Methods, apparatus, and user equipment for determining precoding information used in beamforming

Abstract

A method, apparatus, and user equipment are provided for determining precoding information for beamforming. The method may include: determining a target component of the precoding information for a channel; determining at least a portion of the precoding information based on a decompressed component and the target component; and transmitting the at least portion of the precoding information from the user equipment. Determining at least a portion of the precoding information may be based on the correlation between the target component and the decompressed component. At least a portion of the precoding information may include a vector of coefficients. The vector of coefficients may include a vector of complex numbers. At least one of the complex numbers may include an amplitude representing scaling coefficients and a phase representing rotation coefficients. The target component may include a target matrix. The decompressed component may include a decompressed matrix. The target matrix may include a linear combination coefficient matrix. The precoding information may include a precoding matrix indicator.

Description

[0001] This application claims priority to and the benefit of U.S. Provisional Patent Application No. 63 / 301,049, filed January 19, 2022, and U.S. Patent Application No. 17 / 852,358, filed June 28, 2022, which are incorporated by reference. TECHNICAL FIELD

[0002] The present disclosure relates generally to beamforming, and more specifically to systems, methods, and devices for determining precoding information for beamforming. BACKGROUND

[0003] A wireless communication system can implement a beamforming technique in which a receiver can provide precoding information to a transmitter based on channel conditions between the transmitter and the receiver. The transmitter can use the precoding information to perform beamforming for transmission to the receiver.

[0004] The above information disclosed in this Background section is only for enhancing the understanding of the background of the application, and therefore it can contain information that does not constitute prior art. SUMMARY

[0005] A method can include determining a target component of precoding information for a channel, determining at least a portion of the precoding information based on a decompressed component and the target component of the precoding information, and transmitting the at least a portion of the precoding information from a user equipment. The step of determining the at least a portion of the precoding information can be based on a correlation between the target component and the decompressed component. The at least a portion of the precoding information can include a vector of coefficients. The vector of coefficients can include a vector of complex numbers. At least one of the complex numbers can include an amplitude representing a scaling coefficient and a phase representing a rotation coefficient. The target component can include a target matrix, and the decompressed component can include a decompressed matrix. The target matrix can include a linear combination coefficient (LCC) matrix. The precoding information can include a precoding matrix indicator (PMI). The step of determining the at least a portion of the precoding information can include computing the precoding information based on the target component and the decompressed component. The step of computing the at least a portion of the precoding information can include performing an iterative operation. The iterative operation can include a gradient descent operation. The iterative operation can include a heavy ball operation. The method can further include determining a channel condition, and the step of determining the target component of the precoding information can be based on the channel condition. The decompressed component can include a decompressed version of the target component.

[0006] An apparatus can include a receiver configured to receive a reference signal using a channel, a transmitter configured to transmit precoding information for the channel, and a controller configured to determine a target component of the precoding information based on the reference signal and to determine at least a portion of the precoding information based on a decompressed component of the precoding information and the target component. The controller can be configured to determine the at least a portion of the precoding information based on a correlation between the target component and the decompressed component. The at least a portion of the precoding information can include one or more coefficients, the target component can include a target precoding coefficient matrix, and the decompressed component can include a decompressed version of the target component.

[0007] A user equipment can include at least one processor configured to determine a target component of a precoding matrix for a downlink channel based on a condition of the downlink channel, determine a decompressed component of the precoding matrix, determine at least a portion of the precoding matrix based on the target component and the decompressed component, compress the precoding matrix based on the determined at least a portion of the precoding matrix to generate a compressed component, and transmit the compressed component using an uplink channel. The at least one processor can be configured to determine at least a portion of the precoding matrix based on a correlation between the target component and the decompressed component. The at least one processor can be configured to determine the target component based on a coefficient matrix and vectors of scaling coefficients and rotation coefficients. BRIEF DESCRIPTION OF DRAWINGS

[0008] The accompanying drawings are not necessarily to scale, and in general, are for illustrative purposes only. Like reference numerals used in the various drawings designate like structures or functions throughout the various drawings. The drawings are intended to facilitate description of the various embodiments described herein and are not intended to limit the scope of claims. To prevent the drawings from becoming cluttered, not all components, connections, etc. can be shown, and not all components can have reference numbers. However, the pattern of component configuration can be readily apparent from the drawings. The drawings, together with the description, demonstrate example embodiments of the present disclosure and, together with the description, serve to explain principles of the present disclosure.

[0009] Figure 1 Embodiments of a wireless system showing a UE providing channel condition feedback to a base station.

