Method and apparatus for channel estimation and precoding through imperfect channel observations and channel state information feedback

Abstract

A method implemented by a node, comprising: estimating first downlink (DL) channel information of a first DL channel according to a reference signal received on a first uplink (UL) channel, where the first UL channel is a subset of available UL channels, and the first DL channel corresponds to the first UL channel; receiving channel feedback of a set of DL channels; estimating second DL channel information according to the estimated first DL channel information and the received channel feedback; and performing communication according to the second DL channel information.

Description

Technical Field

[0001] The present disclosure generally relates to methods and apparatuses for digital communication, and in particular embodiments, to methods and apparatuses for channel estimation and precoding through incomplete channel observations and channel state information (CSI) feedback. Background Art

[0002] In modern communication systems, such as in Third Generation Partnership Project (3GPP) Fifth Generation (5G) and 3GPP Long Term Evolution (LTE) compatible communication systems, large-scale multiple input multiple output (MIMO) communication with many transmit antennas significantly increases the overall capacity of the system. The performance of MIMO transmission depends on the knowledge and accuracy of the downlink (DL) channel state information (CSI), which is used to estimate the channel and determine the channel matrix.

[0003] In a time division duplexed (TDD) communication system, information about the DL channel can be obtained by measuring the uplink (UL) channel that utilizes channel reciprocity. In a frequency division duplexed (FDD) communication system, information about the DL channel can be obtained from the quantized CSI feedback measured and reported by a user equipment (UE).

[0004] In a TDD communication system, in order to obtain complete DL channel information using UL measurements, the UE must transmit reference signals, such as sounding reference signals (SRS), on each transmit antenna. This process is called UL channel sounding. However, although the UE typically uses all its antennas for DL reception, the UE typically does not use all its antennas for UL transmission. This limitation may be caused by cost reduction measures, where the UE does not have a transmit radio frequency (RF) chain for each antenna. Thus, in such a UE, the UE can only transmit reference signals on the antennas having a transmit RF chain at any given time.

[0005] In some UE implementations, a switch is used to enable the switching of the transmit RF chains so that all antennas of the UE can be used to transmit reference signals. However, transmitting reference signals using all antennas requires multiple transmission opportunities. In addition, both the complexity and cost of the switch are high. Additionally, the latency associated with switching and transmitting reference signals in different transmission opportunities increases the overhead related to UL measurements based on reference signal transmission.

[0006] Therefore, there is a need for methods and apparatuses for channel estimation and precoding through incomplete channel observations and CSI feedback. Summary of the Invention

[0007] According to a first aspect, there is provided a method implemented by a node. The method includes: the node estimating first DL channel information of a first downlink (DL) channel based on a reference signal received on a first uplink (UL) channel, where the first UL channel is a subset of the UL channel and the first DL channel corresponds to the first UL channel; the node receiving channel feedback of a set of DL channels; the node estimating second DL channel information based on the estimated first DL channel information and the received channel feedback; and the node communicating according to the second DL channel information.

[0008] According to the first aspect, in a first implementation manner of the method, the second DL channel information includes information derived from the estimated first DL channel information and an orthogonal projection of the received channel feedback.

[0009] According to the first aspect or the first aforementioned first implementation manner of the first aspect, in a second implementation manner of the method, estimating the second DL channel information includes: the node projecting channel information derived from the received channel feedback onto an orthogonal subspace of the first DL channel, thereby generating a projection; the node selecting one or more principal eigenvectors of the projection; and the node combining the estimated first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

[0010] According to the first aspect or any of the aforementioned implementation manners of the first aspect, in a third implementation manner of the method, communicating according to the second DL channel information includes: the node precoding data according to the second DL channel information.

[0011] According to the first aspect or any of the aforementioned implementation manners of the first aspect, in a fourth implementation manner of the method, it further includes: the node processing the received channel feedback after receiving the channel feedback, thereby generating processed channel feedback.

[0012] According to the first aspect or any of the foregoing implementations of the first aspect, in the fifth implementation of the method, estimating the second DL channel information includes: the node projects the channel information derived from the processed channel feedback onto the orthogonal subspace of the first DL channel, thereby generating a projection; the node selects one or more principal eigenvectors of the projection; and the node combines the estimated first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

[0013] According to the first aspect or any of the foregoing implementations of the first aspect, in the sixth implementation of the method, processing the received channel feedback includes: filtering or denoising the received channel feedback using channel statistical information.

[0014] According to the first aspect or any of the foregoing implementations of the first aspect, in the seventh implementation of the method, filtering or denoising the received channel feedback includes at least one of the following: performing minimum mean square error (MMSE) filtering on the received channel feedback, projecting the principal subspace of the received channel feedback, performing filtering based on maximum fairness (MF) or maximal ratio combining (MRC) on the received channel feedback, and performing quadratic programming on the received channel feedback.

[0015] According to the first aspect or any of the foregoing implementations of the first aspect, in the eighth implementation of the method, processing the received channel feedback includes: the node determines the conditional covariance of the received channel feedback; and the node filters or denoises the received channel feedback according to the conditional covariance.

[0016] According to the first aspect or any of the foregoing implementations of the first aspect, in the ninth implementation of the method, filtering or denoising the received channel feedback includes at least one of the following: performing MMSE filtering on the received channel feedback, projecting the principal subspace of the received channel feedback, performing filtering based on MF or MRC on the received channel feedback, and performing quadratic programming on the received channel feedback.

[0017] According to the first aspect or any of the foregoing implementations of the first aspect, in the tenth implementation of the method, the second DL channel information includes a DL precoder.

