0162] In some examples, the communication manager 720, the receiver 710, the transmitter 715, or various combinations or components thereof may be implemented in hardware (e.g., in a communication management circuit). The circuit may include a processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described in the present disclosure. In some examples, at least one processor (or processing circuit) and at least one memory (or memory circuit) coupled to the at least one processor may be configured to perform one or more of the functions described herein (e.g., instructions stored in at least one memory are executed by one or more processors individually or collectively).

[0163] In another specific implementation, the communication manager 720, the receiver 710, the transmitter 715, or various combinations or components thereof may be implemented in code executed by at least one processor (e.g., as communication management software or firmware). If implemented in code executed by at least one processor, the functionality of the communication manager 720, the receiver 710, the transmitter 715, or various combinations or components thereof may be performed by a general purpose processor, a DSP, a CPU, an ASIC, an FPGA, a microcontroller, or any combination of these or other programmable logic devices (e.g., configured to or otherwise individually or collectively support components for performing the functions described in the present disclosure).

[0164] In some examples, the communication manager 720 can be configured to perform various operations (e.g., receive, obtain, determine, monitor, output, send) using or otherwise cooperating with the receiver 710, the transmitter 715, or both. For example, the communication manager 720 can receive information from the receiver 710, transmit information to the transmitter 715, or be integrated in conjunction with the receiver 710, the transmitter 715, or both to obtain information, output information, or perform various other operations as described herein.

[0165] According to the examples disclosed herein, the communication manager 720 can support wireless communication. For example, the communication manager 720 can be configured to support components for the following aspects: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for a data signal and a second subset of resources allocated for a reference signal. The communication manager 720 can be configured to support components for the following aspects: generating a first set of multiple channel estimates associated with a corresponding layer in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation. The communication manager 720 can be configured to support components for the following aspects: generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​are associated with a corresponding layer in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The communication manager 720 is capable of, configured to, or operable to support components for: performing a refinement operation on the second set of multiple channel estimates comprising one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, to perform each of the one or more iterations, the communication manager 720 may be configured to or otherwise support components for: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generating a second set of values ​​for the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​for the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0166] By including or configuring a communication manager 720 according to examples as described herein, a device 705 (e.g., at least one processor controlling or otherwise coupled to a receiver 710, a transmitter 715, a communication manager 720, or a combination thereof) may support techniques for more accurate channel estimation, more efficient use of communication resources, and reduced memory and computational overhead.

[0167] Figure 8 A block diagram 800 of a device 805 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The device 805 may be an example of aspects of the device 705, UE 115, or network entity 105 as described herein. The device 805 may include a receiver 810, a transmitter 815, and a communication manager 820. The device 805 or one or more components of the device 805 (e.g., the receiver 810, the transmitter 815, and the communication manager 820) may include at least one processor (or processing circuit), which may be coupled to at least one memory (or memory circuit) to support the described techniques. Each of these components may communicate with each other (e.g., via one or more buses).

[0168] The receiver 810 may provide means for receiving information (such as packets, user data, control information, or any combination thereof) associated with various information channels (e.g., control channels, data channels, information channels associated with a cyclic equivariant inference engine for channel estimation). The information may be communicated to other components of the device 805. The receiver 810 may utilize a single antenna or a collection of multiple antennas.

[0169] The transmitter 815 may provide means for transmitting signals generated by other components of the device 805. For example, the transmitter 815 may transmit information associated with various information channels (e.g., control channels, data channels, information channels associated with a cyclic equivariant inference engine for channel estimation), such as packets, user data, control information, or any combination thereof. In some examples, the transmitter 815 may be co-located with the receiver 810 in a transceiver module. The transmitter 815 may utilize a single antenna or a collection of multiple antennas.

[0170] The device 805 or its various components may be examples of components for performing various aspects of a cyclic equivariant inference engine for channel estimation as described herein. For example, the communication manager 820 may include a scheduling component 825, an MMSE component 830, a coarse network component 835, a refined network component 840, or any combination thereof. The communication manager 820 may be an example of various aspects of the communication manager 720 as described herein. In some examples, the communication manager 820 or its various components may be configured to use or otherwise cooperate with the receiver 810, the transmitter 815, or both to perform various operations (e.g., receive, obtain, monitor, output, send). For example, the communication manager 820 may receive information from the receiver 810, transmit information to the transmitter 815, or integrate with the receiver 810, the transmitter 815, or both to obtain information, output information, or perform various other operations as described herein.

[0171] According to the examples disclosed herein, the communication manager 820 can support wireless communication. The scheduling component 825 can be configured to support components for the following aspects: receiving an assignment of a set of resources associated with a channel, the resource set including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. The MMSE component 830 can be configured to support components for the following aspects: generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation. The coarse network component 835 can be configured to support components for the following aspects: generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​are associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The refinement network component 840 is capable of, configured to, or operable to support means for performing a refinement operation on the second set of multiple channel estimates comprising one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, to perform each of the one or more iterations, the likelihood component 845 may be configured to or otherwise support components for generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; the encoder component 850 may be configured to or otherwise support components for generating a second set of values ​​for the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and the decoder component 855 may be configured to or otherwise support components for modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​for the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0172] Additionally or alternatively, the communication manager 820 may support wireless communications according to examples disclosed herein. The scheduling component 825 may be configured to or otherwise support components for receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. The coarse network component 835 may be configured to or otherwise support components for generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources. The refinement network component 840 may be configured to or otherwise support components for performing a refinement operation on the set of multiple channel estimates comprising one or more iterations. In some examples, for each of the one or more iterations, the likelihood component 845 may be configured to or otherwise support components for generating corresponding gradients associated with the set of multiple channel estimates based on the set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; the encoder component 850 may be configured to or otherwise support components for generating a second set of values ​​for the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and the decoder component 855 may be configured to or otherwise support components for modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0173] Fig. 9 A block diagram 900 of a communication manager 920 supporting a cyclic equivariant inference engine for channel estimation is shown in accordance with one or more aspects of the present disclosure. The communication manager 920 may be an example of aspects of the communication manager 720, the communication manager 820, or both as described herein. The communication manager 920 or its various components may be examples of components for performing various aspects of the cyclic equivariant inference engine for channel estimation as described herein. For example, the communication manager 920 may include a scheduling component 925, an MMSE component 930, a coarse network component 935, a refined network component 940, a likelihood component 945, an encoder component 950, a decoder component 955, or any combination thereof. These components, or each of their components or subcomponents (e.g., one or more processors, one or more memories) may communicate directly or indirectly with each other (e.g., via one or more buses), which communication may include communication within a protocol layer of a protocol stack, communication associated with a logical channel of a protocol stack (e.g., between protocol layers of a protocol stack, within a device, component, or virtualized component associated with the network entity 105, between devices, components, or virtualized components associated with the network entity 105), or any combination thereof.

[0174] According to the examples disclosed herein, the communication manager 920 can support wireless communication. The scheduling component 925 can be configured to support components for the following aspects: receiving an assignment of a set of resources associated with a channel, the resource set including a first subset of resources allocated for a data signal and a second subset of resources allocated for a reference signal. The MMSE component 930 can be configured to support components for the following aspects: generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation. The coarse network component 935 can be configured to support components for the following aspects: generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​are associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The refinement network component 940 is capable of, configured to, or operable to support means for performing a refinement operation on the second set of multiple channel estimates comprising one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, to perform each of the one or more iterations, the likelihood component 945 is capable of being configured to be operable to support components for: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; the encoder component 950 is capable of being configured to be operable to support components for: generating a second set of values ​​for the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and the decoder component 955 is capable of being configured to be operable to support components for: modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​for the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0175] In some examples, the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters, and the refinement network component 940 is capable of, configured to, or operable to support components for: performing a second refinement operation on a second set of multiple channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective attention computations in a set of multiple attention computations.

[0176] In some examples, the set of multiple attention computations includes intra-PRB group computations, inter-PRB group computations, cross-MIMO computations, MLP computations, or any combination thereof.

[0177] In some examples, MMSE component 930 is capable of, configured to, or operable to support means for performing the MMSE operation based on a resource configuration pattern of the second subset of resources allocated for the reference signal, the reference signal comprising a DMRS.

[0178] In some examples, to support generating the corresponding gradients, the likelihood component 945 can be configured to or operable to support components for: generating a corresponding set of values ​​of the residual variable based on a difference between the measured observation for the second subset of resources and a second set of multiple channel estimates for the second subset of resources. In some examples, to support generating the corresponding gradients, the likelihood component 945 can be configured to or operable to support components for: combining the corresponding set of values ​​of the residual variable, the second subset of resources, and the number of mask bits.

