TYPE II CSI

BACKGROUND:

Field:

[0001] Various communication systems may benefit from improved side channel state information management techniques.

Description of the Related Art:

[0002] 3rd Generation Partnership Project (3GPP) release (Rel)-15, new radio (NR) type II CSI channel state information (CSI) is an eigenvector approximation scheme for CSI feedback, also known as a CSI report, defined up to rank=2 transmission. This limitation is primarily due to the large feedback overhead which may result from a higher rank CSI feedback. The feedback overhead of NR type II may scale linearly with the rank of the CSI feedback if the legacy framework were extended. This would require a significant increase of the necessary uplink resources to perform the feedback. Despite this limitation, legacy type II codebooks may achieve up to a 36% performance enhancement over LTE at the cost of higher feedback overhead as compared to the latter.

BRIEF DESCRIPTION OF THE DRAWINGS:

[0003] For proper understanding of this disclosure, reference should be made to the accompanying drawings, wherein:

[0004] FIG. 1 illustrates an example of frequency domain channels between a first transmit beam and a first receive antenna across frequency domains.

[0005] FIG. 2 illustrates an example of an average chordal distance among 210 users. [0006] FIG. 3 illustrates another example of an average chordal distance among 210 users.

[0007] FIG. 4 illustrates an example of user perceived throughput vs uplink overhead performance Rel-16 when M = 7.

[0008] FIG. 5 illustrates an example of a signaling diagram according to certain embodiments.

[0009] FIG. 6 illustrates an example of a flow diagram of a method that may be performed by a user equipment according to certain embodiments.

[0010] FIG. 7 illustrates an example of a flow diagram of a method that may be performed by a network entity according to certain embodiments.

[0011] FIG. 8 illustrates an example of a system architecture according to certain embodiments.

DETAILED DESCRIPTION:

[0012] Under 3GPP Rel-15 NR type II CSI, precoding vectors may be denoted as w = w ₁ w ₂ . The final weighting vector at the network entity may be a weighted linear combination of L orthogonal beams per polarization, denoted as is a long-term 2D discrete fourier transform (DFT) beam; is a beam power scaling factor wideband; is a beam power scaling factor subband; and c _r,l,i is a beam combining coefficient.

[0013] Determining may begin with generating a grid-of-beam matrix W ₁ of size 2N ₁ N ₂ X 2L by choosing L orthogonal vectors/beams per polarization r from a set of oversampled O ₁ O ₂ N ₁ N ₂ DFT beams, wherein N ₁ and N ₂ are the number of antenna ports in horizontal and vertical domains, respectively, and O ₁ and O ₂ may be oversampling factors in both dimensions. This collection of vectors may be used to approximate the eigenvectors of the channel covariance matrix by means of suitable weighted linear combinations. This operation may achieve a compression in the spatial domain (SD); thus, the resulting 2 L beams may also be referred to as SD components. In some embodiments, one or more eigenvectors may be drawn from at least one covariance matrix computed using CSI of one or more subbands.

[0014] A linear combination subband matrix W ₂ may then be built, where for every subband, the coefficients to be used for the weighted linear combination of the columns of W ₁ may be calculated, yielding the approximation of the l strongest eigenvectors of the channel covariance matrix, wherein l denotes the number of layers. A quantization of linear combining coefficients may then be performed, wherein the correlation between the coefficients of the different W ₂ across all the subbands may be exploited to achieve a reduction of the overall number of coefficients to feedback by means of a differential wideband+subband quantization. [0015] 3GPP Rel-16 includes an enhancement of type II CSI feedback based on exploiting this frequency correlation. A frequency domain compression scheme may be applied on subband matrix W ₂ . The precoder for each layer, as well as across frequency-domain units W, may be derived as , wherein the frequency domain (FD) basis subset W _ƒN3xM may be derived from a DFT codebook; may be a matrix of linear combination coefficients; N ₃ may be the number of subbands; and M < N ₃ may be the number of FD coefficients. For example, may be a sparse matrix; thus, all elements inside may not need to be fed back, and instead only a subset of the elements inside may be quantized and fed back. The location of the subset of the elements inside may then be signaled via a bitmap.