[0018] According to a second aspect, a method implemented by a node is provided. The method includes: the node estimating first DL channel information of a first DL channel according to a reference signal received on a first UL channel, where the first UL channel is a subset of UL channels, and the first DL channel is associated with the first UL channel; the node receiving channel feedback of a set of DL channels; the node performing eigenvalue decomposition on the first DL channel information; the node selecting one or more first principal eigenvectors of the eigenvalue decomposition of the first DL channel information as a partial precoding procedure; the node projecting channel information derived from the channel feedback onto an orthogonal subspace of the first DL channel; the node selecting one or more second principal eigenvectors of the channel information after projection of the received feedback channel information as second DL channel information of a second DL channel, where the second DL channel is not associated with the first UL channel; and the node combining the partial precoding procedure and the one or more second principal eigenvectors to form a precoding procedure.

[0019] According to the second aspect, in a first implementation manner of the method, the set of DL channels includes the first DL channel and the second DL channel.

[0020] According to the second aspect or the aforementioned first implementation manner of the second aspect, in a second implementation manner of the method, the reference signal includes a sounding reference signal (SRS).

[0021] According to the second aspect or any of the aforementioned implementation manners of the second aspect, in a third implementation manner of the method, the method further includes: after receiving the channel feedback, the node processing the received channel feedback.

[0022] According to the second aspect or any of the aforementioned implementation manners of the second aspect, in a fourth implementation manner of the method, processing the received channel feedback includes: filtering or denoising the received channel feedback using channel statistical information.

[0023] According to the second aspect or any of the aforementioned implementation manners of the second aspect, in a fifth implementation manner of the method, the method further includes: the node determining a conditional covariance of the received channel feedback, where the filtering or denoising of the received channel feedback is performed according to the conditional covariance.

[0024] According to the second aspect or any of the foregoing implementations of the second aspect, in the sixth implementation of the method, filtering or denoising the received channel feedback includes at least one of the following: performing MMSE filtering on the received channel feedback, projecting the principal subspace of the received channel feedback, performing MF- or MRC-based filtering on the received channel feedback, and performing quadratic programming on the received channel feedback.

[0025] According to a third aspect, there is provided a node, the node comprising: a non-transitory memory storage including instructions; and one or more processors in communication with the memory storage, wherein the one or more processors execute the instructions to cause the node to perform the following operations: estimating first DL channel information of a first DL channel based on a reference signal received on a first UL channel, wherein the first UL channel is a subset of UL channels and the first DL channel corresponds to the first UL channel; receiving channel feedback of a set of DL channels; estimating second DL channel information based on the estimated first DL channel information and the received channel feedback; and communicating based on the second DL channel information.

[0026] According to the third aspect, in a first implementation of the node, the one or more processors further execute the instructions to cause the node to project channel information derived from the received channel feedback onto an orthogonal subspace of the first DL channel, thereby generating a projection; select one or more principal eigenvectors of the projection; and combine the estimated first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

[0027] According to the third aspect or the foregoing first implementation of the third aspect, in a second implementation of the node, the one or more processors further execute the instructions to cause the node to precode data based on the second DL channel information.

[0028] According to the third aspect or any of the foregoing implementations of the third aspect, in a fourth implementation of the node, the one or more processors further execute the instructions to cause the node to process the received channel feedback after receiving the channel feedback.

[0029] According to the third aspect or any of the foregoing implementations of the third aspect, in a fifth implementation of the node, the one or more processors further execute the instructions to cause the node to project channel information derived from the processed channel feedback onto an orthogonal subspace of the first DL channel, thereby generating a projection; select one or more principal eigenvectors of the projection; and combine the estimated first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

[0030] According to the third aspect or any of the foregoing implementations of the third aspect, in a sixth implementation of the node, the one or more processors further execute the instructions to cause the node to filter or denoise the received channel feedback using channel statistics.

[0031] According to the third aspect or any of the foregoing implementations of the third aspect, in a seventh implementation of the node, the received channel feedback is filtered or denoised by at least one of the following: performing MMSE filtering on the received channel feedback, projecting a principal subspace of the received channel feedback, performing MF- or MRC-based filtering on the received channel feedback, and performing quadratic programming on the received channel feedback.

[0032] According to the third aspect or any of the foregoing implementations of the third aspect, in an eighth implementation of the node, the one or more processors further execute the instructions to cause the node to determine a conditional covariance of the received channel feedback; and the node filters or denoises the received channel feedback according to the conditional covariance.

[0033] According to the third aspect or any of the foregoing implementations of the third aspect, in a ninth implementation of the node, the received channel feedback is filtered or denoised by at least one of the following: performing MMSE filtering on the received channel feedback, projecting a principal subspace of the received channel feedback, performing MF- or MRC-based filtering on the received channel feedback, and performing quadratic programming on the received channel feedback.

[0034] According to the third aspect or any of the foregoing implementations of the third aspect, in a tenth implementation of the node, the second DL channel information includes a DL precoding procedure.

[0035] According to a fourth aspect, there is provided a node, the node comprising: a non-transitory memory storage including instructions; and one or more processors in communication with the memory storage, wherein the one or more processors execute the instructions to cause the node to perform the following operations: estimate first DL channel information of a first DL channel according to a reference signal received on a first UL channel, wherein the first UL channel is a subset of UL channels and the first DL channel is associated with the first UL channel; receive channel feedback of a set of DL channels; perform eigen decomposition on the first DL channel information; select one or more first principal eigenvectors of the eigen decomposition of the first DL channel information as a partial precoding procedure; project channel information derived from the channel feedback onto an orthogonal subspace of the first DL channel; select one or more second principal eigenvectors of the channel information after projection of the received feedback channel information as second DL channel information of a second DL channel, wherein the second DL channel is not associated with the first UL channel; and combine the partial precoding procedure and the one or more second principal eigenvectors to form a precoding procedure.