[0179] In some examples, to support generating a second set of values ​​for the latent variable, the encoder component 950 is capable of, configured to, or operable to support components for combining the second set of multiple channel estimates for the second subset of resources, the corresponding gradients, and corresponding values ​​in the first set of values ​​for the latent variable based on generating the corresponding gradients.

[0180] In some examples, to support generating a second set of values ​​for latent variables, encoder component 950 is capable, configured, or operable to support components for modeling dependencies between resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0181] In some examples, to support generating a second set of values ​​for the latent variables, the encoder component 950 is capable of, configured to, or operable to support components for modeling correlations between resources of each group in a set of multiple resource groups of the resource set and other groups in the set of multiple resource groups, wherein each group in the set of multiple resource groups comprises a set of multiple resource blocks.

[0182] In some examples, to support generating a second set of values ​​for the latent variables, the encoder component 950 is capable, configured, or operable to support means for modeling a correlation between each layer in the set of multiple layers for the set of resources.

[0183] In some examples, to support modifying the second set of multiple channel estimates, the encoder component 950 is capable of, configured to, or operable to support components for combining the second set of values ​​of the latent variable, the second set of multiple channel estimates, and the corresponding gradients based on the set of machine learning parameters.

[0184] In some examples, the nonlinear two-dimensional interpolation of the channel is based on a machine learning model.

[0185] In some examples, the first set of multiple channel estimates and the second set of multiple channel estimates are associated with a set of multiple single-input and single-output antenna pairs.

[0186] In some examples, each of the one or more iterations is performed by a refinement network comprising a likelihood module, an encoder module, and a decoder module, the refinement network comprising a machine learning model. In some examples, each refinement network is performed according to the same set of machine learning parameters.

[0187] According to examples disclosed herein, the communication manager 920 may support wireless communications. The scheduling component 925 may be configured to or otherwise support components for receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. The coarse network component 935 may be configured to or otherwise support components for generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources. The refinement network component 940 may be configured to or otherwise support components for performing a refinement operation on the set of multiple channel estimates comprising one or more iterations. In some examples, for each of the one or more iterations, the likelihood component 945 may be configured to or otherwise support components for generating corresponding gradients associated with the set of multiple channel estimates based on the set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; the encoder component 950 may be configured to or otherwise support components for generating a second set of values ​​for the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and the decoder component 955 may be configured to or otherwise support components for modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0188] In some examples, to support generating the corresponding gradients, the likelihood component 945 may be configured or otherwise support means for generating a corresponding set of values ​​of the residual variable based on a difference between an observation of the measurement for the second subset of resources and a set of multiple channel estimates for the second subset of resources. In some examples, to support generating the corresponding gradients, the likelihood component 945 may be configured or otherwise support means for combining the corresponding set of values ​​of the residual variable, the observation of the measurement for the second subset of resources, and the number of mask bits.

[0189] In some examples, to support generating a second set of values ​​for the latent variable, the encoder component 950 may be configured as or otherwise support components for combining the set of multiple channel estimates for the second subset of resources, the corresponding gradients, and corresponding values ​​from the first set of values ​​for the latent variable based on generating the corresponding gradients.

[0190] In some examples, to support generating a second set of values ​​for latent variables, encoder component 950 may be configured or otherwise support components for modeling correlations between resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0191] In some examples, to support generating a second set of values ​​for the latent variables, encoder component 950 may be configured to or otherwise support components for modeling correlations between resources of each group in a set of multiple resource groups of the resource set and other groups in the set of multiple resource groups, wherein each group in the set of multiple resource groups comprises a set of multiple resource blocks.

[0192] In some examples, to support generating the second set of values ​​for the latent variables, encoder component 950 may be configured or otherwise support means for modeling a correlation between each layer in the set of multiple layers for the set of resources.

[0193] In some examples, to support modifying the set of multiple channel estimates, decoder component 955 may be configured or otherwise support means for combining the set of values ​​of the latent variable, the second set of multiple channel estimates, and the corresponding gradients.

[0194] In some examples, an initial value for the set of multiple channel estimates is associated with a SISO antenna pair.

[0195] In some examples, the second subset of resources is configured according to a resource configuration pattern in a set of resource configuration patterns.

[0196] In some examples, the resource configuration pattern set is a DMRS pattern set.

[0197] In some examples, each iteration is performed by a refinement network that includes a likelihood module, an encoder module, and a decoder module, and each refinement network also includes corresponding parameters associated with the machine learning operation.

[0198] In some examples, the set of resources includes one or more resource groups, and each respective layer in the set of multiple layers is associated with a respective antenna pair in the set of multiple SISO antenna pairs.

[0199] Fig.10A diagram of a system 1000 including a device 1005 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The device 1005 may be an example of a device 705, a device 805, or a UE 115 as described herein, or include components thereof. The device 1005 may communicate with one or more network entities 105, one or more UEs 115, or any combination thereof (e.g., wirelessly). The device 1005 may include components for two-way voice and data communications, including components for sending and receiving communications, such as a communication manager 1020, an input / output (I / O) controller 1010, a transceiver 1015, an antenna 1025, at least one memory 1030 (or memory circuit), a code 1035, and at least one processor 1040 (or processing circuit). These components may be electronically communicated or otherwise coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more buses (e.g., a bus 1045).

[0200] I / O controller 1010 can manage input signals and output signals of device 1005. I / O controller 1010 can also manage peripheral devices that are not integrated into device 1005. In some cases, I / O controller 1010 can represent a physical connection or port to an external peripheral device. In some cases, I / O controller 1010 can utilize an operating system, such as or another known operating system. Additionally or alternatively, I / O controller 1010 may represent or interact with a modem, keyboard, mouse, touch screen, or similar device. In some cases, I / O controller 1010 may be implemented as part of one or more processors (such as at least one processor 1040) (or processing circuitry). In some cases, a user may interact with device 1005 via I / O controller 1010 or via hardware components controlled by I / O controller 1010.

[0201] In some cases, the device 1005 may include a single antenna 1025. However, in some other cases, the device 1005 may have more than one antenna 1025, which may be capable of concurrently transmitting or receiving multiple wireless transmissions. The transceiver 1015 may communicate bidirectionally via one or more antennas 1025, a wired or wireless link, as described herein. For example, the transceiver 1015 may represent a wireless transceiver and may communicate bidirectionally with another wireless transceiver. The transceiver 1015 may also include a modem for: modulating packets; providing the modulated packets to one or more antennas 1025 for transmission; and demodulating packets received from one or more antennas 1025. The transceiver 1015 or the transceiver 1015 and one or more antennas 1025 may be examples of transmitters 715, transmitters 815, receivers 710, receivers 810, or any combination thereof or components thereof as described herein.

[0202] At least one memory 1030 (or memory circuit) may include random access memory (RAM) and read-only memory (ROM). At least one memory 1030 may store computer-readable, computer-executable code 1035 including instructions that, when executed by at least one processor 1040, cause the device 1005 to perform various functions described herein. The code 1035 may be stored in a non-transitory computer-readable medium such as a system memory or another type of memory. In some cases, the code 1035 may not be directly executable by at least one processor 1040, but may cause a computer (e.g., when compiled and executed) to perform the functions described herein. In some cases, among other things, at least one memory 1030 may also include a basic I / O system (BIOS) that may control basic hardware or software operations, such as interaction with peripheral components or devices.

[0203] At least one processor 1040 (or processing circuit) may include an intelligent hardware device (e.g., a general-purpose processor, a DSP, a CPU, a microcontroller, an ASIC, an FPGA, a programmable logic device, a discrete gate or transistor logic component, a discrete hardware component, or any combination thereof). In some cases, at least one processor 1040 may be configured to operate a memory array using a memory controller. In some other cases, the memory controller may be integrated into at least one processor 1040. At least one processor 1040 may be configured to execute computer-readable instructions stored in a memory (e.g., at least one memory 1030) to enable the device 1005 to perform various functions (e.g., various functions or tasks of a loop-equivariant inference engine for channel estimation). For example, the device 1005 or a component of the device 1005 may include at least one processor 1040 (or processing circuit) and at least one memory 1030 (or memory circuit) coupled to or to the at least one processor 1040, and the at least one processor 1040 and the at least one memory 1030 are configured to perform various functions described herein. In some examples, at least one processor 1040 may include multiple processors, and at least one memory 1030 may include multiple memories. One or more of the multiple processors may be coupled to one or more of the multiple memories, which may be configured individually or collectively to perform various functions herein. In some examples, at least one processor 1040 may be a component of a processing system, which may refer to a machine (such as a series of machines) that receives or obtains input and processes the input to produce, generate or obtain an output set, a circuit (including, for example, one or both of a processor circuit (which may include at least one processor 1040) and a memory circuit (which may include at least one memory 1030)) or a system of components. The processing system may be configured to perform one or more of the functions described herein. Therefore, at least one processor 1040 or a processing system including at least one processor 1040 may be configured, configurable, or operable to cause the device 1005 to perform one or more of the functions described herein. Furthermore, as described herein, "configured to," "configurable to," and "operable to" may be used interchangeably and may relate to the ability to perform one or more of the functions described herein when executing code stored in at least one memory 1030 or otherwise.