[0010] Figure 2 Embodiments of spatial compression and frequency compression of channel condition feedback in a wireless system.

[0011] Figure 3Embodiments are shown illustrating a scheme for determining precoding information based on a target component of precoding information and a decompressed component of precoding information in accordance with the principles of the present disclosure.

[0012] Figure 4 Example embodiments of a user equipment in accordance with the principles of the present disclosure are shown.

[0013] Figure 5 Example embodiments of a base station in accordance with the principles of the present disclosure are shown.

[0014] Figure 6 Embodiments are shown illustrating a method for determining precoding information in accordance with the principles of the present disclosure. DETAILED DESCRIPTION

[0015] Wireless base stations can use channel state information (CSI) feedback from user equipment (UE) for beamforming in a downlink (DL) channel to the UE. The UE can provide feedback to the base station in the form of precoding information, such as a precoding matrix or a precoding matrix indicator (PMI). However, reporting precoding information to the base station can involve a large amount of uplink (UL) overhead.

[0016] To reduce the UL overhead, the UE can compress the precoding information using feedback compression (e.g., using spatial compression and / or frequency compression) before sending the precoding information to the base station in a UL channel. The base station can then decompress the precoding information and use the precoding information for beamforming in the DL channel. However, compressing the precoding information can reduce the effectiveness of the precoding information.

[0017] Some inventive principles of the present disclosure relate to techniques for determining at least a portion of precoding information based on a target component of precoding information and a corresponding decompressed component of precoding information. For example, in some embodiments, values (e.g., optimized values) of a vector of scaling coefficients and rotation coefficients for the precoding information can be determined based on a correlation between a target linear combination coefficient (LCC) matrix (e.g., an ideal LCC matrix) computed at the UE and a corresponding decompressed LCC matrix at the base station. In some embodiments, the values of the vector of scaling coefficients and rotation coefficients can be optimized, e.g., by maximizing the correlation between the target LCC matrix and the decompressed LCC matrix. In some embodiments, one or more scaling coefficients and / or rotation coefficients can be optimized on a per-subband basis. Depending on implementation details, using one or more techniques for determining precoding information in accordance with the principles of the present disclosure can improve the effectiveness of the precoding information (e.g., block error rate (BLER), signal-to-noise ratio (SNR), etc.) on the DL channel.

[0018] In some embodiments, the correlation between the target component of the precoding information and the respective decompressed component of the precoding information can be expressed in terms of mathematical principles, e.g., as an objective function. The objective function or other mathematical expression can be solved in any suitable manner. For example, if the expression does not readily yield a closed-form solution, an iterative algorithm such as gradient descent (GD) can be used to solve it.

[0019] The present disclosure encompasses a number of inventive principles related to precoding information for beamforming. These principles can have independent utility and can be embodied individually, and not every embodiment can utilize each principle. Moreover, the principles can also be embodied in various combinations, some of which can amplify the benefits of the individual principles in a synergistic manner.

[0020] For illustrative purposes, some embodiments can be described in the context of particular implementation details and / or applications, such as coefficients for precoding information can be determined by a UE and transmitted to a gNodeB (gNB) base station in a Fifth Generation (5G) New Radio (NR) system for beamforming in a DL channel. However, the inventive principles are not limited to these details and / or applications, and can be applied in any other situation in which a first device can provide channel feedback to a second device in the form of precoding information, whether either of the devices can be a base station, a UE, a peer device, etc., and whether the channel can be a UL channel, a DL channel, a peer-to-peer channel, etc. Moreover, the inventive principles can be applied in any type of wireless communication system that can utilize beamforming, such as other types of cellular networks (e.g., 4G LTE, 6G, and / or any future generation of cellular networks), Bluetooth, Wi-Fi, etc.

[0021] In some wireless communication systems, a first device can rely on a second device to provide information about channel conditions to enable the second device to perform beamforming in a channel between the two devices. For example, in a 5G NR system operating in a frequency domain duplex (FDD) mode, the lack of reciprocity between UL and DL channels can cause a gNB to rely on channel state information (CSI) feedback from a UE to determine channel conditions in a DL channel for precoding (e.g., for beamforming).

[0022] Figure 1 Embodiments of a wireless system in which a UE provides channel condition feedback to a base station are shown. Figure 1The embodiments shown in the middle can include a base station (e.g., gNB) 102 and a UE 104 connected through a DL channel 106 and a UL channel 108. The base station 102 can transmit one or more reference signals 110 to the UE 104 in the DL channel 106. The UE 104 can determine the DL channel conditions by measuring how the DL channel 106 affects the reference signals 110, and then transmit CSI feedback 112 to the base station 102 in the UL channel 108. The base station 102 can then use the CSI feedback 112 to implement a DL precoding scheme for beamforming in the DL channel 106 (e.g., to reduce interference between users).