[0016] In NR type II CSI under Rel-15 and Rel-16, approximated eigenvectors may be compressed, quantized, and fed back to the network entity. However, channel eigenvectors may not be predicted on the Euclidean space, such as in an explicit CSI case. Two non-zero vectors s ₁ , s ₂ ∈ C ^M may be equivalent s ₁ ~ s ₂ in terms of subspaces, but only if there exists a non-zero a ∈ C such that s ₁ = as ₂ . A different criterion to judge the deviation between two normalized eigenvectors s ₁ and s ₂ may be to compute a chordal distance, wherein the sin of the angle between the two eigenvectors on the Grassmannian manifold G _{M , 1} may be computed as

[0017] In LTE and NR, only a portion of the spectrum may be utilized, while the remained of the spectrum may be suppressed to abide by standardized spectral masks. For example, in a 10MHz bandwidth and 15kHz subcarrier spacing, only N _a = 624 subcarriers out of N _FFT = 1024 subcarriers may be used, while the remaining subcarriers move to zero, where N _FFT may be the fast Fourier transform (FFT) size. On the user equipment side, there may be no knowledge of the channel frequency response (CFR) outside of the used band, which may lead to a jump across the edges of the CFR, as illustrated in FIG. 1.

[0018] A similar increase may occur in the columns of W ₂ , which may lead to a poor CSI compression performance in the edge subbands, particularly when applying DFT-based FD compression steps, as illustrated in FIG. 2. FIG. 2 depicts the chordal distance for each subband eigenvector before and after CSI compression. Consequently, as illustrated in FIG. 2, for 3GPP ReI-16 users which are allocated at the edge subbands, illustrated as subbands #1 and #13, poorer CSI information may be available at the network entity side, and may result in a chordal distance loss of approximately 3.5dB.

[0019] Certain embodiments described herein may have various benefits and/or advantages to overcome the disadvantages described above. For example, some embodiments may provide a better estimate of the edge subbands or any set of subband(s) where the CSI quality is insufficient. Thus, certain embodiments are directed to improvements in computer-related technology.

[0020] Some embodiments described herein may also provide an enhancement to 3GPP ReI-15 FD compression, where side CSI (SCSI) may be transmitted, along with the CSI of Rel-16, in order to improve the estimation of edge subbands. SCSI information may be constructed by first generating quantized differential information. For example, following CSI compression, the edge subband eigenvector of subband i estimate may be denoted as , and the original subband eigenvector of subband I may be denoted by S _i . A differential vector may be built on an element-based division operation, using may first be properly rotated to be able to perform this division by a complex factor of The differential vector may be quantized and fed back to the network entity, as well as the CSI of Rel-16, allowing the network entity to more accurately estimate the CSI in the edge subbands, as illustrated in FIG. 4.

[0021] FIG. 5 illustrates an example of a signaling diagram showing communications between UE 530 and NE 540 where UE 530 generates CSI, and signalling between UE 530 and NE 540 is required to transfer the CSI from UE 530 to NE 540. UE 530 may be similar to UE 810, and NE 540 may be similar to NE 820, both illustrated in FIG. 8.

[0022] In step 501, NE 540 may transmit at least one CSI-RS resource configured to compute CSI feedback, also known as a CSI report, to UE 530, wherein the at least one CSI-RS resource may be configured to compute CSI feedback. In step 503, UE 530 may compute one or more of at least one CSI-RS reception and/or at least one CSI ^c omputation.