[0204] According to the examples disclosed herein, the communication manager 1020 can support wireless communication. For example, the communication manager 1020 can be configured to support components for the following aspects: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for a data signal and a second subset of resources allocated for a reference signal. The communication manager 1020 can be configured to support components for the following aspects: generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation. The communication manager 1020 can be configured to support components for the following aspects: generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​are associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The communication manager 1020 is capable of, configured to, or operable to support components for: performing a refinement operation on the second set of multiple channel estimates comprising one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, to perform each of the one or more iterations, the communication manager 1020 may be configured to or otherwise support components for: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generating a second set of values ​​for the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​for the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0205] Additionally or alternatively, the communication manager 1020 may support wireless communications according to examples disclosed herein. For example, the communication manager 1020 may be configured to or otherwise support components for receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. The communication manager 1020 may be configured to or otherwise support components for generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources. The communication manager 1020 may be configured to or otherwise support components for performing a refinement operation on the set of multiple channel estimates comprising one or more iterations. In some examples, for each of the one or more iterations, the communication manager 1020 may be configured to or otherwise support components for: generating corresponding gradients associated with the set of multiple channel estimates based on the set of multiple channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​of the latent variable based on the first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients.

[0206] By including or configuring a communication manager 1020 according to the examples described herein, the device 1005 can support techniques for improving communication reliability, reducing latency, improving user experience associated with reduced processing, reducing power consumption, more efficiently utilizing communication resources, more accurate channel estimation, and reducing memory and computational overhead.

[0207] In some examples, the communication manager 1020 may be configured to perform various operations (e.g., receive, monitor, transmit) using or otherwise cooperating with the transceiver 1015, one or more antennas 1025, or any combination thereof. Although the communication manager 1020 is illustrated as a separate component, in some examples, one or more functions described with reference to the communication manager 1020 may be supported or performed by at least one processor 1040, at least one memory 1030, code 1035, or any combination thereof. For example, the code 1035 may include instructions that are executable by at least one processor 1040 to cause the device 1005 to perform various aspects of the cyclic equivariant inference engine for channel estimation as described herein, or the at least one processor 1040 and the at least one memory 1030 may be otherwise configured to perform or support such operations individually or collectively.

[0208] Fig.11A diagram of a system 1100 including a device 1105 supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The device 1105 may be an example of a device 705, a device 805, or a network entity 105 as described herein, or include components thereof. The device 1105 may communicate with one or more network entities 105, one or more UEs 115, or any combination thereof, which may include communication through one or more wired interfaces, through one or more wireless interfaces, or any combination thereof. The device 1105 may include components that support output and acquisition of communications, such as a communication manager 1120, a transceiver 1110, an antenna 1115, at least one memory 1125 (or memory circuit), a code 1130, and at least one processor 1135 (or processing circuit). These components may be electronically communicated or otherwise coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more buses (e.g., bus 1140).

[0209] The transceiver 1110 may support bidirectional communication via a wired link, a wireless link, or both as described herein. In some examples, the transceiver 1110 may include a wired transceiver and may communicate bidirectionally with another wired transceiver. Additionally or alternatively, in some examples, the transceiver 1110 may include a wireless transceiver and may communicate bidirectionally with another wireless transceiver. In some examples, the device 1105 may include one or more antennas 1115, which may be capable of sending or receiving wireless transmissions (e.g., concurrently). The transceiver 1110 may also include a modem for: modulating a signal; providing the modulated signal for transmission (e.g., via one or more antennas 1115, via a wired transmitter); receiving the modulated signal (e.g., from one or more antennas 1115, from a wired receiver); and demodulating the signal. In some implementations, the transceiver 1110 may include one or more interfaces, such as one or more interfaces coupled to one or more antennas 1115 configured to support various receiving or obtaining operations, or one or more interfaces coupled to one or more antennas 1115 configured to support various transmitting or outputting operations, or a combination thereof. In some implementations, the transceiver 1110 may include or be configured to be coupled to one or more processors or one or more memory components, which are operable to perform or support operations based on received or obtained information or signals, or to generate information or other signals for transmission or other output, or any combination thereof. In some implementations, the transceiver 1110, or the transceiver 1110 and one or more antennas 1115, or the transceiver 1110 and one or more antennas 1115 and one or more processors or one or more memory components (e.g., at least one processor 1135, at least one memory 1125, or both) may be included in a chip or chip assembly installed in the device 1105. In some examples, transceiver 1110 is operable to support communications via one or more communication links (eg, communication link 125 , backhaul communication link 120 , midhaul communication link 162 , fronthaul communication link 168 ).

[0210] At least one memory 1125 may include RAM, ROM, or any combination thereof. At least one memory 1125 may store computer-readable, computer-executable code 1130 including instructions that, when executed by one or more processors in at least one processor 1135, cause the device 1105 to perform various functions described herein. Code 1130 may be stored in a non-transitory computer-readable medium such as a system memory or another type of memory. In some cases, code 1130 may not be directly executable by a processor of at least one processor 1135, but may cause a computer (e.g., when compiled and executed) to perform the functions described herein. In some cases, among other things, at least one memory 1125 may also include a BIOS that may control basic hardware or software operations, such as interactions with peripheral components or devices. In some examples, at least one processor 1135 may include multiple processors, and at least one memory 1125 may include multiple memories. One or more of the multiple processors may be coupled to one or more of the multiple memories, which may be configured individually or collectively to perform various functions herein (e.g., as part of a processing system).

[0211] At least one processor 1135 may include an intelligent hardware device (e.g., a general-purpose processor, a DSP, an ASIC, a CPU, an FPGA, a microcontroller, a programmable logic device, a discrete gate or transistor logic component, a discrete hardware component, or any combination thereof). In some cases, at least one processor 1135 may be configured to operate a memory array using a memory controller. In some other cases, the memory controller may be integrated into one or more processors of at least one processor 1135. At least one processor 1135 may be configured to execute computer-readable instructions stored in a memory (e.g., one or more memories in at least one memory 1125) to enable the device 1105 to perform various functions (e.g., various functions or tasks of a cyclic equivariant inference engine for channel estimation). For example, the device 1105 or a component of the device 1105 may include at least one processor 1135 (or processing circuit) and at least one memory 1125 (or memory circuit) coupled to one or more processors in the at least one processor 1135 or coupled to the processor, and the at least one processor 1135 and the at least one memory 1125 are configured to perform various functions described herein. At least one processor 1135 may be an example of a cloud computing platform (e.g., one or more physical nodes and supporting software such as an operating system, virtual machine, or container example) that may host functions for performing functions of the device 1105 (e.g., by executing code 1130). At least one processor 1135 may be any one or more suitable processors capable of executing scripts or instructions of one or more software programs stored in the device 1105 (such as in one or more memories of the at least one memory 1125). In some examples, the at least one processor 1135 may include multiple processors, and the at least one memory 1125 may include multiple memories. One or more of the multiple processors may be coupled to one or more memories of the multiple memories, which may be configured individually or collectively to perform various functions herein. In some examples, the at least one processor 1135 may be a component of a processing system, which may refer to a machine (such as a series of machines) that receives or obtains input and processes the input to produce, generate, or obtain an output set, a circuit (including, for example, one or both of a processor circuit (which may include at least one processor 1135) and a memory circuit (which may include at least one memory 1125)), or a system of components. The processing system may be configured to perform one or more of the functions described herein. Thus, at least one processor 1135 or a processing system including at least one processor 1135 may be configured, configurable, or operable to cause the device 1105 to perform one or more of the functions described herein.Furthermore, as described herein, "configured to," "configurable to," and "operable to" may be used interchangeably and may relate to the ability to perform one or more of the functions described herein when executing code stored in at least one memory 1125 or otherwise.