[0036] According to the fourth aspect, in a first implementation manner of the node, the set of DL channels includes the first DL channel and the second DL channel.

[0037] According to the fourth aspect or the foregoing first implementation manner of the fourth aspect, in a second implementation manner of the node, the reference signal includes SRS.

[0038] According to the fourth aspect or any of the foregoing implementation manners of the fourth aspect, in a third implementation manner of the node, the one or more processors further execute the instructions to cause the node to process the channel feedback after receiving the channel feedback.

[0039] According to the fourth aspect or any of the foregoing implementation manners of the fourth aspect, in a fourth implementation manner of the node, the one or more processors further execute the instructions to cause the node to determine a conditional covariance of the channel feedback, wherein the channel feedback is filtered or denoised according to the conditional covariance.

[0040] According to the fourth aspect or any of the foregoing implementation manners of the fourth aspect, in a fifth implementation manner of the node, the channel feedback is filtered or denoised by at least one of the following: performing MMSE filtering on the channel feedback, projecting a principal subspace of the channel feedback, performing MF- or MRC-based filtering on the channel feedback, and performing quadratic programming on the channel feedback.

[0041] The advantages of the preferred embodiment are that channel estimation is performed using channel state information (CSI) feedback that may have quantization errors and incomplete channel information based on UL channel sounding. Using CSI feedback and incomplete channel information based on UL channel sounding helps to overcome the quantization errors present in CSI feedback while overcoming the limitations associated with obtaining full channel information based on UL channel sounding due to hardware constraints.

[0042] Another advantage of the preferred embodiment is that channel statistical information is used to further improve the accuracy of channel estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] To more fully understand the present disclosure and its advantages, reference is now made to the following description taken in conjunction with the accompanying drawings, in which:

[0044] Figure 1 An exemplary wireless communication system is shown;

[0045] Figure 2 An exemplary communication system is shown, providing a mathematical expression of a signal transmitted in the communication system;

[0046] Figure 3 An exemplary user equipment (UE) is shown;

[0047] Figure 4 A flowchart of an exemplary operation performed at a node communicating using a channel matrix according to an exemplary embodiment presented herein, where the channel matrix is generated based on channel information obtained from both channel state information (CSI) feedback and an uplink (UL) reference signal;

[0048] Figure 5 A flowchart of an exemplary operation performed at a node estimating downlink (DL) channel information with orthogonal subspace projection according to an exemplary embodiment provided herein;

[0049] Figure 6 A flowchart of an exemplary operation performed at a node determining a precoding procedure for a DL channel based on estimated channel information according to an exemplary embodiment provided herein;

[0050] Figure 7 An exemplary node highlighting circuit according to an exemplary embodiment provided herein to enhance channel estimation using UL channel sounding-based techniques and CSI feedback for determining DL channel information, where the node improves the quality of DL channel estimation by processing the CSI feedback;

[0051] Figure 8An exemplary node highlighting circuit according to an exemplary embodiment provided herein is shown to enhance channel estimation by using UL channel sounding-based techniques and CSI feedback for determining DL channel information, wherein the node improves the quality of DL channel estimation by processing CSI feedback with conditional covariance;

[0052] Figure 9A and Figure 9B A data graph of the downlink signal-to-noise ratio (SNR) and the sum rate R according to an exemplary embodiment proposed herein is shown, comparing the embodiment techniques with channel information having only CSI feedback;

[0053] Figure 10 An exemplary communication system according to an exemplary embodiment proposed herein is shown;

[0054] Figure 11A and Figure 11B An exemplary device that can implement the methods and teachings provided by the present disclosure is shown; and

[0055] Figure 12 A block diagram of a computing system that can be used to implement various devices and methods disclosed herein. Detailed Description

[0056] The structures and uses of the disclosed embodiments are discussed in detail below. However, it should be understood that the present disclosure provides many applicable concepts that can be embodied in a variety of specific contexts. The specific embodiments discussed merely illustrate the specific structures and uses of the embodiments and do not limit the scope of the present disclosure.

[0057] Figure 1 An exemplary wireless communication system 100 is shown. The communication system 100 includes an access node 110 having a coverage area 111. The access node 110 serves a plurality of user equipments (UEs), including UE 120 and UE 122. The UEs in the communication system can be mobile terminals, PCs, tablets, wearable devices (such as watches), and other devices having the ability to communicate wirelessly with the access node. Transmissions from the access node 110 to the UEs are referred to as downlink (DL) transmissions and occur on the downlink channel ( Figure 1 shown as solid arrow line 112 therein), while transmissions from the UEs to the access node 110 are referred to as uplink (UL) transmissions and occur on the uplink channel ( Figure 1 shown as dashed arrow line 114 therein). Services can be provided to the plurality of UEs by a service provider connected to the access node 110 through a backhaul network 130 (such as the Internet). The wireless communication system 100 can include a plurality of distributed access nodes 110.

[0058] In a typical communication system, there are several operating modes. In the cellular operating mode, communication is carried out with multiple UEs through an access node 110, while in the device-to-device communication mode, for example, in the proximity service (ProSe) operating mode, UEs can communicate directly with each other. The access node can usually also be referred to as NodeB, evolved NodeB (eNB), next generation (NG) NodeB (gNB), master eNB (MeNB), secondary eNB (SeNB), master gNB (MgNB), secondary gNB (SgNB), network controller, control node, base station, access point, transmission point (TP), transmission-reception point (TRP), cell, carrier, macro cell, femto cell, pico cell, relay station, customer premises equipment (CPE), etc. The UE can usually also be referred to as mobile station, mobile phone, terminal, user, subscriber, station, communication device, CPE, relay station, Integrated Access and Backhaul (IAB) relay station, etc. It should be noted that when using relays (such as based on relay stations, pico cells, CPEs, etc.), especially in multi-hop relays, the boundary between the controller and the nodes controlled by the controller may become blurred, and in a two-node (controller or node controlled by the controller) deployment, the first node that provides configuration or control information to the second node is considered the controller. Similarly, the concepts of UL and DL transmissions can also be extended. In some cases (such as in sidelink, vehicle-to-vehicle (V2V), vehicle-to-anything (V2X) communications, etc.), the UE can operate in a manner similar to an access node. In this case, the UE can also be referred to as a node.