[0023] In 5G NR systems, Type-I codebooks can be used to provide implicit CSI feedback in the form of indices to a base station, where the indices can point to a pre-defined PMI selected by a UE based on DL channel conditions. Implicit CSI feedback can provide sufficient performance, for example, in embodiments with a limited number of antenna ports and / or users. However, in systems with a larger number of antenna ports and / or users (e.g., multiple-input multiple-output (MIMO) systems), the relatively low resolution of Type-I codebooks can not provide CSI feedback with sufficient accuracy.

[0024] In 5G NR systems, Type-II codebooks can be used to provide explicit CSI feedback, where a UE can determine one or more principal eigenvectors or singular vectors per subband based on DL channel conditions. The UE can then use the principal eigenvectors or singular vectors to derive a PMI, where the PMI can be fed back to a gNB, which can use the PMI for beamforming in the DL channel. However, in some embodiments, and depending on implementation details, reporting Type-II CSI feedback information to a gNB can involve a large amount of overhead (e.g., sending a large amount of data over the UL channel).

[0025] In some embodiments of wireless systems, one or more types of compression can be applied to channel condition information to reduce the overhead associated with providing channel condition feedback to a base station. For example, 5G NR systems starting from Release 15 (Rel 15) can implement spatial compression, where a singular vector matrix E (which can have columns corresponding to dominant singular vectors for a channel for N3subbands (SBs)) can be compressed in the spatial domain (SD) by performing linear combinations (LCs) in the spatial domain.

[0026] As another example, to provide further compression, 5G NR systems starting from Release 16 (Rel 16) can implement eType-II codebooks that can use joint spatial and frequency compression, where dominant singular vectors for one or more subbands (SBs) can be jointly compressed in the spatial domain (SD) and the frequency domain (FD).

[0027] In some embodiments, one or more additional techniques can be used to reduce the amount of data sent to the base station to report channel conditions. For example, quantization can be used to reduce the number of bits used to represent a numerical value. As another example, coefficient pruning can treat relatively small coefficient values as zero, and thus not report those values to the base station. In some embodiments, one or more additional techniques used to reduce the amount of data sent to the base station to report channel conditions can be considered to be in the form of compression.

[0028] In some embodiments, an eType-II codebook configuration can be used to select a precoding matrix indicator (PMI) as follows:

[0029] W = W1W2 (1)

[0030] where W can represent a precoding matrix (e.g., for a given layer), W1may represent a beam selection matrix such that

[0031]

[0032] and W2may represent a linear combination coefficient (LCC) matrix such that

[0033]

[0034] where N T may indicate a number of transmit (Tx) ports, L can indicate a number of beams, N3may indicate a number of PMI subbands, and the notation may indicate a matrix with dimensions a and b.

[0035] With the eType-II codebook configuration, a precoding matrix can be determined, for example, using a three-part process. During the first part of the process, the UE can determine a beam selection matrix W1, for example, by selecting one or more beams based on a correlation to a channel matrix H or a principal singular vector of a channel matrix H (e.g., a singular vector matrix E), such as selecting L beams from a grid of N1N2possible beams, where N1may represent a number of horizontal beams and N2may represent a number of vertical beams. For example, in some embodiments, the one or more beams of W1may be selected by maximizing the correlation.

[0036] During the second part of the three-part process, the LCC matrix W2may be determined. In some embodiments, the target LCC matrix V is such that

[0037]

[0038] V can be computed, e.g., using a singular value decomposition (SVD) of W1. However, other techniques for determining W1 and / or V can be used. In some embodiments, the target LCC matrix V can represent an ideal LCC matrix, an optimized LCC matrix, a desired LCC matrix, etc.

[0039] In embodiments where uplink overhead is not a concern, the target LCC matrix V can be used as W2, and W1 and W2 can be reported to a base station (e.g., a gNB) that can use the precoding matrix W = W1W2 to beamform in a DL channel.

[0040] Thus, a first part of the process for determining a precoding matrix W can involve selecting L beams from a grid of N1N2 possible beams (e.g., determining a beam selection matrix W1), and a second part of the process can involve determining how to combine the selected beams (e.g., by determining an LCC matrix W2). In some embodiments, these first two parts can be utilized to implement a Type-II codebook configuration.