[0212] In some examples, bus 1140 may support communications of protocol layers (e.g., within protocol layers) of a protocol stack. In some examples, bus 1140 may support communications associated with logical channels of a protocol stack (e.g., between protocol layers of a protocol stack), which may include communications performed within components of device 1105, or communications performed between different components of device 1105 that may be co-located or may be located in different locations (e.g., where device 1105 may refer to a system in which one or more of communication manager 1120, transceiver 1110, at least one memory 1125, code 1130, and at least one processor 1135 may be located in one of the different components or divided between the different components).

[0213] In some examples, the communication manager 1120 may manage aspects of communications with the core network 130 (e.g., via one or more wired or wireless backhaul links). For example, the communication manager 1120 may manage the delivery of data communications for client devices such as one or more UEs 115. In some examples, the communication manager 1120 may manage communications with other network entities 105 and may include a controller or scheduler for controlling communications with the UE 115 in coordination with the other network entities 105. In some examples, the communication manager 1120 may support an X2 interface within an LTE / LTE-A wireless communication network technology to provide communications between network entities 105.

[0214] According to the examples disclosed herein, the communication manager 1120 can support wireless communication. For example, the communication manager 1120 can be configured to support components for the following aspects: receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for a data signal and a second subset of resources allocated for a reference signal. The communication manager 1120 can be configured to support components for the following aspects: generating a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the resource set from the reference signal received on the second subset of resources according to an MMSE operation. The communication manager 1120 can be configured to support components for the following aspects: generating a second set of multiple channel estimates and a set of multiple values ​​of latent variables according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second set of multiple channel estimates and the set of multiple values ​​are associated with corresponding layers in the set of multiple layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The communication manager 1120 is capable of, configured to, or operable to support components for: performing a refinement operation on the second set of multiple channel estimates comprising one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, to perform each of the one or more iterations, the communication manager 1120 may be configured to or otherwise support components for: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generating a second set of values ​​for the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​for the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0215] By including or configuring a communication manager 1120 according to the examples described herein, the device 1105 may support techniques for improving communication reliability, reducing latency, improving user experience associated with reduced processing, reducing power consumption, more efficiently utilizing communication resources, more accurate channel estimation, and reducing memory and computational overhead.

[0216] In some examples, the communication manager 1120 may be configured to perform various operations (e.g., receive, obtain, monitor, output, transmit) using or otherwise cooperating with the transceiver 1110, one or more antennas 1115 (e.g., where applicable), or any combination thereof. Although the communication manager 1120 is illustrated as a separate component, in some examples, one or more functions described with reference to the communication manager 1120 may be supported or performed by the transceiver 1110, one or more processors in at least one processor 1135, one or more memories in at least one memory 1125, code 1130, or any combination thereof (e.g., by a processing system including at least a portion of at least one processor 1135, at least one memory 1125, code 1130, or any combination thereof). For example, the code 1130 may include instructions that are executable by one or more processors in at least one processor 1135 to cause the device 1105 to perform various aspects of a cyclic equivariant inference engine for channel estimation as described herein, or the at least one processor 1135 and the at least one memory 1125 may be otherwise configured to perform or support such operations individually or collectively.

[0217] Fig.12 A flow chart illustrating a method 1200 for supporting a cyclic equivariant inference engine for channel estimation according to aspects of the present disclosure is shown. The operations of the method 1200 may be implemented by a UE or a network entity or a component thereof as described herein. For example, the operations of the method 1200 may be implemented by a UE or a network entity or a component thereof as described in reference to Figures 1 to 11 The UE 115 or network entity described herein performs. In some examples, the UE or network entity may execute a set of instructions to control the functional elements of the UE or network entity to perform the described functions. Additionally or alternatively, the UE or network entity may use dedicated hardware to perform various aspects of the described functions.

[0218] At 1205, the method may include receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. The operations of block 1205 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1205 may be performed by Fig. 9 The scheduling component 925 is used to execute.

[0219] At 1210, the method may include generating, from the reference signal received on the second subset of resources according to an MMSE operation, a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources. The operations of block 1210 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1210 may be performed as described in reference to Fig. 9The MMSE component 930 is used to perform.

[0220] At 1215, the method may include generating a second set of multiple channel estimates and a set of multiple values ​​of the latent variable based on the nonlinear two-dimensional interpolation of the channel for the set of resources, the second set of multiple channel estimates and the set of multiple values ​​being associated with a corresponding layer in the set of multiple layers for the channel for the set of resources, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The operations of block 1215 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1215 may be performed as described in reference to Fig. 9 The coarse network component 935 is used to perform the above.

[0221] At 1220, the method may include: performing a refinement operation on the second set of multiple channel estimates including one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, each of the one or more iterations may include: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generating a second set of values ​​of the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients. The operations of box 1220 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1220 may be performed by reference to Fig. 9 The refinement network component 940 is used to perform the above.

[0222] Fig.13 A flow chart illustrating a method 1300 for supporting a cyclic equivariant inference engine for channel estimation according to aspects of the present disclosure is shown. The operations of the method 1300 may be implemented by a UE or a network entity or a component thereof as described herein. For example, the operations of the method 1300 may be implemented by a UE or a network entity or a component thereof as described in reference to Figures 1 to 11 The UE 115 or network entity described herein performs. In some examples, the UE or network entity may execute a set of instructions to control the functional elements of the UE or network entity to perform the described functions. Additionally or alternatively, the UE or network entity may use dedicated hardware to perform various aspects of the described functions.

[0223] At 1305, the method may include receiving an assignment of a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. The operations of block 1305 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1305 may be performed by Fig. 9 The scheduling component 925 is used to execute.

[0224] At 1310, the method may include generating, from the reference signal received on the second subset of resources according to an MMSE operation, a first set of multiple channel estimates associated with corresponding layers in a set of multiple layers of the channel for the set of resources. The operations of block 1310 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1310 may be performed as described in reference to Fig. 9 The MMSE component 930 is used to perform.

[0225] At 1315, the method may include generating a second set of multiple channel estimates and a set of multiple values ​​of the latent variable based on the nonlinear two-dimensional interpolation of the channel for the set of resources, the second set of multiple channel estimates and the set of multiple values ​​being associated with a corresponding layer in the set of multiple layers for the channel for the set of resources, wherein the nonlinear two-dimensional interpolation of the channel is based on the first set of multiple channel estimates. The operations of block 1315 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1315 may be performed as described in reference to Fig. 9 The coarse network component 935 is used to perform the above.

[0226] At 1320, the method may include: performing a refinement operation on the second set of multiple channel estimates including one or more iterations, wherein each of the one or more iterations is performed according to the same set of machine learning parameters. In some examples, each of the one or more iterations may include: generating corresponding gradients associated with the second set of multiple channel estimates based on the second set of multiple channel estimates for the second subset of resources and observations of measurements of the second subset of resources; generating a second set of values ​​of the latent variable based on a first set of values ​​in the set of multiple values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the second set of multiple channel estimates, the set of machine learning parameters, and the corresponding gradients. The operations of box 1320 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1320 may be performed as described in reference to Fig. 9 The refinement network component 940 is used to perform the above.

[0227] At 1325, the method may include: performing a second refinement operation on the second set of multiple channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective attention calculations in the set of multiple attention calculations. The operations of block 1325 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of block 1325 may be performed as described in reference to Fig. 9 The refinement network component 940 is used to perform the above.

[0228] Fig.14 A flowchart illustrating a method 1400 for supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The operations of the method 1400 may be implemented by a UE or a component thereof as described herein. For example, the operations of the method 1400 may be implemented by a UE or a component thereof as described in reference to Figures 1 to 13 The UE 115 described herein may be executed. In some examples, the UE may execute a set of instructions to control the functional elements of the UE to perform the described functions. Additionally or alternatively, the UE may use dedicated hardware to perform various aspects of the described functions.

[0229] At 1405, the method may include receiving an assignment to a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. Receiving the assignment may include identifying time-frequency resources on which the assignment is sent, demodulating the transmissions on those time-frequency resources, and decoding the demodulated transmissions to obtain bits indicating the assignment. The assignment may be received via a DCI in a downlink control channel. The operations of 1405 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1405 may be performed by as described in reference Fig. 9 The scheduling component 925 is used to execute.

[0230] At 1410, the method may include generating a set of multiple channel estimates associated with respective layers in a set of multiple layers of channels for the set of resources. Generating the set of multiple channel estimates may include performing the steps described herein with reference to Figure 3 The various interpolation techniques described above are used to calculate estimates (e.g., SISO channel estimates) for each antenna pair per PRG. The operations of 1410 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1410 may be performed as described in reference to Fig. 9 The coarse network component 935 is used to perform the above.