[0059] Figure 2 An exemplary communication system 200 is shown, providing a mathematical expression of the signal transmitted in the communication system. The communication system 200 includes an access node 205 that communicates with a UE 210. As Figure 2 shown, the DL transmission received by the UE 210 can be expressed as

[0060] y = H H x + n

[0061] where H:n T ×nR is a multiple input multiple output (MIMO) channel matrix, H H is the Hermitian of H, n T is the number of transmit antennas, n R is the number of receive antennas, x:n T ×1 is the transmit signal vector, y:n R ×1 is the receive signal vector, n:n R ×1 is the receive noise vector. In addition, x is the precoded signal and can be expressed as

[0062] x = Ws,

[0063] where W:n T ×r is the precoding matrix or weight, r is the transmission rank, and s:r×1 is the vector of transmission data symbols. For example, the precoding matrix W is selected together with the channel matrix H to maximize the sum data rate.

[0064] For DL channel estimation at UE 210, a typical technique includes transmitting pilots on each antenna port over one or a set of orthogonal time-frequency resources. Let h i ′ be the i-th row of H, then the pilot signal received at the UE can be expressed as

[0065] y i = h′ i H x i + n, i = 1,…,n T .

[0066] In this case, the pilot x i is known to UE 210. Then, UE 210 can determine or estimate h′ i according to the equation y i H , i = 1,…,n T , and thus, determine or estimate the entire H H .

[0067] Although Figure 2 only one access node and one UE are shown, the communication system 200 is not limited to this case. Multiple UEs can be served by the access node on different time-frequency resources (e.g., in an FDM-TDM communication system, such as in a typical cellular system) or on the same time-frequency resources (e.g., in a multi-user MIMO (MU-MIMO) communication system, where multiple UEs are paired together and the transmission to each UE is precoded separately).

[0068] Other mathematical symbols used in the discussion are as follows:

[0069] Feedback rank;

[0070] Optimal precoding matrix mapped from the feedback precoding matrix index (PMI);

[0071] From the i-th layer of the feedback channel quality indicator (CQI) Mapped signal to interference plus noise ratio (SINR);

[0072] Diagonal matrix with diagonal entries being

[0073] A column of DL channels or a set of DL channel columns estimated from UL channel sounding. Sometimes replaced by actual values h1, H1, assuming good channel estimation with high precision;

[0074] Normalized

[0075] Approximate channel matrix from CSI feedback;

[0076] Z: Project Matrix after projecting onto the orthogonal subspace;

[0077] w1, W1: Transmit precoding column or part 1, channel estimated from SRS Obtained;

[0078] w2, W2: Transmit precoding column or part 2, obtained from the approximate channel and projection;

[0079] R T :n T ×n T Channel space covariance at the transmitter;

[0080] R R :n R ×n R Channel space covariance at the receiver;

[0081] U: Eigenvector with non-zero eigenvalues; Is RT rank;

[0082] U d :n T × d, d main eigenvectors from U,

[0083] Λ: diagonal matrix with non - zero eigenvalues λ i , ;

[0084] C: independent and identically - distributed (i.i.d.) Gaussian complex matrix for fading channel generation;

[0085] channel subset defined by feedback PMI;

[0086] transmission conditional covariance, conditioned on feedback PMI at ;

[0087] eigenvectors with non - zero eigenvalues; is rank of;

[0088] n T × d, d from main eigenvectors of, and

[0089] diagonal matrix with non - zero eigenvalues λ i , .

[0090] As mentioned above, many UEs, especially low - cost UEs, cannot transmit simultaneously using all available antennas, while they can generally receive simultaneously using all available antennas. This limitation is due to the fact that the number of transmit radio frequency (RF) chains in a UE is less than the number of antennas. This limitation may be a cost - reduction measure.

[0091] Figure 3 illustrates an exemplary UE 300. As Figure 3As shown, UE 300 includes two antennas, antenna 305 and antenna 307. UE 300 is capable of receiving simultaneously using both antennas. However, due to the limited number of RF transmit chains present in UE 300, UE 300 can only transmit simultaneously on one antenna. For example, some UEs have a switching circuit such that the UE transmits on the first antenna and then switches to the second antenna for subsequent transmissions. However, these UEs still cannot transmit simultaneously using all available antennas and require multiple transmission opportunities to transmit on all antennas.

[0092] UE 300 is a specific example of a hardware - limited UE, where the UE has a total of Y antennas, all of which are capable of receiving simultaneously. However, the UE has only X RF transmit chains (where X is less than Y), so the UE can only transmit simultaneously on X antennas. To enable the UE to transmit on all Y antennas, the UE requires multiple transmission opportunities, enabling the UE to transmit on the first set of X antennas, switch to the second set of X antennas and transmit on the second set of X antennas, and so on.

[0093] In addition, some low - cost UEs do not include a switching circuit that enables them to switch between different sets of transmit antennas. In such cases, these UEs can only transmit on a specific set of antennas and cannot switch to other antennas. The exemplary embodiments presented herein focus on these low - cost UEs. However, the exemplary embodiments are also applicable to UEs with a switching circuit that do not switch the transmit antennas, for example, perhaps due to time constraints.