[0041] However, in some embodiments, reporting an entire LCC matrix to a base station as channel condition feedback can involve sending a number of coefficients determined by 2 x L x N3 times the number of layers. Depending on implementation details, this can result in an unacceptable amount of overhead.

[0042] Thus, in some embodiments, a process for implementing an eType-II codebook configuration can include a third part, in which the LCC matrix W2 can be subjected to frequency domain (FD) compression to reduce the amount of data reported to a base station as channel condition feedback. For example, in some embodiments, FD compression can involve determining M < N3 DFT bases, where M can indicate the number of compressed discrete Fourier transform (DFT) components.

[0043] In some embodiments, the third part of the process for implementing an eType-II codebook configuration can further reduce the amount of data reported to a base station by using coefficient pruning, in which coefficients with relatively small values can be set to zero and not reported to the base station. For example, a UE can report up to K0< 2LM non-zero coefficients. In some embodiments, one or more values in the LCC matrix W2 can be quantized before being sent to the base station.

[0044] In some embodiments, for convenience, the various techniques for reducing the amount of data reported to a base station as channel condition feedback (e.g., SD compression, FD compression, coefficient pruning, quantization, etc.) can be collectively referred to as feedback compression, or simply compression. In some embodiments, for convenience, PMI and / or precoding information can also refer to any type of precoding information used for beamforming in any type of wireless communication system, such as cellular, Wi-Fi, Bluetooth, etc.

[0045] In some embodiments, an eType-II codebook configuration can exploit frequency domain correlation in W2 to enable feedback compression. For example, in a system with 2L spatial domain (SD) beams and N3subbands, a frequency compression matrix W f (e.g., selected from columns of an oversampled DFT codebook) of size N3x M f to form an orthogonal subset of the basis set of the codebook) can be applied to the LCC matrix W2 f (e.g., at each layer) to obtain a matrix of linear combination coefficients

[0046]

[0047] which can be rewritten as

[0048]

[0049] where H can represent a conjugate transpose (also known as the Hermitian operator).

[0050] In some embodiments, because the frequency compression matrix W f may have a size of N3x M and the LCC matrix W2may have a size of 2Lx N3, applying DFT compression to W2may result in selecting M < N3frequency domain (FD) basis vectors after compression. In some embodiments, FD compression can be applied to each layer to obtain a matrix of linear combination coefficients In some embodiments, one or more elements within the matrix of linear combination coefficients may be referred to as FD coefficients.

[0051] Thus, in some embodiments, the final precoder W obtained with the eType-II codebook configuration can be written as

[0052]

[0053] where, in some embodiments, W f may be obtained by taking M columns from a DFT matrix F, and thus W f may be described as selecting M columns from N3columns of the full DFT matrix F.

[0054] In some embodiments, and depending on implementation details, feedback compression can reduce UL overhead by reducing the amount of channel condition feedback data sent to the base station. However, compressing the feedback data can reduce the accuracy, effectiveness, etc. of the feedback information, and thus the accuracy, effectiveness, etc. of the PMI, beamforming, etc. obtained on the DL channel.

[0055] Figure 2 Embodiments are shown that illustrate spatial compression and frequency compression of channel condition feedback in a wireless system. In Figure 2 In the embodiment shown in FIG. 2, one or more columns of the singular vector matrix E (shown as element 220) can be spatially compressed by performing linear combinations using an SD-based matrix W1 (shown as element 222). In the resulting spatially compressed matrix W2 (shown as element 224), the linear combination coefficients can be correlated.

[0056] The resulting spatially compressed matrix W2 can then be frequency compressed using a frequency compression matrix W f (shown as element 226) to generate a spatially and frequency compressed matrix W (shown as element 228) such that

[0057]

[0058] Some inventive principles of the present disclosure relate to techniques for determining precoding information based on a target component of precoding information and a corresponding decompressed component of precoding information. In some embodiments, this can be understood with reference to equation (9), where, once the UE computes the target LCC matrix V (for a given layer), the decompressed LCC matrix W2 at the base station (e.g., gNB) end can be written for each row k e {1,.., 2L} as:

[0059] W2(k, :) = V(k, :) UFS k F H (9)

[0060] where

[0061]

[0062] The kth row of the target LCC matrix V can be denoted as

[0063] U = diag(u) (11)

[0064] A diagonal matrix U can be denoted as

[0065]

[0066] A vector of per-subband rotation and / or scaling coefficients can be denoted as

[0067]

[0068] A DFT matrix F can be denoted as

[0069] S k = diag(s k ) (14)

[0070] may represent a diagonal matrix, where

[0071]

[0072] may represent a binary vector of zeros and ones. In some embodiments, one or more of the N3coefficients (e.g., relatively small coefficients) can be set to zero to achieve coefficient reduction. Thus, in some embodiments, S k may contain < M ones.