[0023] In step 505, UE 530 may compute one or more of at least one subband matrix, at least one bitmap, at least one FD basis subset matrix W _ƒ , and/or at least one linear combination coefficient matrix for subband and layer index over which the CSI is computed. In various embodiments, for each layer, UE 530 may derive W ₁ , W _ƒ bitmap and according to Rel-16. Additionally or alternatively, for each layer, UE 530 may generate , wherein may be obtain ed from the i ^th column inside after at least one proper phase rotation, and s _i may be obtained from the i ^th column inside W ₂ before CSI compression is performed. Furthermore, for each layer l and edge subband i, UE 530 may generate element-based division operation

[0024] In certain embodiments, the dynamic range of the phase quantizer may be chosen to be smaller than [—π, π] since the phase info inside may represent the deviation between the phases of the actual subband eigenvector and the compressed subband eigenvector. For example, all elements inside may first be rotated by e ^-jΦmin, where Φ _min may be the smallest angle inside such that to ensure that the smallest angle inside is zero. The quantizer range may then be [0, π] .

[0025] In some embodiments, for all subbands with CSI, the side information may be grouped as . The elements inside may be sparse, so in order to save overhead, a bitmap b _c2LxMe where only non-zero elements in R ¹ may be fed back to NE 540. The bitmap may also be chosen to be common across all subbands with CSI, wherein the size of the bitmap is b _c2Lx1 .

[0026] In step 507, UE 530 may transmit the one or more of at least one subband matrix , at least one bitmap, at least one FD basis subset matrix W _ƒ , and/or at least one linear combination coefficient matrix to NE 540. In some embodiments, the transmission may be according to at least one release version of UE 530. For example, when UE 530 is with a first release version, such as Release 15, UE 530 may transmit the one or more of at least one subband matrix to NE 540. In various embodiments, when UE 530 is associated with a second release, such as Release 16, UE 530 may transmit at least one linear combination coefficient matrix , at least one bitmap, and at least one FD basis subset matrix W _ƒ to NE 540.

[0027] In step 509, NE 540 may generate . In step 511, UE 530 may generate for each layer. In step 513, UE 530 may generate and quantize at least one differential vector. In certain embodiments, UE 530 may quantize and/or may generate is performed according to element-based multiplication operation.

[0028] In step 515, UE 530 may generate at least enhanced CSI by taking into account differential side info For example, in some embodiments, UE 530 may compute

[0029] In step 517, UE 530 may compare In various embodiments, if , UE 530 may assign bitmap vector , UE 530 may assign bitmap vector b _e (i, l)=0. Furthermore, the threshold γ _th may be configured via radio resource control (RRC) signalling and/or may be predetermined or computed according to at least one rule. However, any other metric may be used to determine whether applying CSI is desired. This step may be a UE implementation part.

[0030] In step 519, UE 530 may transmit one or more of at least one bitmap vector and at least one element-based operation to NE 540. In certain embodiments, UE 530 may transmit bitmap vector b _e (i, l) in uplink control information (UCI) part 1, which may be configured to enable NE 540 to predict the overhead needed in UCI part 2, as described in step 521. In various embodiments, for all the layers l and subbands t at which b _e (i, l) = 1, UE 530 may feedback in UCI part 2 to NE 540, in addition to the CSI from Rel-16.

[0031] In step 521, upon receiving bitmap vector b _e (i, l) in UCI part 1 in step 519, NE 540 may estimate at least one overhead. In step 523, upon receiving feedback in UCI part 2 in step 519, NE 540 may generate at least one merged CSI. For example, for all b _e (i, l) = 1, NE 540 may generate at least one merged version of the CSI at subbands

[0032] In step 525, NE 540 may update at least one second matrix. For example, NE 540 may update at subbands where b _e (i, l) = 1 with from step 515. In certain embodiments, NE 540 may generate from the main Rel-16 feedback.

[0033] In step 527, NE 540 may generate at least one precoding vector for each layer according to

[0034] FIG. 6 illustrates an example of a flow diagram of a method that may be performed by a UE, such as UE 810 illustrated in FIG. 8, according to certain embodiments.

[0035] In step 601, the UE may receive at least one reference signal configured for measurement of channel state information from a network entity (NE), such as NE 820 illustrated in FIG. 8, wherein the at least one reference signal may be configured to compute CSI feedback. In step 603, the UE may compute one or more of at least one CSI-RS reception and/or at least one CSI computation.