[0094] The first technique that can be used to obtain information about the DL channel is CSI feedback. CSI feedback is commonly used in FDD communication systems. In CSI feedback, the access node transmits reference signals (such as CSI reference signals (CSI - RS)), and the UE that receives the reference signals uses these reference signals to measure the DL channel. The UE generates channel information from the measurements and reports the channel information to the access node. The channel information reported to the access node includes:

[0095] – Rank feedback

[0096] – PMI feedback And

[0097] – CQI feedback

[0098] However, CSI feedback is quantized to reduce communication overhead, and the quantization of the channel information introduces quantization errors. The quantization error is the difference between the actual value and its quantized value. The quantization error reduces the accuracy of the channel information reported by the UE, resulting in channel estimation errors determined based on the channel information.

[0099] A second technique that can be used to obtain information about the DL is to utilize the channel reciprocity of channel measurements based on UL channel sounding. Techniques based on UL channel sounding are typically used in TDD communication systems. In techniques based on UL channel sounding, the UE transmits reference signals (e.g., sounding reference signals (SRS)), and these reference signals are used by the access node that receives the reference signals to perform measurements on the UL channel. The access node generates channel information for the DL channel based on the measurements. Since in a TDD communication system, the UL channel and the DL channel share a common frequency band, channel reciprocity enables the measurements performed on the UL channel to be applied to the DL channel.

[0100] However, since some UEs, especially low-cost UEs, cannot transmit simultaneously on all antennas, the access node cannot measure all UL channels in a single transmission opportunity. Therefore, the access node cannot generate channel information for all DL channels.

[0101] A complete channel matrix can be represented as

[0102]

[0103] However, in techniques based on UL channel sounding for UEs with hardware constraints that limit the number of simultaneous transmissions that the UE can perform, the access node can only measure a subset of the multiple UL channels. In other words, the channel information for DL channels that are not related to the UL channels on which the UE transmits reference signals is lost. Therefore, the channel information for the DL channel can be represented as

[0104]

[0105] where s1, s2, … ∈ {1, …, N R}.

[0106] When determining the channel information for the DL channel based on CSI feedback, the following information is available

[0107] – Rank feedback [[ID=***]]

[0108] – PMI feedback and

[0109] – CQI feedback

[0110] Among them, the CQI feedback is the quantized signal-to-noise ratio (SNR) of each layer. In addition, in 3GPP LTE or 5G, the CQI feedback corresponds to a codeword, and this codeword can be mapped to multiple layers. Then, for these layers, the CQI values are the same. For example, in 3GPP LTE, when the rank is 4, γ1 = γ2, γ3 = γ4, and in 5G, γ1 = γ2 = γ3 = γ4.

[0111] According to an exemplary embodiment, there are provided methods and apparatuses for enhancing channel estimation by using UL channel sounding techniques and CSI feedback for determining DL channel information. Performing channel estimation by using UL channel sounding techniques and CSI feedback helps to solve the quantization errors existing in the channel information reported in the CSI feedback and the lost information of the DL channel that is not related to the UL channel of the UE transmitting reference signals.

[0112] In one embodiment, the channel information of the DL channel derived from the CSI feedback is projected onto the orthogonal subspace of the DL channel associated with the UL channel of the UE transmitting reference signals. Projecting the channel information of the DL channel derived from the CSI feedback onto the orthogonal subspace of the DL channel associated with the UL channel of the UE transmitting reference signals provides an estimation of the channel information of the DL channel that is not related to the UL channel of the UE transmitting reference signals. Orthogonal projection helps to reduce the quantization errors existing in the CSI feedback, which can degrade the performance.

[0113] In one embodiment, one or more principal eigenvectors of the orthogonal projection are selected as the channel information of the DL channel that is not associated with the UL channel of the UE transmitting reference signals.

[0114] In one embodiment, a combination of the channel information of the DL channel associated with the UL channel of the UE transmitting reference signals and one or more principal eigenvectors of the orthogonal projection is used as the channel matrix H. One or more principal eigenvectors of the orthogonal projection can be used as the lost channel information due to the UE being unable to transmit reference signals using all of its antennas.

[0115] For discussion purposes, consider a rank-2 link between the access node and the UE, but the UE has only one transmit RF chain and two antennas. Therefore, the UE can only transmit reference signals on one of its two antennas. This limitation may be due to hardware limitations, for example, a single transmit RF chain or no switching circuit. Therefore, the UE can only transmit on one antenna in a single transmission opportunity. The channel matrix H can be expressed as H = [h1, h2]. The channel information determined from the reference signals transmitted by the UE is expressed as Assume that there is no channel estimation error when determining the channel information based on the reference signals transmitted by the UE. Therefore, While the discussion focuses on the first of the two antennas being used by the UE to transmit reference signals, the exemplary embodiments can operate in the case where the second of the two antennas is used by the UE to transmit reference signals.

[0116] The CSI feedback reported by the UE is expressed as: rank and CQI γ1,γ2.

[0117] A better matrix W can be found to increase the sum rate, i.e.,

[0118] R = log2det(I + W H HH H W),

[0119] where R is the sum rate. In other words, W is a precoding matrix used to precode the transmission over the rank-2 link.

[0120] Ideally, W is a semi-unitary matrix generated by the singular value decomposition (SVD) of H. W can be expressed as

[0121]

[0122] Therefore, since is known (channel information determined from the reference signals transmitted by the UE), only w2 needs to be determined to find W.