[0073] In some embodiments, the process characterized by equation (9) can be applied on a per layer basis, and thus, the result can be multiplied by the number of layers.

[0074] In some embodiments, equation (9) can be characterized as describing the overall process for compressing and / or selecting precoding information at the UE, sending the precoding information to the base station, and decompressing the precoding information at the base station. For example, in some embodiments, the target LCC matrix V can represent an ideal or optimized LCC matrix.

[0075] To send row k of the target LCC matrix V to the base station, row k of the target LCC matrix V can be multiplied by a diagonal matrix U to provide one or more rotation and / or scaling coefficients per subband, and then multiplied by a DFT matrix F to perform FD compression and / or domain conversion (e.g., to the frequency or compressed domain on the subbands). In the converted domain, the compressed result can be multiplied by a diagonal matrix S k , where the diagonal matrix S k may perform a selection function, e.g., for coefficient reduction. In some embodiments, some or all of the values generated during the overall process can be quantized.

[0076] The resulting row of the target LCC matrix V that has been processed with FD compression and / or coefficient reduction (V(k, :) UFS k ) can then be sent to the base station, e.g., using an indicator or other technique that can include less than the entire vector of coefficients, the base station can perform decompression, e.g., by multiplying by the inverse (e.g., Hermitian) DFT matrix F H to generate a decompressed LCC matrix W2at the base station.

[0077] In some embodiments, the compression, coefficient reduction, quantization, decompression, etc. described above with respect to equation (9) can result in a suboptimal precoding matrix, which can result in lower data rates on the DL channel.

[0078] In some embodiments, one or more techniques based on the principles of this disclosure can determine the precoding information based on the target component of the precoding information and the corresponding decompression component of the precoding information. For example, in some embodiments, the UE can adjust one or more coefficients in the vector u of each subband rotation coefficient and / or scaling coefficient in equation (12) to make one or more aspects of the decompressed LCC matrix W2 at the base station closer to the target LCC matrix V in equation (9). In some embodiments, this can be achieved, for example, by optimizing the vector u to maximize the correlation between the target LCC matrix V (e.g., the ideal matrix) and the normalized matrix W2 at the base station (e.g., the decompressed precoding matrix after power normalization applied by the gNB).

[0079] Figure 3 An embodiment of a scheme for determining precoded information based on a target component and a decompressed component of the precoded information, according to the principles of this disclosure, is shown. Figure 3 In the illustrated embodiments, UE 334 may determine a target component 338 of precoding information. For example, in some embodiments, UE 334 may calculate a target LCC matrix and / or PMI to be used by base station 336 for beamforming in the DL channel to UE 334. In operation 340, UE 334 may determine precoding information 341 based on the target component 338 and the calculated decompressed component 350′ (e.g., using one or more parameters such as gradient descent parameters that may be used to determine the precoding information), wherein the calculated decompressed component may be a locally computed version of the decompressed component 350 that may be applied by base station 336. For example, precoding information 341 may include a vector of coefficients, wherein the vector of coefficients includes a vector of complex numbers. At least one of the complex numbers may include an amplitude representing a scaling factor and a phase representing a rotation factor. For example, in some embodiments, UE 334 may determine one or more coefficients (e.g., one or more optimized rotation factors and / or scaling factors) based on the correlation between the target LCC matrix and a decompressed version of the target LCC matrix that may be generated by base station 336.

[0080] At operation 342, the UE 334 can compress one or more components of the precoding information, and at operation 344, the UE 334 can transmit the compressed precoding information 352 to the base station 336. For example, in some embodiments, the compression operation 342 can perform spatial domain (SD) and / or frequency domain (FD) compression, coefficient reduction, quantization, etc. on the LCC matrix of the PMI. In some embodiments, the FD compression can be performed by multiplication, where the multiplication can be a frequency compression matrix. In some embodiments, the transmission operation 344 can transmit the compressed precoding information to the base station 336 over a UL channel. At operation 346, the base station 336 can receive the compressed precoding information. At operation 348, the base station 336 can perform a decompression operation on the received precoding information to generate decompressed components 350 of the precoding information. For example, in some embodiments, the decompression operation 346 can be performed by multiplying the received compressed precoding information by an inverse (e.g., Hermitian) frequency compression matrix.