[0036] In step 605, the UE may compute one or more of at least one subband matrix, at least one bitmap, and/or at least one linear combination coefficient matrix. In various embodiments, for each layer, the UE may derive W ₁ , W _ƒ bitmap and , according to Rel-16. Additionally or alternatively, for each layer, the UE may generate wherein may be obtained from the i ^th column inside after at least one proper phase rotation, and s _i may be obtained from the i ^th column inside W ₂ before CSI compression is performed. Furthermore, for each layer / and edge subband i, the UE may generate element-based division operation

[0037] In certain embodiments, the dynamic range of the phase quantizer may be chosen to be smaller than [—π, π] since the phase info inside may represent the deviation between the phases of the actual subband eigenvector and the compressed subband eigenvector. For example, all elements inside may first be rotated by e ^-jΦmin _, where Φ _min may be the smallest angle insider such that to ensure that the smallest angle insider is zero. The quantizer range may then be [0, π] .

[0038] In some embodiments, for all subbands with CSI, the side information may be grouped as The elements inside may be sparse, so in order to save overhead, a bitmap b _c2LxMe where only non-zero elements in R ^l may be fed back to NE 540. The bitmap may also be chosen to be common across all subbands with CSI, wherein the size of the bitmap is b _c2Lx1 .

[0039] In step 607, the UE may transmit the computed one or more of at least one subband matrix , at least one bitmap, at least one ED basis subset matrix W _ƒ , and/or at least one linear combination coefficient matrix to the NE. In some embodiments, the transmission may be according to at least one release version of UE 530. For example, when the UE is with a first release version, such as Release 15, the UE may transmit the one or more of at least one subband matrix to the NE. In various embodiments, when the UE is associated with a second release, such as Release 16, the UE may transmit at least one linear combination coefficient matrix at least one bitmap, and at least one FD basis subset matrix W _ƒ to the NE.

[0040] In step 609, the UE may generate at least one differential vector based at least in part on channel state information. In step 611 , the UE may quantize the at least one differential vector. In certain embodiments, the UE may quantize and/or may generate is performed according to an element-based multiplication operation.

[0041] In step 613, the UE may generate at least one merged CSI. For example, in some embodiments, the UE may compute

[0042] In step 615, the UE may compare . In various embodiments, if the UE may assign bitmap vector the UE may assign bitmap vector b _e (i, l)=0. Furthermore, the threshold γ _th may be configured via radio resource control (RRC) signalling and/or may be predetermined or computed according to at least one rule. However, any other metric may be used to determine whether applying CSI is desired. This step may be a UE implementation part.

[0043] In step 617, the UE may transmit the at least one quantized differential vector to the NE. In certain embodiments, the UE may transmit bitmap vector b _e (i, l) in uplink control information (UCI) part 1, which may be configured to enable the NE to predict the overhead needed in UCI part 2. In various embodiments, for all the layers l and subbands i at which b _e (i, l) = 1 , the UE may feedback in UCI part 2 to the NE, in addition to the CSI from Rel-16. [0044] FIG. 7 illustrates an example of a flow diagram of a method that may be performed by a NE, such as NE 820 illustrated in FIG. 8, according to certain embodiments. In step 701, the NE may transmit at least one reference signal to a user equipment (UE), wherein the at least one reference signal may be configured to compute CSI feedback. In step 703, the NE may receive one or more of at least one subband matrix at least one bitmap, at least one FD basis subset matrix W _ƒ , and at least one linear combination coefficient matrix to from the UE. In some embodiments, the transmission may be according to at least one release version of the UE. For example, when the UE is with a first release version, such as Release 15, the NE may receive the one or more of at least one subband matrix from the UE. In various embodiments, when the UE is associated with a second release, such as Release 16, the NE may receive at least one linear combination coefficient matrix , at least one bitmap, and at least one FD basis subset matrix W _ƒ from the UE.

[0045] In step 705, the NE may generate . In step 707, the NE may receive at least one quantized differential vector from the UE, wherein the at least one quantized differential vector is generated based at least, in part, on channel state information. In certain embodiments, the NE may receive bitmap vector b _e (i, l) in uplink control information (UCI) part 1, which may be configured to enable the NE to predict the overhead needed in UCI part 2. In various embodiments, for all the layers / and subbands i at which b _e (i, l) = 1, the NE may receive feedback in UCI part 2 to from the UE, in addition to the CSI from Rel-16.