[0123] Assume that w2 is in the orthogonal subspace of h1, is the estimated channel matrix determined according to the CSI feedback, and can be expressed as

[0124] and

[0125] The channel matrix from the CSI feedback, projected onto the orthogonal subspace of h1, can be expressed as

[0126] and

[0127] Then one or more principal eigenvectors of Z are selected as the solution of w2. Since the solution of w2 is found, the precoding matrix W is known. The transmission from the access node can be precoded using W and transmitted to the UE over the rank-2 link.

[0128] Typically, consider a rank-M link between an access node and a UE (where M is greater than or equal to 2), but the UE's transmit RF chains are not as many as the antennas. Therefore, the UE can only transmit reference signals on a subset of its antennas. The channel matrix H can be expressed as H = [H1 H2], where H1 is the channel information determined based on the reference signals transmitted by the UE and H2 is lost. Assume that there is no channel estimation error when determining the channel information based on the reference signals transmitted by the UE. Therefore,

[0129] The channel matrix derived from CSI feedback can be expressed as

[0130]

[0131] where,

[0132] Ideally, W is a semi-unitary matrix generated by the SVD of H. W can be expressed as

[0133] W = [W1 W2],

[0134] where W1 is the dominant eigenvector of H1. Therefore, since is known (the channel information determined from the reference signals transmitted by the UE), W1 is known, and only W2 needs to be determined to find W.

[0135] The channel matrix from CSI feedback, projected onto the orthogonal subspace of H1, can be expressed as

[0136] and

[0137] Then, one or more dominant eigenvectors of Z are selected as the solution for W2. Since the solution for W2 is found, the precoding matrix W is known. Transmissions from the access node can be precoded using W and transmitted to the UE over the rank-M link.

[0138] Alternatively, some dominant eigen-directions of H1 and W1 can be taken, and projected onto the orthogonal subspace of W1. Then, the eigenvalue decomposition of H1 can be expressed as

[0139]

[0140] and d1 dominant eigenvectors are selected from V1 to form W1.

[0141] The projection onto the orthogonal subspace is defined as

[0142]

[0143] Select d2 principal eigenvectors of Z as the solution of W2. The final solution of W is W = [W1 W2], as described above. The sizes d1 and d2 can be determined based on the eigenvalues of H1 and Z and the total rank of W.

[0144] Figure 4 A flowchart of an exemplary operation 400 performed in a node that communicates with, for example, other nodes or UEs is shown. The node uses a channel matrix generated based on channel information obtained from both CSI feedback and UL reference signals. Operation 400 may represent operations performed in the node when the node communicates using a channel matrix generated based on channel information obtained from both CSI feedback and UL reference signals. The node may be an access node that performs DL transmission or some other device that estimates the DL channel of an access node that performs DL transmission.

[0145] Operation 400 begins with the node estimating the DL channel based on reference signals received on the UL channel (e.g., reference signals received from a UE) (block 405). The node may measure the UL channel on which the reference signal (e.g., SRS) is transmitted and estimate the UL channel based on the channel measurement results. Using channel reciprocity, the node is able to estimate the DL channel corresponding to the UL channel. However, due to limitations of the UE (e.g., the UE cannot transmit on all UL channels), the node may not be able to estimate all DL channels. Therefore, the DL channel estimation result using the UL reference signal is referred to as partial channel measurement. For example, if the UE can only transmit on two UL channels simultaneously, the node can only estimate two DL channels corresponding to the two UL channels, regardless of the total number of DL channels between the node and the UE.

[0146] The node receives CSI feedback (block 407). The CSI feedback may be received from the UE. The CSI feedback may include rank indication, PMI feedback, and CQI feedback. The CSI feedback may be quantized to reduce communication overhead. The CSI feedback may optionally be filtered or denoised (block 409). The filtering or denoising of the CSI feedback may be performed by the node or other devices in the communication system. Filtering or denoising the CSI feedback may improve the quality of the DL channel estimation. The filtering or denoising of the CSI feedback may utilize channel statistics, such as conditional covariance or correlation, or the instantaneous fading channel. The filtering or denoising of the CSI feedback is discussed in detail below.

[0147] The node estimates the DL channel with orthogonal subspace projection (block 411). The node estimates the DL channel using CSI feedback and the UL reference signal-based DL channel estimate. The CSI feedback (after quantization) and the UL reference signal-based (essentially partial) DL channel estimate are used to approximate all DL channels. For example, the node projects the quantized DL channel estimate (from the CSI feedback) onto the orthogonal subspace of the DL channel estimated based on the UL reference signal to estimate the DL channel not estimated based on the UL reference signal.

[0148] The node communicates using the DL channel information (block 413). The DL channel information includes the DL channel estimate based on the UL reference signal (e.g., as obtained in block 405) and the DL channel estimate of the DL channel independent of the UL channel transmitting the reference signal. For example, the node precodes the data transmitted using the precoding procedure derived from the DL channel information and transmits the precoded data.

[0149] Figure 5 A flowchart of an exemplary operation 500 performed when the node estimates the DL channel information with orthogonal subspace projection is shown. Operation 500 may represent the operations performed at the node when the node estimates the DL channel information with orthogonal subspace projection. Operation 500 may be an exemplary implementation of block 411 of FIG. 4 for generating a partial precoding procedure based on the UL reference signal-derived channel information.

[0150] The node uses (block 505). As described above, W is a semi-unitary matrix generated by the SVD of H, and W can be expressed as W = [W1 W2], where W1 is the principal eigenvector of H1, and these eigenvectors are determined by the UL reference signal transmitted by the UE and are known.