[0081] For illustrative purposes, Figure 3 The embodiments shown in FIG. 3 illustrate various operations being performed by various particular components in a particular order. However, the illustrated components and / or operations are merely exemplary. Some embodiments can involve various additional components and / or operations not depicted, and some embodiments can omit some components and / or operations. Further, the order of the components’ arrangement and / or the timing of the operations can be altered in some embodiments. Although some components can be shown as separate components, some components shown separately can be integrated into a single component in some embodiments, and some components shown as a single component can be implemented with multiple components in some embodiments.

[0082] In some embodiments, the UE 334 can utilize one or more known aspects of spatial compression and / or frequency compression, coefficient reduction, quantization, etc. that can be applied to the target components 338 at operation 342 to make one or more aspects of the decompressed components 350 at the base station more similar to the target components 338. However, in other embodiments, the UE 334 can implement one or more learning algorithms (e.g., machine learning, heuristics, etc.) to reduce the variation between the decompressed components 350 at the base station and the target components 338 at the UE.

[0083] In some embodiments, the process of compression at the UE (which can include quantization, coefficient selection and / or reduction, etc.) and subsequent decompression at the base station can be characterized as distorting the precoding information (e.g., distorting the PMI or components of the precoding information). Thus, in some embodiments, determining the precoding information based on the target components of the precoding information and the decompressed components of the precoding can be characterized as pre-distorting or compensating the precoding information or components of the precoding information in a manner that can counteract or cancel the distortion caused by the compression process and / or the decompression process.

[0084] In some embodiments, one or more mathematical principle techniques can be used to determine the precoding information based on the target component of the precoding information and the decompressed component of the precoding information. For example, an objective function can be used to adjust (e.g., optimize) the vector of rotation coefficients and / or scaling coefficients for each subband to increase (e.g., maximize) the correlation between the target LCC matrix V and the decompressed LCC matrix W2 (e.g., after power normalization at the base station) as follows:

[0085]

[0086] which can be rewritten as

[0087]

[0088] wherein, may refer to a computed value (e.g., a value that can be optimized, solved, estimated, estimated, etc.) of the vector u of rotation coefficients and / or scaling coefficients for each subband,

[0089]

[0090]

[0091]

[0092] such that

[0093] W2(:, i) = Q i u (21)

[0094] and

[0095]

[0096] In some embodiments, one or more iterative algorithms such as Gradient Descent (GD) and / or Heavy Ball (HB) variation of GD can be used to solve, estimate, estimate, optimize, etc. equation (16) and / or equation (17). (The one or more iterative algorithms can be used, for example, in embodiments where equation (16) and / or equation (17) can not have a closed-form solution and / or can be non-convex.)

[0097] In some embodiments (e.g., in equations (9) through (11)), u can be a part of the precoding information, and the precoding information can refer to a larger set (e.g., the entire set) of PMI coefficients that can be reported by the UE to the base station. Thus, in some embodiments, u can be a subset of the precoding information.

[0098] In some embodiments, to apply one or more iterative methods such as GD and / or HB, the update rule at the n-th iteration can be as follows:

[0099]

[0100] where γ n may represent a step size, β n may represent a HB coefficient, and may represent a starting point of the iterative method. In some embodiments where β n may be zero, equation (23) can implement a simple GD.

[0101] Depending on implementation details, e.g., step size, HB coefficient, etc., the selection of the starting value can or can not have a significant impact on whether equations (16) and / or (17) converge to a solution (e.g., can be optimal, near-optimal, approximate, etc.), a speed of convergence, etc.

[0102] In some embodiments, the starting point may be implemented with a heuristic value, such as, for example, rotating the phase of the first beam and / or polarization as follows:

[0103]

[0104] where the angle operator ∠may indicate a phase rotation. Alternatively or additionally, the starting point may be implemented with another heuristic value, such as, for example, rotating the phase of the average over one or more beams and / or polarizations as follows:

[0105]

[0106] In some embodiments, d i (u)may be implemented as a derivative of G i (u), which can be expressed in closed form as follows:

[0107]

[0108] Depending on implementation details, it can be set to zero to determine a local minimum and / or maximum.

[0109] Within the framework of the iterative algorithm as set forth herein, one or more of the following parameters can be adjusted: (1) the number of iterations; (2) one or more values of the step size γ n (which can be fixed and / or variable); (3) the HB coefficient β none or more values (e.g., for GD and / or HB); (4) one or more normalized or non- normalized gradients; and / or the like. The principles of the application are not limited to any particular values of the parameters disclosed herein. However, for purposes of illustration, the following example using the following parameters can provide a relatively low complexity implementation with significant improvement in BLER and / or SNR: 5 iterations; fixed step size γ_n = 1; simple GD (i.e., β n = 0); and non-normalized gradients.