[0046] In step 709, upon receiving bitmap vector b _e (i, l) in UCI part 1 in step 707, the NE may estimate at least one overhead. In step 711, upon receiving feedback in UCI part 2 in step 707, the NE may generate at least one merged CSI. For example, for all b _e (i, l) = 1, the NE may generate at least one merged version of the CSI at subbands

[0047] In step 713, the NE may update at least one second matrix. For example, the NEmayupdate at subbands where . In certain embodiments, the NE may generate from the main Rel-16 feedback. In step 715, the NE may generate at least one precoding vector for each layer according to [0048] FIG. 8 illustrates an example of a system according to certain embodiments. In one embodiment, a system may include multiple devices, such as, for example, user equipment 810 and/or network entity 820.

[0049] User equipment 810 may include one or more of a mobile device, such as a mobile phone, smart phone, personal digital assistant (PDA), tablet, or portable media player, digital camera, pocket video camera, video game console, navigation unit, such as a global positioning system (GPS) device, desktop or laptop computer, single- location device, such as a sensor or smart meter, or any combination thereof.

[0050] Network entity 820 may be one or more of a base station, such as an evolved node B (eNB) or 5G or New Radio node B (gNB), a serving gateway, a session management function (SMF), a user plane function (UPF), a 5G NG-RAN node, a server, and/or any other access node or combination thereof. Furthermore, network entity 820 and/or user equipment 810 may be one or more of a citizens broadband radio service device (CBSD).

[0051] One or more of these devices may include at least one processor, respectively indicated as 811 and 821. Processors 811 and 821 may be embodied by any computational or data processing device, such as a central processing unit (CPU), application specific integrated circuit (ASIC), or comparable device. The processors may be implemented as a single controller, or a plurality of controllers or processors. [0052] At least one memory may be provided in one or more of devices indicated at 812 and 822. The memory may be fixed or removable. The memory may include computer program instructions or computer code contained therein. Memories 812 and 822 may independently be any suitable storage device, such as a non-transitory computer-readable medium. A hard disk drive (HDD), random access memory (RAM), flash memory, or other suitable memory may be used. The memories may be combined on a single integrated circuit as the processor, or may be separate from the one or more processors. Furthermore, the computer program instructions stored in the memory and which may be processed by the processors may be any suitable form of computer program code, for example, a compiled or interpreted computer program written in any suitable programming language. Memory may be removable or non-removable.

[0053] Processors 811 and 821 and memories 812 and 822 or a subset thereof, may be configured to provide means corresponding to the various blocks of FIGS. 5-7. Although not shown, the devices may also include positioning hardware, such as GPS or micro electrical mechanical system (MEMS) hardware, which may be used to determine a location of the device. Other sensors are also permitted and may be included to determine location, elevation, orientation, and so forth, such as barometers, compasses, and the like.

[0054] As shown in FIG. 8, transceivers 813 and 823 may be provided, and one or more devices may also include at least one antenna, respectively illustrated as 814 and 824. The device may have many antennas, such as an array of antennas configured for multiple input multiple output (MIMO) communications, or multiple antennas for multiple radio access technologies. Other configurations of these devices, for example, may be provided. Transceivers 813 and 823 may be a transmitter, a receiver, or both a transmitter and a receiver, or a unit or device that may be configured both for transmission and reception.

[0055] The memory and the computer program instructions may be configured, with the processor for the particular device, to cause a hardware apparatus such as user equipment to perform any of the processes described below (see, for example, FIGS. 5-7). Therefore, in certain embodiments, a non-transitory computer-readable medium may be encoded with computer instructions that, when executed in hardware, perform a process such as one of the processes described herein. Alternatively, certain embodiments may be performed entirely in hardware.