[0231] At 1415, the method may include performing a refinement operation on the set of multiple channel estimates including one or more iterations. Performing the refinement operation may include performing a refinement operation as described herein with reference to Figures 4 to 6The refinement network calculates (e.g., updates, generates) various channel estimates (e.g., MIMO channel estimates) and refines the estimates on various iterations of the machine learning operation. In some examples, each of the one or more iterations may include: generating corresponding gradients associated with the set of multiple channel estimates based on the set of multiple channel estimates for the second resource subset and observations of measurements of the second resource subset; generating a second set of values ​​of the latent variable based on a first set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients; and modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients. The operation of 1415 may be performed according to examples as disclosed herein. In some examples, various aspects of the operation of 1415 may be performed as described in reference to Fig. 9 The refinement network component 940 is used to perform the above.

[0232] Fig.15 A flowchart illustrating a method 1500 for supporting a cyclic equivariant inference engine for channel estimation according to one or more aspects of the present disclosure is shown. The operations of the method 1500 may be implemented by a UE or a component thereof as described herein. For example, the operations of the method 1500 may be implemented by a UE or a component thereof as described in reference to Figures 1 to 14 The described UE 115 is performed. In some examples, the UE may execute a set of instructions to control the functional elements of the UE to perform the described functions. Additionally or alternatively, the UE may use dedicated hardware to perform various aspects of the described functions.

[0233] At 1505, the method may include receiving an assignment to a set of resources associated with a channel, the set of resources including a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals. Receiving the assignment may include identifying time-frequency resources on which the assignment is sent, demodulating the transmissions on those time-frequency resources, and decoding the demodulated transmissions to obtain bits indicating the assignment. The assignment may be received via a DCI in a downlink control channel. The operations of 1505 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1505 may be performed by as described in reference Fig. 9 The scheduling component 925 is used to execute.

[0234] At 1510, the method may include generating a set of multiple channel estimates associated with corresponding layers in a set of multiple layers of channels for the set of resources. Generating the set of multiple channel estimates may include performing the steps described herein with reference to Figure 3 The various interpolation techniques described above are used to calculate estimates (e.g., SISO channel estimates) for each antenna pair per PRG. The operations of 1510 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1510 may be performed as described in reference to Fig. 9 The coarse network component 935 is used to perform the above.

[0235] At 1515, the method may include performing a refinement operation on the set of multiple channel estimates including one or more iterations. Performing the refinement operation may include performing a refinement operation as described herein with reference to Figures 3 to 5 The refinement network calculates (e.g., updates, generates) various channel estimates (e.g., MIMO channel estimates) and refines the estimates on various iterations of the machine learning operation. In some examples, each of the one or more iterations may include: generating corresponding gradients associated with the set of multiple channel estimates based on the set of multiple channel estimates for the second resource subset and the measured observations of the second resource subset; generating a second set of values ​​of the latent variables (e.g., z based at least in part on the first set of values ​​of the latent variables, the multiple channel estimates, the corresponding gradients, and the modeled correlations between the resources of each group in the multiple resource groups of the resource set and the other groups in the multiple resource groups; τ ), wherein each of the plurality of resource groups comprises a plurality of resource blocks; and modifying the set of multiple channel estimates associated with the set of multiple layers based on the second set of values ​​of the latent variable, the set of multiple channel estimates, and the corresponding gradients. The operations of 1515 may be performed according to examples as disclosed herein. In some examples, aspects of the operations of 1515 may be performed by a likelihood module (e.g., module 425), an encoder module (e.g., module 430), or a decoder module (e.g., module 435). In some examples, aspects of the operations of 1515 may be performed by a likelihood module (e.g., module 425), an encoder module (e.g., module 430), or a decoder module (e.g., module 435). Fig. 9 The refinement network component 940 is used to perform the above.

[0236] The following provides an overview of various aspects of the disclosure:

[0237] Aspect 1: An apparatus for performing wireless communications at a wireless communication device, the apparatus comprising: one or more memories; and one or more processors, the one or more processors being coupled to the one or more memories, the one or more memories being configured to execute code that causes the wireless communication device to perform the following operations: receiving an assignment to a resource set associated with a channel, the resource set comprising a first resource subset allocated for a data signal and a second resource subset allocated for a reference signal; generating a first plurality of channel estimates associated with corresponding layers in a plurality of layers of the channel for the resource set from the reference signal received on the second resource subset according to an MMSE operation; generating a second plurality of channel estimates and a plurality of values ​​of a latent variable according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second plurality of channel estimates and the plurality of values ​​being associated with corresponding layers in the plurality of layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is at least Based in part on the first plurality of channel estimates; and performing a refinement operation comprising one or more iterations on the second plurality of channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters, and wherein, in order to perform each of the one or more iterations, the one or more processors are configured to cause the wireless communication device to: generate corresponding gradients associated with the second plurality of channel estimates based at least in part on the second plurality of channel estimates for the second subset of resources and measured observations of the second subset of resources; generate a second set of values ​​of the latent variable based at least in part on a first set of values ​​among the plurality of values ​​of the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients; and modify the second plurality of channel estimates associated with the plurality of layers based at least in part on the second set of values ​​of the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0238] Aspect 2: An apparatus according to Aspect 14, wherein the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters, and wherein the one or more processors are configured to cause the wireless communication device to: perform a second refinement operation on the second plurality of channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with corresponding attention calculations in a plurality of attention calculations.

[0239] Aspect 3: The apparatus according to aspect 15, wherein the plurality of attention calculations include intra-PRB group calculations, inter-PRB group calculations, cross-MIMO calculations, MLP calculations, or any combination thereof.

[0240] Aspect 4: An apparatus according to any one of Aspects 14 to 16, wherein the one or more processors are configured to cause the wireless communication device to: perform the MMSE operation based on a resource configuration pattern of the second subset of resources allocated for the reference signal, wherein the reference signal includes a DMRS.

[0241] Aspect 5: An apparatus according to any one of Aspects 14 to 17, wherein, in order to generate the corresponding gradient, the one or more processors are configured to enable the wireless communication device to: generate a corresponding set of values ​​of residual variables based at least in part on the difference between the measured observations of the second resource subset and the second plurality of channel estimates for the second resource subset; and combine the corresponding set of values ​​of the residual variables, the second resource subset, and the number of mask bits.

[0242] Aspect 6: An apparatus according to any one of Aspects 14 to 18, wherein, in order to generate the second set of values ​​of the latent variable, the one or more processors are configured to cause the wireless communication device to: combine the second plurality of channel estimates for the second subset of resources, the corresponding gradients, and the corresponding values ​​in the first set of values ​​of the latent variable based at least in part on generating the corresponding gradients.

[0243] Aspect 7: An apparatus according to any one of Aspects 14 to 19, wherein, in order to generate the second set of values ​​of the latent variable, the one or more processors are configured to enable the wireless communication device to: model the correlation between resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0244] Aspect 8: An apparatus according to any one of Aspects 14 to 19, wherein, in order to generate the second set of values ​​of the latent variable, the one or more processors are configured to enable the wireless communication device to: model the correlation between resources of each of a plurality of resource groups of the resource set and other groups of the plurality of resource groups, wherein each of the plurality of resource groups includes a plurality of resource blocks.

[0245] Aspect 9: An apparatus according to any one of Aspects 14 to 19, wherein, in order to generate the second set of values ​​of the latent variable, the one or more processors are configured to enable the wireless communication device to: model the correlation between each of the multiple layers of the resource set.

[0246] Aspect 10: An apparatus according to any one of Aspects 14 to 22, wherein, in order to modify the second plurality of channel estimates, the one or more processors are configured to cause the wireless communication device to: combine the second set of values ​​of the latent variables, the second plurality of channel estimates, and the corresponding gradients based at least in part on the set of machine learning parameters.

[0247] Aspect 11: An apparatus according to any one of Aspects 14 to 23, wherein the nonlinear two-dimensional interpolation of the channel is based at least in part on a machine learning model.

[0248] Aspect 12: The apparatus of any one of Aspects 14 to 24, wherein the first plurality of channel estimates and the second plurality of channel estimates are associated with a plurality of SISO antenna pairs.

[0249] Aspect 13: An apparatus according to any one of Aspects 14 to 25, wherein each of the one or more iterations is performed by a refinement network comprising a likelihood module, an encoder module, and a decoder module, the refinement network comprising a machine learning model; and each refinement network is executed according to the same set of machine learning parameters.