[0110] In some embodiments, one or more parameters can be adjusted by the UE 334 in real-time during operation of the UE 334, but in other embodiments, one or more parameters can be adjusted offline and / or by any other device, such as the base station 336.

[0111] Figure 4 An example embodiment of a user equipment (UE) according to the present disclosure is shown. Figure 4 The illustrated embodiment 400 can include a radio transceiver 402 and a controller 404, where the controller 404 can control operation of the transceiver 402 and / or any other components in the UE 400. The UE 400 can be used, for example, to implement any of the functionality described in the present disclosure, including determining DL channel conditions based on one or more reference signals from a base station, determining one or more parameters of precoding information (e.g., a vector of optimized rotation and / or scaling coefficients) based on a target component (e.g., a target LCC matrix) and a decompressed component (e.g., a decompressed LLC matrix at the base station), compression, quantization, coefficient reduction, etc.

[0112] The transceiver 402 can transmit one or more signals to / from a base station and can include an interface unit for such transmission / reception. For example, the transceiver 402 can receive one or more signals from a base station and / or can transmit precoding information, e.g., one or more PMI components, to a base station on an UL channel.

[0113] The controller 404 can include, for example, one or more processors 406 and a memory 408, where the memory 408 can store instructions for execution by the one or more processors 406 to implement any of the functionality described in the present disclosure. For example, the controller 404 can be configured to implement functionality for determining DL channel conditions based on one or more reference signals from a base station, determining one or more parameters of precoding information (e.g., a vector of optimized rotation and / or scaling coefficients) based on a target component (e.g., a target LCC matrix) and a decompressed component (e.g., a decompressed LLC matrix at the base station), compression, quantization, coefficient reduction, etc.

[0114] Figure 5An example embodiment of a base station according to the present disclosure is shown. Figure 5 The illustrated embodiment 500 can include a radio transceiver 502 and a controller 504, where the controller 504 can control operation of the transceiver 502 and / or any other components in the base station 500. The base station 500 can be used, for example, to implement any of the functionality described in the present disclosure, including transmitting one or more reference signals to a UE on a DL channel, receiving precoding information from a UE on a UL channel, decompressing the precoding information, etc.

[0115] The transceiver 502 can transmit one or more signals to / from a user equipment and can include an interface unit for such transmission / reception. For example, the transceiver 502 can transmit one or more reference signals to a UE on a DL channel and / or receive precoding information from a UE on a UL channel.

[0116] The controller 504 can include, for example, one or more processors 506 and memory 508, which can store instructions for execution by the one or more processors 506 to implement any of the base station functionality described in the present disclosure. For example, the controller 504 can be used to implement functionality for decompressing precoding information, etc.

[0117] In Figure 4 and Figure 5In the illustrated embodiment, transceiver 402 and transceiver 502 can implement receiving and / or transmitting RF signals using various components such as amplifiers, filters, modulators and / or demodulators, A / D and / or DA converters, antennas, switches, phase shifters, detectors, couplers, conductors, transmission lines, etc. Controller 404 and / or controller 504 can be implemented using hardware, software, and / or any combination thereof. For example, all or part of a hardware implementation can include combinational logic, sequential logic, timers, counters, registers, gate arrays, amplifiers, synthesizers, multiplexers, modulators, demodulators, filters, vector processors, complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), systems on a chip (SOCs), state machines, data converters such as ADCs and DACs, etc. All or part of a software implementation can include one or more processor cores, memory, programs, and / or data stores, etc., which can be local and / or remote, and which can be programmed to execute instructions to perform one or more functions of a controller. Some embodiments can include one or more processors such as microcontrollers, CPUs (such as complex instruction set computer (CISC) processors such as x86 processors and / or reduced instruction set computer (RISC) processors such as ARM processors), etc., executing instructions stored in any type of memory, graphics processors (GPUs), neural processing units (NPUs), tensor processing units (TPUs), etc.

[0118] Figure 6 An embodiment of a method for determining precoding information according to the present disclosure is shown. The method can begin at operation 602. At operation 604, the method can determine a target component of precoding information for a channel. At operation 606, the method can determine at least a portion of the precoding information based on a decompressed component and the target component of the precoding information. At operation 608, the method can transmit the at least a portion of the precoding information from a user equipment. The method can end at operation 610.