[0056] In certain embodiments, an apparatus may include circuitry configured to perform any of the processes or functions illustrated in FIGS. 5-7. For example, circuitry may be hardware-only circuit implementations, such as analog and/or digital circuitry. In another example, circuitry may be a combination of hardware circuits and software, such as a combination of analog and/or digital hardware circuit(s) with software or firmware, and/or any portions of hardware processor(s) with software (including digital signal processor(s)), software, and at least one memory that work together to cause an apparatus to perform various processes or functions. In yet another example, circuitry may be hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that include software, such as firmware for operation. Software in circuitry may not be present when it is not needed for the operation of the hardware. [0057] The features, structures, or characteristics of certain embodiments described throughout this specification may be combined in any suitable manner in one or more embodiments. For example, the usage of the phrases “certain embodiments,” “some embodiments,” “other embodiments,” or other similar language, throughout this specification refers to the fact that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present invention. Thus, appearance of the phrases “in certain embodiments,” “in some embodiments,” “in other embodiments,” or other similar language, throughout this specification does not necessarily refer to the same group of embodiments, and the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. Furthermore, notations including a hat (^) indicate an estimate of a value, i.e., IV is an estimate of W.

[0058] One having ordinary skill in the art will readily understand that certain embodiments discussed above may be practiced with steps in a different order, and/or with hardware elements in configurations which are different than those which are disclosed. Therefore, it would be apparent to those of skill in the art that certain modifications, variations, and alternative constructions would be apparent, while remaining within the spirit and scope of the invention. In order to determine the metes and bounds of the invention, therefore, reference should be made to the appended claims.

[0059] Partial Glossary

[0060] 3GPP 3rd Generation Partnership Project

[0061] BS Base Station [0062] BWP Bandwidth Part [0063] CFR Channel Frequency Response [0064] CSI Channel State Information [0065] DFT Discrete Fourier Transform [0066] DL Downlink [0067] eMBB Enhanced Mobile Broadband [0068] eNB Evolved Node B [0069] EPS Evolved Packet System [0070] FDD Frequency Division Duplex [0071] FD Frequency Domain [0072] gNB Next Generation Node B [0073] GPS Global Positioning System [0074] LC Linear Combination [0075] LTE Long-Term Evolution [0076] MAC Medium Access Control [0077] MIMO Multiple-Input Multiple- Output [0078] MR Maximum Rank [0079] NR New Radio [0080] PMI Precoding Matrix Indicator [0081] PRB Physical Resource Block [0082] RAN Radio Access Network [0083] RB Resource Block [0084] SB Subband [0085] SCSI Side Channel State Information [0086] SD Spatial Domain [0087] TS Technical Specification [0088] UCI Uplink Control Information [0089] UE User Equipment [0090] UL Uplink [0091] UPT User Perceived Throughput [0092] URLLC Ultra-Reliable and Low-Latency Communication [0093] WB Wideband [0094] WLAN Wireless Local Area Network [0095] According to a first embodiment, a method may include receiving, by a user equipment, at least one reference signal configured for measurement of channel state information from a network entity (NE). The method may further include generating, by the user equipment, at least one differential vector based at least in part on channel state information. The method may further include quantizing the at least one differential vector. The method may further include transmitting, by the user equipment, the at least one quantized differential vector to the NE.

[0096] In a variant, the generating at least one differential vector may comprise rotating at least one eigenvector estimate of channel state information of at least one first subband.

[0097] In a variant, the reference signal may comprise at least one channel state information reference signal.

[0098] In a variant, the at least one quantized differential vector may be generated based on at least one of a first bitmap or an element-based operation.

[0099] In a variant, the element-based operation may comprise element-based division operation.

[0100] In a variant, the element-based division operation may further comprise dividing at least one eigenvector estimate of the channel state information of the at least one first subband by at least one eigenvector of the channel state information of the at least one first subband.

[0101] In a variant, the at least one eigenvector estimate of the channel state information comprises one or more of quantizing and compressing the at least one eigenvector of the channel state information.

[0102] In a variant, the rotating may be performed prior to the dividing.

[0103] In a variant, the at least one first subband may comprise at least one subband of at least one frequency band.