[0151] The node projects the channel information provided by the CSI feedback onto the orthogonal subspace of H1 (block 507). Projecting the channel information provided by the CSI feedback (which is complete but may include errors due to the quantization process for reducing communication overhead) onto the orthogonal subspace helps to mitigate the impact of quantization errors. The projection onto the orthogonal subspace can be expressed as where, The principal eigenvector of node identifier Z (block 509). Identify the principal eigenvector of Z and use it as the solution for the remainder of the precoding procedure W. In other words, the principal eigenvector of Z is used as the solution for W2. In one embodiment, the number of identified principal eigenvectors of Z is equal to the number of DL channels lacking channel information estimation (i.e., the number of columns of W2). For example, if there are a total of three DL channels and the UE can only transmit using one transmit RF chain, then there are two DL channels lacking channel information estimation (since only one DL channel has channel information derived for the UL reference signal). In this case, two principal eigenvectors of Z are identified and used as the solution for W2.

[0152] In one embodiment, the principal eigenvector direction of the precoding procedure and the DL channel information derived from the UL reference signal are used instead of the actual precoding procedure and DL channel information in the orthogonal subspace projection. Utilizing the principal eigenvector direction may help simplify the orthogonal subspace projection.

[0153] Figure 6 A flowchart of an exemplary operation 600 performed at a node for determining a precoding procedure for a DL channel based on estimated channel information is shown. Operation 600 may represent the operations performed at the node when the node determines a precoding procedure for a DL channel based on estimated channel information. Operation 600 may be Figure 4 an exemplary implementation of block 411.

[0154] Operation 600 begins with the node performing an eigen - decomposition H1 of the channel information of the DL channels derived from the UL reference signal (block 605). The eigen - decomposition of H1 can be expressed as The node selects d1 principal eigenvectors from V1 and forms W1, where d1 is the number of DL channels derived from the UL reference signal (block 607).

[0155] The node projects the channel information provided by the CSI feedback onto the orthogonal subspace of W1 (block 609). Projecting the channel information provided by the CSI feedback (which is complete but may include errors due to the quantization process used to reduce communication overhead) onto the orthogonal subspace helps mitigate the impact of quantization errors. The projection onto the orthogonal subspace can be expressed as The node selects the principal eigenvectors of Z as the solution for W2 (block 611). In one embodiment, the node selects d2 principal eigenvectors as the solution for W2, where d2 is the number of DL channels not associated with the UL reference signal. The precoding procedure for the DL channel is determined as W = [W1 W2] (block 613).

[0156] In one embodiment, CSI feedback can be processed to improve the quality of DL channel estimation. Processing of CSI feedback includes filtering or denoising using channel statistics such as channel covariance or correlation, or the instantaneous fading channel. For example, the impact of channel covariance or correlation on the sum rate at the transmitter can be expressed as eigenvalue decomposition

[0157] R T =UΛU H 。

[0158] For the instantaneous fading channel, the channel information can be expressed as

[0159]

[0160] Under the assumption of no receive correlation, C is an i.i.d. random matrix, and the channel information can be re-expressed as

[0161] H=UΛ 1 / 2 C。

[0162] The same correlation can be applied to any channel vector on any receive antenna on the UE side. Additionally, for most cases, R is not a full-rank matrix, Λ includes non-zero eigenvalues, and then U is a semi-unitary matrix. In one embodiment, the dominant eigenvalues can be selected.

[0163] Figure 7 An exemplary node 700 is shown highlighting circuitry to enhance channel estimation using UL channel sounding-based techniques and CSI feedback for determining DL channel information, where the node 700 improves the quality of DL channel estimation by processing CSI feedback. Examples of processing include filtering, denoising, or filtering and denoising.

[0164] Node 700 includes a CSI feedback processing unit 705 for processing the CSI feedback received by node 700. For example, the CSI feedback processing unit 705 filters, denoises, or filters and denoises the CSI feedback received by node 700. The processed CSI feedback, denoted as is a refined version of the channel information reported in the CSI feedback, which is quantized and may include quantization error. Due to the processing performed by the CSI feedback processing unit 705, the processed CSI feedback generally has a smaller quantization error (compared to the unprocessed CSI feedback). For example, the CSI feedback processing unit 705 can average, filter, or smooth the channel information reported in the CSI feedback.

[0165] Node 700 further includes a projection unit 707 for projecting the channel information (e.g., the processed CSI feedback )(e.g., inFigure 5 in the frame 507 or Figure 6 in the frame 609). The node 700 also includes a selection unit 709 for selecting one or more principal eigenvectors of the projected Z provided by the projection unit 707. The number of principal eigenvectors selected by the selection unit 709 may depend on the number of DL channels not associated with the UL channel transmitting the UL reference signal. In other words, the selection unit 709 selects the same number of principal eigenvectors as the number of DL channels for which the channel information is not derived from the UL reference signal.

[0166] Example processes performed by the CSI feedback processing unit 705 include, but are not limited to:

[0167] – Minimum mean square error (MMSE) filtering;

[0168] – Denoising by projecting onto the dominant subspace;

[0169] – Filtering based on maximum fairness (MF) / maximum ratio combining (MRC);

[0170] – Quadratic programming.

[0171] In relation to MMSE or linear MMSE (LMMSE) filtering, using CSI feedback, the signal model of the feedback PMI with quantization noise can be expressed as

[0172]

[0173] After MMSE or LMMSE filtering, the fading channel on the receiver side (denoted as ) and the processed CSI feedback (denoted as ) can be expressed as

[0174] and

[0175]

[0176] The projection onto the orthogonal subspace of the channel information derived from the UL reference signal H1 determined by the projection unit 707 can be expressed, for example, as

[0177]

[0178] The selection unit 709 selects one or more principal eigenvectors of Z as the solution of W2.