[0119] In Figure 6 In the illustrated embodiment, as well as any embodiment disclosed herein, the illustrated components and / or operations are merely exemplary. Some embodiments can involve various additional components and / or operations not shown. And some embodiments can omit some of the components and / or operations. Further, the order the components are arranged in, and / or the order in which operations are performed, can be changed in some embodiments. Although some components can be shown as separate components, in some embodiments, some components shown separately can be integrated into a single component, and / or some components shown as a single component can be implemented with multiple components.

[0120] The embodiments disclosed herein may be described in the context of various implementation details, but the principles of this disclosure are not limited to these or any other specific details. Some functions have been described as being implemented by certain components, but in other embodiments, functions may be distributed among different systems and components in different locations. References to components or elements may refer only to a portion of a component or element. Unless otherwise apparent from the context, the use of terms such as “first” and “second” in this disclosure and claims is for the purpose of distinguishing what they modify and may not indicate any spatial or temporal order. A reference to a first thing may not imply the presence of a second thing. Furthermore, the various details and embodiments described above may be combined to produce additional embodiments based on the inventive principles disclosed herein. For convenience, various organizational aids such as chapter headings may be provided, but the subject matter arranged according to these aids and the principles of this disclosure are not defined or limited by these organizational aids.

[0121] Since the inventive principles disclosed in this patent can be modified in terms of arrangement and details without departing from the inventive concept, such changes and modifications are considered to fall within the scope of the appended claims.

Claims

1. A method performed by a user equipment for determining precoding information, comprising: determining a target component of the precoding information for a downlink channel; determining at least a portion of the precoding information based on a correlation between a decompressed component of the precoding information and the target component; and transmitting the at least a portion of the precoding information; wherein the at least a portion of the precoding information provides a beamforming performance for the channel. The at least a portion of the precoding information comprises a vector of coefficients.

2. The method of claim 1, wherein, The vector of coefficients comprises a vector of complex numbers.

3. The method of claim 2, wherein, At least one of the complex numbers comprises:

4. The method of claim 3, wherein, an amplitude representing a scaling coefficient; and a phase representing a rotation coefficient.

5. The method of claim 1, wherein: the target component comprises a target matrix; and the decompressed component comprises a decompressed matrix. The target matrix comprises a linear combination coefficient (LCC) matrix.

6. The method of claim 5, wherein, The precoding information comprises a precoding matrix indicator (PMI).

7. The method of claim 1, wherein, The step of determining the at least a portion of the precoding information comprises computing the at least a portion of the precoding information based on the decompressed component and the target component.

8. The method of claim 1, wherein, The step of computing the at least a portion of the precoding information comprises performing an iterative operation.

9. The method of claim 8, wherein, The iterative operation comprises a gradient descent operation.

10. The method of claim 9, wherein, The iterative operation comprises a heavy ball operation.

11. The method of claim 10, wherein, 12. The method of claim 1, further comprising: determining a channel condition; and The step of determining the target component of the precoding information is based on the channel condition. The decompressed component comprises a decompressed version of the target component. The target component of the precoding information is determined using singular value decomposition.

13. The method of claim 1, wherein, 15. The method of claim 1, wherein:

14. The method of claim 1, wherein, the target component of the precoding information comprises target components of the precoding information for a first subband and a second subband; and the at least a portion of the precoding information is determined for a first subband and a second subband.

16. An apparatus for determining precoding information, comprising: a receiver configured to receive a reference signal using a channel; a transmitter configured to transmit the precoding information for the channel; and a controller configured to: determine a target component of the precoding information based on the reference signal; and determine at least a portion of the precoding information based on a decompressed component of the precoding information and the target component. The controller is configured to determine the at least a portion of the precoding information based on a correlation between the target component and the decompressed component.

18. The apparatus of claim 16, wherein: the at least a portion of the precoding information comprises one or more coefficients; the target component comprises a target precoding coefficient matrix; and 17. The apparatus of claim 16, wherein, the decompressed component comprises a decompressed version of the target component.

19. A user equipment, comprising: at least one processor configured to: determine a target component of a precoding matrix for a downlink channel based on a condition of the downlink channel; determine a decompressed component of the precoding matrix; and ​ ​ ​ ​ determining at least a portion of the precoding matrix based on the target component and the decompressed component; compressing the precoding matrix to generate a compressed component; and transmitting the compressed component using an uplink channel.

20. The user equipment of claim 19, wherein, the at least one processor is configured to determine the at least a portion of the precoding matrix based on a correlation between the target component and the decompressed component.

21. The user equipment of claim 20, wherein, the at least one processor is configured to determine the target component based on a coefficient matrix and vectors of scaling coefficients and rotation coefficients.

Images