[0250] Aspect 14: A method for performing wireless communications at a wireless communication device, the method comprising: receiving an assignment of a resource set associated with a channel, the resource set comprising a first resource subset allocated for a data signal and a second resource subset allocated for a reference signal; generating a first plurality of channel estimates associated with corresponding layers in a plurality of layers of the channel for the resource set from the reference signal received on the second resource subset according to an MMSE operation; generating a second plurality of channel estimates and a plurality of values ​​of a latent variable according to a nonlinear two-dimensional interpolation of the channel for the resource set, the second plurality of channel estimates and the plurality of values ​​being associated with corresponding layers in the plurality of layers of the channel for the resource set, wherein the nonlinear two-dimensional interpolation of the channel is based at least in part on the first plurality of channel estimates; and The channel estimation execution includes one or more iterations of a refinement operation, wherein each of the one or more iterations is performed according to the same set of machine learning parameters, and wherein each of the one or more iterations includes: generating corresponding gradients associated with the second plurality of channel estimates based at least in part on the second plurality of channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​of the latent variable based at least in part on a first set of values ​​among the plurality of values ​​of the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients; and modifying the second plurality of channel estimates associated with the plurality of layers based at least in part on the second set of values ​​of the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients.

[0251] Aspect 15: A method according to Aspect 14, wherein the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters, the method further comprising: performing a second refinement operation on the second plurality of channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with corresponding attention calculations in a plurality of attention calculations.

[0252] Aspect 16: The method according to aspect 15, wherein the multiple attention calculations include intra-PRB group calculations, inter-PRB group calculations, cross-MIMO calculations, MLP calculations, or any combination thereof.

[0253] Aspect 17: The method according to any one of aspects 14 to 16, the method further comprising: performing the MMSE operation based on a resource configuration pattern of the second subset of resources allocated for the reference signal, the reference signal comprising a DMRS.

[0254] Aspect 18: A method according to any one of Aspects 14 to 17, wherein generating the corresponding gradient includes: generating a corresponding set of values ​​of residual variables based at least in part on the difference between the measured observations of the second resource subset and the second plurality of channel estimates for the second resource subset; and combining the corresponding set of values ​​of the residual variables, the second resource subset, and the number of mask bits.

[0255] Aspect 19: A method according to any one of Aspects 14 to 18, wherein generating the second set of values ​​of the latent variable includes: combining the second plurality of channel estimates for the second resource subset, the corresponding gradients, and the corresponding values ​​in the first set of values ​​of the latent variable based at least in part on generating the corresponding gradients.

[0256] Aspect 20: The method according to any one of aspects 14 to 19, wherein generating the second set of values ​​of the latent variable comprises: modeling the correlation between resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0257] Aspect 21: A method according to any one of Aspects 14 to 19, wherein generating the second set of values ​​of the latent variable includes: modeling the correlation between resources of each of multiple resource groups in the resource set and other groups in the multiple resource groups, wherein each of the multiple resource groups includes multiple resource blocks.

[0258] Aspect 22: The method according to any one of aspects 14 to 19, wherein generating the second set of values ​​of the latent variable comprises: modeling the correlation between each of the multiple layers of the resource set.

[0259] Aspect 23: A method according to any one of Aspects 14 to 22, wherein modifying the second plurality of channel estimates comprises: combining the second set of values ​​of the latent variables, the second plurality of channel estimates, and the corresponding gradients based at least in part on the set of machine learning parameters.

[0260] Aspect 24: A method according to any one of Aspects 14 to 23, wherein the nonlinear two-dimensional interpolation of the channel is based at least in part on a machine learning model.

[0261] Aspect 25: The method according to any one of aspects 14 to 24, wherein the first plurality of channel estimates and the second plurality of channel estimates are associated with a plurality of SISO antenna pairs.

[0262] Aspect 26: A method according to any one of Aspects 14 to 25, wherein each of the one or more iterations is performed by a refinement network comprising a likelihood module, an encoder module and a decoder module, the refinement network comprising a machine learning model; and each refinement network is executed according to the same set of machine learning parameters.

[0263] Aspect 27: An apparatus for wireless communication at a wireless communication device, the apparatus comprising: at least one component for performing the method according to any one of aspects 14 to 26.

[0264] Aspect 28: A non-transitory computer-readable medium storing code for wireless communication at a wireless communication device, the code comprising instructions executable by one or more processors to cause the wireless communication device to perform a method according to any one of aspects 14 to 26.

[0265] Aspect 29: A method for wireless communication, the method comprising: receiving an assignment of a resource set associated with a channel, the resource set comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating multiple channel estimates associated with corresponding layers in multiple layers of the channel for the resource set; and performing a refinement operation on the multiple channel estimates comprising one or more iterations, wherein each of the one or more iterations comprises: generating corresponding gradients associated with the multiple channel estimates based at least in part on the multiple channel estimates for the second resource subset and measured observations of the second resource subset; generating a second set of values ​​of the latent variables based at least in part on a first set of values ​​of the latent variables, the multiple channel estimates, and the corresponding gradients; and modifying the multiple channel estimates associated with the multiple layers based at least in part on the second set of values ​​of the latent variables, the multiple channel estimates, and the corresponding gradients.

[0266] Aspect 30: A method according to Aspect 14, wherein generating the corresponding gradient includes: generating a corresponding set of values ​​of residual variables based at least in part on the difference between the measured observations of the second resource subset and the multiple channel estimates for the second resource subset; and combining the corresponding set of values ​​of the residual variables, the measured observations of the second resource subset, and the number of mask bits.

[0267] Aspect 31: A method according to any one of Aspects 14 to 15, wherein generating the second set of values ​​of the latent variable includes: combining the multiple channel estimates for the second subset of resources, the corresponding gradients, and the corresponding values ​​in the first set of values ​​of the latent variable based at least in part on generating the corresponding gradients.

[0268] Aspect 32: The method according to any one of aspects 14 to 16, wherein generating the second set of values ​​of the latent variable comprises: modeling the correlation between the resources of each resource block in a resource block group and other resource blocks in the resource block group.

[0269] Aspect 33: A method according to any one of Aspects 14 to 17, wherein generating the second set of values ​​of the latent variable includes: modeling the correlation between the resources of each of the multiple resource groups of the resource set and other groups in the multiple resource groups, wherein each of the multiple resource groups includes multiple resource blocks.

[0270] Aspect 34: The method according to any one of aspects 14 to 18, wherein generating the second set of values ​​of the latent variable comprises: modeling the correlation between each of the multiple layers of the resource set.

[0271] Aspect 35: The method according to any one of aspects 14 to 19, wherein modifying the multiple channel estimates comprises: combining the second set of values ​​of the latent variables, the multiple channel estimates, and the corresponding gradients.

[0272] Aspect 36: The method according to any one of aspects 14 to 20, wherein the initial values ​​of the plurality of channel estimates are associated with a SISO antenna pair.

[0273] Aspect 37: The method according to any one of Aspects 14 to 21, wherein the second subset of resources is configured according to a resource configuration pattern in a resource configuration pattern set.

[0274] Aspect 38: The method according to aspect 22, wherein the resource configuration pattern set is a demodulation reference signal pattern set.

[0275] Aspect 39: A method according to any one of Aspects 14 to 23, wherein each iteration is performed by a refinement network, the refinement network comprising a likelihood module, an encoder module and a decoder module, and each refinement network also includes corresponding parameters associated with the machine learning operation.

[0276] Aspect 40: The method according to any one of aspects 14 to 24, wherein the set of resources comprises one or more resource groups, and each respective layer of the plurality of layers is associated with a respective antenna pair of a plurality of SISO antenna pairs.

[0277] Aspect 41: A device for wireless communication, the device comprising: one or more memories, the one or more memories storing processor executable code; and one or more processors, the one or more processors being coupled to the one or more memories and capable of operating individually or collectively to execute the code so that the device performs a method according to any one of Aspects 14 to 25.

[0278] Aspect 42: An apparatus for wireless communication, the apparatus comprising at least one component for performing the method according to any one of aspects 14 to 25.

[0279] Aspect 43: A non-transitory computer-readable medium storing a code for wireless communication, the code comprising instructions executable by a processor to perform a method according to any one of aspects 14 to 25.

[0280] It should be noted that the methods described herein describe possible implementations, and that the various operations and steps may be rearranged or otherwise modified and that other implementations are possible. Furthermore, aspects from two or more methods may be combined.