[0104] In a variant, the method may further include generating, by the UE, the channel state information feedback.

[0105] In a variant, the method may further include generating, by the UE, one or more of at least one subband matrix, at least one frequency domain subset matrix, a second bitmap, or at least one linear combination coefficient matrix.

[0106] In a variant, the method may further include transmitting, by the UE, the generated one or more of at least one subband matrix, at least one frequency domain subset matrix, a second bitmap, or at least one linear combination coefficient matrix. [0107] In a variant, the method may further include generating, by the UE, at least one linear combination subband matrix for one layer.

[0108] In a variant, the method may further include generating, by the UE, at least one merged channel state information based at least partially on the channel state information feedback and at least partially on the at least one quantized differential vector. [0109] In a variant, the method may further include comparing, by the UE, the at least one merged channel state information to the channel state information feedback.

[0110] In a variant, the method may further include determining, by the UE, whether to send the at least one quantized differential vector on a subband or not.

[0111] In a variant, the method may further include transmitting, by the UE, the first bitmap which indicates at least one subband configured to transmit the at least one quantized differential vector.

[0112] In a variant, the method may further include transmitting the at least one quantized differential vector according to the first bitmap.

[0113] According to a second embodiment, a method may include transmitting, by a network entity (NE), at least one reference signal to a user equipment (UE). The method may further include receiving, by the network entity (NE), at least one quantized differential vector from the user equipment, wherein the at least one quantized differential vector is generated based at least, in part, on channel state information. The at least one quantized differential vector may be a quantization result based on at least one differential vector.

[0114] In a variant, the reference signal may comprise at least one channel state information reference signal.

[0115] In a variant, the at least one quantized differential vector may be generated based on at least one of the first bitmap or an element-based operation.

[0116] In a variant, the element-based operation may comprise element-based division operation.

[0117] In a variant, the at least one first subband may comprise at least one of edge subband, side subband, or non-central subband.

[0118] In a variant, the element-based division operation may further comprise dividing at least one eigenvector estimate of the channel state information of the at least one first subband by at least one eigenvector of the channel state information of the at least one first subband.

[0119] In some variants, the element-based operation may further comprise rotation of the at least one eigenvector estimate of channel state information of the at least one first subband. [0120] In some variants, the rotation may occur prior to the element-based division operation.

[0121] In a variant, the method may further include receiving, by the NE, one or more of at least one subband matrix, at least one frequency domain subset matrix, a second bitmap, or at least one linear combination coefficient matrix from the UE.

[0122] In a variant, the method may further include generating, by the NE, at least one linear combination coefficient matrix.

[0123] In a variant, the method may further include estimating, by the NE, overhead in at least part of an uplink control information.

[0124] In a variant, the method may further include generating, by the NE, at least one merged channel state information.

[0125] According to a third embodiment and a fourth embodiment, an apparatus can include at least one processor and at least one memory and computer program code. The at least one memory and the computer program code can be configured to, with the at least one processor, cause the apparatus at least to perform a method according to the first embodiment, the second embodiment, and the third embodiment, and any of its variants.

[0126] According a fifth embodiment and a sixth embodiment, an apparatus can include means for performing the method according to the first embodiment, the second embodiment, and the third embodiment, and any of its variants.

[0127] According to a seventh embodiment and an eighth embodiment, a computer program product may encode instructions for performing a process including a method according to the first embodiment, the second embodiment, and the third embodiment, and any of its variants.

[0128] According to a ninth embodiment and a tenth embodiment, a non-transitory computer-readable medium may encode instructions that, when executed in hardware, perform a process including a method according to the first embodiment, the second embodiment, and the third embodiment, and any of its variants.

[0129] According to an eleventh embodiment and a twelfth embodiment, a computer program code may include instructions for performing a method according to the first embodiment, the second embodiment, and the third embodiment, and any of its variants. [0130] According to a thirteenth embodiment and a fourteenth embodiment, an apparatus may include circuitry configured to perform a process including a method according to the first embodiment, the second embodiment, and the third embodiment, and any of its variants.