[0281] Although aspects of LTE, LTE-A, LTE-A Pro, or NR systems may be described for example purposes, and LTE, LTE-A, LTE-A Pro, or NR terminology may be used in much of the description, the techniques described herein may also be applicable to networks other than LTE, LTE-A, LTE-A Pro, or NR networks. For example, the described techniques may be applicable to various other wireless communication systems, such as Ultra Mobile Broadband (UMB), Institute of Electrical and Electronics Engineers (IEEE) 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20, Flash-OFDM, and other systems and radio technologies not explicitly mentioned herein.

[0282] The information and signals described herein may be represented using any of a variety of different techniques and methods. For example, data, instructions, commands, information, signals, bits, symbols, and chips mentioned throughout the specification may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or optical particles, or any combination thereof.

[0283] The various illustrative blocks and components described in conjunction with the disclosure herein may be implemented or executed using at least one general purpose processor (or processing circuit), DSP, ASIC, CPU, FPGA or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in an alternative, the processor may be any processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices (e.g., a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in combination with a DSP core, or any other such configuration). Any function or operation described herein as being capable of being performed by a processor may be performed by multiple processors capable of performing the described functions or operations individually or collectively.

[0284] The functions described herein may be implemented using hardware, software executed by a processor, processing circuitry, firmware, or any combination thereof. If implemented using software executed by a processor, the functions may be stored as one or more instructions or codes of a computer-readable medium, or sent using one or more instructions or codes of a computer-readable medium. Other examples and specific implementations are within the scope of this disclosure and the appended claims. For example, due to the nature of software, the functions described herein may be implemented using software executed by a processor, hardware, firmware, hard wiring, or a combination of any of these items. Features that implement the functions may also be physically located at different locations, including being distributed so that the various parts of the functions are implemented at different physical locations.

[0285] Computer-readable medium includes both non-transient computer storage medium and communication medium, and the communication medium includes any medium that promotes the transfer of computer programs from one location to another location.Non-transient storage medium can be any available medium that can be accessed by a general or special computer.By way of example and not limitation, non-transient computer-readable medium can include RAM, ROM, electrically erasable programmable ROM (EEPROM), flash memory, compact disc (CD) ROM or other optical disc storage device, disk storage device or other magnetic storage device, or can be used for carrying or storing desired program code components and any other non-transient medium that can be accessed by a general or special computer or a general or special processor in the form of an instruction or data structure.Moreover, any connection is appropriately referred to as computer-readable medium.For example, if software is sent from a website, a server or other remote source using coaxial cable, optical fiber cable, twisted pair, digital subscriber line (DSL) or wireless technologies such as infrared, radio and microwave, then coaxial cable, optical fiber cable, twisted pair, DSL or wireless technologies such as infrared, radio and microwave are included in the definition of computer-readable medium. As used herein, disks and optical disks include CDs, laser disks, optical disks, digital versatile disks (DVDs), floppy disks, and Blu-ray disks. Disks can reproduce data magnetically, and optical disks can reproduce data optically using lasers. Combinations of the above are also included within the scope of computer-readable media. Any function or operation described herein as being capable of being performed by a memory may be performed by a memory circuit and / or multiple memories that can perform the function or operation individually or collectively.

[0286] As used herein (including in the claims), "or" used in a list of items (e.g., a list of items followed by a phrase such as "at least one of" or "one or more of") indicates an inclusive list, so that, for example, a list of at least one of A, B, or C means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). In addition, as used herein, the phrase "based on" should not be interpreted as a reference to a closed set of conditions. For example, an example step described as "based on condition A" can be based on both condition A and condition B without departing from the scope of the present disclosure. In other words, as used herein, the phrase "based on" should be interpreted in the same manner as the phrase "based at least in part on."

[0287] As used herein, including in the claims, the article "a" before a noun is open ended and is understood to mean "at least one" of those nouns or "one or more" of those nouns. Thus, the terms "a", "at least one", "one or more", "atleast one of one or more" are interchangeable. For example, where a claim lists a "component" that performs one or more functions, each of the individual functions may be performed by a single component or by any combination of multiple components. Thus, the term "component" having a characteristic or performing a function may refer to "at least one of one or more components" having a particular characteristic or performing a particular function. Subsequent references to components introduced using the term "the" or "said" together with the article "a" may refer to any or all of the one or more components. For example, a component introduced with the article "a" may be understood to mean "one or more components", and a subsequent reference in a claim to "the component" may be understood to be equivalent to a reference to "at least one of the one or more components". Similarly, subsequent references to components introduced as "one or more components" using the term "the" or "said" may refer to any or all of the one or more components. For example, subsequent references in a claim to "the one or more components" may be understood as equivalent to reference to "at least one of the one or more components." As used herein, including in the claims, the terms "set" or "subset" may be understood to refer to one or more items. For example, reference to "a set of objects" or "a subset of objects" may be understood to refer to one object or multiple objects equivalently.

[0288] The term "determining" encompasses a variety of actions, and thus, "determining" may include calculating, computing, processing, deriving, investigating, looking up (such as via looking up in a table, database or other data structure), ascertaining, and the like. Furthermore, "determining" may include receiving (e.g., receiving information), accessing (e.g., accessing data stored in a memory), etc. Additionally, "determining" may include parsing, obtaining, selecting, choosing, establishing, and other such similar actions.

[0289] In the drawings, similar components or features may have the same reference label. In addition, various components of the same type may be distinguished by following the reference label with a dash and a second label to distinguish between similar components. If only the first reference label is used in the specification, the description may apply to any of the similar components having the same first reference label, regardless of the second or other subsequent reference labels.

[0290] The descriptions set forth herein in conjunction with the accompanying drawings describe example configurations and do not represent all examples that may be implemented or within the scope of the claims. The term "example" as used herein means "used as an example, instance, or illustration," rather than "preferred" or "advantageous over other examples." The specific implementation includes specific details to provide an understanding of the described techniques. However, these techniques may be practiced without these specific details. In some cases, known structures and devices are shown in block diagram form to avoid obscuring the concepts of the described examples.

[0291] The description herein is provided to enable one of ordinary skill in the art to implement or use the present disclosure. Various modifications to the present disclosure will be apparent to one of ordinary skill in the art, and the general principles defined herein may be applied to other variations without departing from the scope of the present disclosure. Thus, the present disclosure is not limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An apparatus for wireless communication at a wireless communication device, the apparatus comprising: one or more memories; and one or more processors coupled to the one or more memories and configured to cause the wireless communication device to: receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating, from the reference signal received on the second subset of resources according to a least mean square estimation operation, a first plurality of channel estimates associated with respective ones of a plurality of layers of the channel for the set of resources; generating a second plurality of channel estimates and a plurality of values ​​of latent variables based on a nonlinear two-dimensional interpolation of the channels for the set of resources, the second plurality of channel estimates and the plurality of values ​​being associated with respective layers of the plurality of layers for the channels for the set of resources, wherein the nonlinear two-dimensional interpolation of the channels is based at least in part on the first plurality of channel estimates; as well as performing a refinement operation comprising one or more iterations on the second plurality of channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters, and wherein, to perform each of the one or more iterations, the one or more processors are configured to cause the wireless communication device to: generating respective gradients associated with the second plurality of channel estimates based at least in part on the second plurality of channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​for the latent variable based at least in part on the first set of values ​​in the plurality of values ​​for the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients; and The second plurality of channel estimates associated with the plurality of layers are modified based at least in part on the second set of values ​​of the latent variables, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients.

2. The apparatus of claim 1 , wherein the refinement operation is a first refinement operation, and the set of machine learning parameters is a first set of machine learning parameters, and the one or more processors are configured to cause the wireless communication device to: Performing a second refinement operation on the second plurality of channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective ones of the plurality of attention computations.

3. The apparatus according to claim 2, wherein the plurality of attention calculations comprise intra-physical resource block group calculations, inter-physical resource block group calculations, cross-multiple input multiple output calculations, multi-layer perceptron calculations, or any combination thereof.

4. The apparatus of claim 1 , wherein the one or more processors are configured to cause the wireless communication device to: The least mean square estimation operation is performed based on a resource configuration pattern of the second subset of resources allocated for the reference signal, the reference signal comprising a demodulation reference signal.

5. The device according to claim 1, wherein: To generate the corresponding gradient, the one or more processors are configured to cause the wireless communication device to: generating a respective set of values ​​of residual variables based at least in part on differences between the measured observations of the second subset of resources and the second plurality of channel estimates for the second subset of resources; as well as The corresponding set of values ​​of the residual variable, the second subset of resources, and a number of mask bits are combined.

6. The device according to claim 1, wherein: To generate the second set of values ​​for the latent variables, the one or more processors are configured to cause the wireless communication device to: The second plurality of channel estimates for the second subset of resources, the respective gradients, and respective values ​​in the first set of values ​​of the latent variable are combined based at least in part on the generation of the respective gradients.

7. The device according to claim 1, wherein: To generate the second set of values ​​for the latent variables, the one or more processors are configured to cause the wireless communication device to: Dependencies between resources of each resource block in a resource block group and other resource blocks in the resource block group are modeled.

8. The device according to claim 1, wherein: To generate the second set of values ​​for the latent variables, the one or more processors are configured to cause the wireless communication device to: Dependencies between resources of each of a plurality of resource groups of the resource set and other groups of the plurality of resource groups are modeled, wherein each of the plurality of resource groups includes a plurality of resource blocks.

9. The device according to claim 1, wherein: To generate the second set of values ​​for the latent variables, the one or more processors are configured to cause the wireless communication device to: Dependencies between each of the plurality of layers of the collection of resources are modeled.

10. The device according to claim 1, wherein: To modify the second plurality of channel estimates, the one or more processors are configured to cause the wireless communication device to: The second set of values ​​for the latent variables, the second plurality of channel estimates, and the corresponding gradients are combined based at least in part on the set of machine learning parameters.

11. The apparatus of claim 1 , wherein the nonlinear two-dimensional interpolation of the channels is based at least in part on a machine learning model.

12. The apparatus of claim 1, wherein the first plurality of channel estimates and the second plurality of channel estimates are associated with a plurality of single-input and single-output antenna pairs.

13. The device according to claim 1, wherein: Each of the one or more iterations is performed by a refinement network comprising a likelihood module, an encoder module, and a decoder module, the refinement network comprising a machine learning model; and Each refinement network is executed according to the same set of machine learning parameters.

14. A method for wireless communication at a wireless communication device, the method comprising: receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating, from the reference signal received on the second subset of resources according to a least mean square estimation operation, a first plurality of channel estimates associated with respective ones of a plurality of layers of the channel for the set of resources; generating a second plurality of channel estimates and a plurality of values ​​of latent variables based on a nonlinear two-dimensional interpolation of the channels for the set of resources, the second plurality of channel estimates and the plurality of values ​​being associated with respective layers of the plurality of layers for the channels for the set of resources, wherein the nonlinear two-dimensional interpolation of the channels is based at least in part on the first plurality of channel estimates; as well as performing a refinement operation comprising one or more iterations on the second plurality of channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters, and wherein each of the one or more iterations comprises: generating respective gradients associated with the second plurality of channel estimates based at least in part on the second plurality of channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​for the latent variable based at least in part on the first set of values ​​in the plurality of values ​​for the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients; and The second plurality of channel estimates associated with the plurality of layers are modified based at least in part on the second set of values ​​of the latent variables, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients.

15. The method of claim 14, wherein the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters, the method further comprising: Performing a second refinement operation on the second plurality of channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective ones of the plurality of attention computations.

16. The method of claim 15, wherein the plurality of attention calculations comprise intra-physical resource block group calculations, inter-physical resource block group calculations, cross-MIMO calculations, multi-layer perceptron calculations, or any combination thereof.

17. The method according to claim 14, further comprising: The least mean square estimation operation is performed based on a resource configuration pattern of the second subset of resources allocated for the reference signal, the reference signal comprising a demodulation reference signal.

18. The method of claim 14, wherein generating the corresponding gradient comprises: generating a respective set of values ​​of residual variables based at least in part on differences between the measured observations of the second subset of resources and the second plurality of channel estimates for the second subset of resources; as well as The corresponding set of values ​​of the residual variable, the second subset of resources, and a number of mask bits are combined.

19. The method of claim 14, wherein generating the second set of values ​​for the latent variable comprises: The second plurality of channel estimates for the second subset of resources, the respective gradients, and respective values ​​in the first set of values ​​of the latent variable are combined based at least in part on generating the respective gradients.

20. The method of claim 14, wherein generating the second set of values ​​for the latent variable comprises: Dependencies between resources of each resource block in a resource block group and other resource blocks in the resource block group are modeled.

21. The method of claim 14, wherein generating the second set of values ​​for the latent variable comprises: Dependencies between resources of each of a plurality of resource groups of the resource set and other groups of the plurality of resource groups are modeled, wherein each of the plurality of resource groups includes a plurality of resource blocks.

22. The method of claim 14, wherein generating the second set of values ​​for the latent variable comprises: Dependencies between each of the plurality of layers of the collection of resources are modeled.

23. The method of claim 14, wherein modifying the second plurality of channel estimates comprises: The second set of values ​​for the latent variables, the second plurality of channel estimates, and the corresponding gradients are combined based at least in part on the set of machine learning parameters.

24. The method of claim 14, wherein the nonlinear two-dimensional interpolation of the channels is based at least in part on a machine learning model.

25. The method of claim 14, wherein the first plurality of channel estimates and the second plurality of channel estimates are associated with a plurality of single-input and single-output antenna pairs.

26. The method of claim 14, wherein: Each of the one or more iterations is performed by a refinement network comprising a likelihood module, an encoder module, and a decoder module, the refinement network comprising a machine learning model; and Each refinement network is executed according to the same set of machine learning parameters.

27. A wireless communication device for wireless communication, the wireless communication device comprising: means for receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; means for generating, from the reference signal received on the second subset of resources in accordance with a least mean square estimation operation, a first plurality of channel estimates associated with respective ones of a plurality of layers of the channel for the set of resources; means for generating a second plurality of channel estimates and a plurality of values ​​of latent variables based on a nonlinear two-dimensional interpolation of the channels for the set of resources, the second plurality of channel estimates and the plurality of values ​​being associated with respective layers of the plurality of layers for the channels for the set of resources, wherein the nonlinear two-dimensional interpolation of the channels is based at least in part on the first plurality of channel estimates; as well as means for performing a refinement operation comprising one or more iterations on the second plurality of channel estimates, wherein each of the one or more iterations is performed according to a same set of machine learning parameters, and wherein the means for performing each of the one or more iterations comprises: means for generating respective gradients associated with the second plurality of channel estimates based at least in part on the second plurality of channel estimates for the second subset of resources and measured observations of the second subset of resources; means for generating a second set of values ​​for the latent variable based at least in part on the first set of values ​​in the plurality of values ​​for the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients; and means for modifying the second plurality of channel estimates associated with the plurality of layers based at least in part on the second set of values ​​of the latent variables, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients.

28. The wireless communication device of claim 27, wherein the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters, the wireless communication device further comprising: means for performing a second refinement operation on the second plurality of channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective ones of the plurality of attention computations.

29. A non-transitory computer readable medium storing code for wireless communication at a wireless communication device, the code comprising instructions executable by one or more processors to cause the wireless communication device to: receiving an assignment of a set of resources associated with a channel, the set of resources comprising a first subset of resources allocated for data signals and a second subset of resources allocated for reference signals; generating, from the reference signal received on the second subset of resources according to a least mean square estimation operation, a first plurality of channel estimates associated with respective ones of a plurality of layers of the channel for the set of resources; generating a second plurality of channel estimates and a plurality of values ​​of latent variables based on a nonlinear two-dimensional interpolation of the channels for the set of resources, the second plurality of channel estimates and the plurality of values ​​being associated with respective layers of the plurality of layers for the channels for the set of resources, wherein the nonlinear two-dimensional interpolation of the channels is based at least in part on the first plurality of channel estimates; as well as performing a refinement operation comprising one or more iterations on the second plurality of channel estimates, wherein each of the one or more iterations is performed according to the same set of machine learning parameters, and wherein the instructions to perform each of the one or more iterations are executable to: generating respective gradients associated with the second plurality of channel estimates based at least in part on the second plurality of channel estimates for the second subset of resources and measured observations of the second subset of resources; generating a second set of values ​​for the latent variable based at least in part on the first set of values ​​in the plurality of values ​​for the latent variable, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients; as well as The second plurality of channel estimates associated with the plurality of layers are modified based at least in part on the second set of values ​​of the latent variables, the second plurality of channel estimates, the set of machine learning parameters, and the corresponding gradients.

30. The non-transitory computer readable medium of claim 29, wherein the refinement operation is a first refinement operation and the set of machine learning parameters is a first set of machine learning parameters, and the instructions are further executable by the one or more processors to cause the wireless communication device to: Performing a second refinement operation on the second plurality of channel estimates, the second refinement operation comprising one or more second iterations performed according to the same second set of machine learning parameters, wherein the first refinement operation and the second refinement operation are associated with respective ones of the plurality of attention computations.

Images