[0179] In relation to denoising by projection onto the principal subspace, the projection of the principal eigenspace of the covariance (e.g., performed by the CSI feedback processing unit 705) can be expressed as

[0180]

[0181] The CSI feedback can be denoised through projection (similarly, for example, by the CSI feedback processing unit 705), generating a processed CSI feedback

[0182]

[0183] The projection unit 707 projects the processed CSI feedback onto the orthogonal subspace of the channel information derived from the UL reference signal H1 For example, the CSI feedback can be expressed as

[0184]

[0185] The selection unit 709 selects one or more principal eigenvectors of Z as the solution of W2

[0186] Related to MF / MRC-based filtering, the quantization model can be expressed as

[0187]

[0188] The CSI feedback processing unit 705 can denoise the CSI feedback using an MRC filter, and the fading channel on the receiver side (denoted as ) and the processed CSI feedback (denoted as ) can be expressed as

[0189] And

[0190]

[0191] The projection unit 707 projects the processed CSI feedback onto the orthogonal subspace of the channel information derived from the UL reference signal H1 For example, the CSI feedback can be expressed as

[0192]

[0193] The selection unit 709 selects one or more principal eigenvectors of Z as the solution of W2

[0194] Related to quadratic programming, high-rank channel reconstruction can be performed. The channel reconstruction can be expressed as

[0195]

[0196] Assume that C' is semi-unitary, i.e., C' is

[0197] the principal eigenvector of

[0198] Then the channel can be reconstructed as

[0199]

[0200] The projection unit 707 projects the processed CSI feedback onto the orthogonal subspace of the channel information derived from the UL reference signal H1. For example, the CSI feedback can be expressed as

[0201]

[0202] The selection unit 709 selects one or more principal eigenvectors of Z as the solution of W2.

[0203] Some or all of the CSI feedback processing unit 705, the projection unit 707, and the selection unit 709 can be implemented in software executed in one or more processors, one or more integrated circuits (e.g., a field programmable logic array (FPGA) or an application-specific integrated circuit (ASIC)), etc. In performance-critical implementations, computationally intensive parts can be implemented as integrated circuits operating in cooperation with software executed on one or more processors.

[0204] In one embodiment, the CSI feedback can be processed through conditional covariance to improve the quality of DL channel estimation. In the presence of CSI feedback, the channel and the feedback PMI are not independent. Then, for a set of channels with a given feedback the following conditional covariance (conditioned on the feedback PMI) can be considered

[0205]

[0206]

[0207] The eigenvalue decomposition of can be expressed as

[0208]

[0209] Figure 8 An exemplary node 800 is shown highlighting circuitry to enhance channel estimation using UL channel sounding-based techniques and CSI feedback for determining DL channel information, where the node 800 improves the quality of DL channel estimation by processing the CSI feedback with conditional covariance. Examples of the processing include filtering, denoising, or filtering and denoising.

[0210] The node 800 includes a covariance unit 805 for determining the feedback PMI (denoted as ) conditional covariance. For example, CSI feedback defines a subset of channels that are measured to derive the feedback of the CSI feedback report. Thus, the channel statistics can be the covariance conditioned on the CSI feedback (in particular, the PMI feedback part of the CSI feedback).

[0211] Node 800 also includes a CSI feedback processing unit 807 for processing the CSI feedback received by node 800. For example, the CSI feedback processing unit 807 filters, denoises, or filters and denoises the CSI feedback received by node 800 in combination with the conditional covariance provided by the covariance unit 805. The processed CSI feedback, denoted as is a refined version of the channel information reported in the CSI feedback, which is quantized and may include quantization errors. Due to the processing performed by the CSI feedback processing unit 807, the processed CSI feedback usually has a smaller quantization error. The CSI feedback processing unit 807 can, for example, average or smooth the channel information reported in the CSI feedback.

[0212] Node 800 also includes a projection unit 809 for projecting the channel information (e.g., the processed CSI feedback )(e.g., in Figure 5 box 507 or Figure 6 box 609) on the orthogonal subspace of the channel information derived from the UL reference signal H1. Node 800 also includes a selection unit 811 for selecting one or more principal eigenvectors of the projection Z provided by the projection unit 809. The number of principal eigenvectors selected by the selection unit 811 can depend on the number of DL channels not associated with the UL channel transmitting the UL reference signal. In other words, the selection unit 811 selects a number of principal eigenvectors equal to the number of DL channels that do not have channel information derived from the UL reference signal.

[0213] Example processing performed by the CSI feedback processing unit 807 includes but is not limited to:

[0214] – Minimum Mean Square Error (MMSE) filtering through the conditional covariance;

[0215] – Denoising through principal subspace projection with the conditional covariance;

[0216] – Maximum fairness (MF) / Maximum Ratio Combining (MRC) based filtering through the conditional covariance;

[0217] – Quadratic programming through the conditional covariance.

[0218] When the conditional covariance is used to further enhance the feedback, for the above by Figure 8The discussion of the exemplary processing performed by the CSI feedback processing unit 807 also applies. However, the general covariance matrix presented previously is replaced by a conditional matrix.

[0219] Some or all of the covariance unit 805, the CSI feedback processing unit 807, the projection unit 809, and the selection unit 811 may be implemented in software executed in one or more processors, one or more integrated circuits (e.g., FPGA or ASIC), etc. In performance-critical implementations, computationally intensive portions may be implemented as integrated circuits operating in cooperation with software executed on one or more processors.

[0220] In relation to MMSE or linear MMSE (LMMSE) filtering, the processed CSI feedback can be expressed as

[0221]

[0222] In relation to denoising by projection onto the principal subspace, the projection onto the orthogonal subspace can be expressed as

[0223]

[0224]

[0225] In relation to MF / MRC-based filtering, the processed CSI feedback can be expressed as

[0226]

[0227] In relation to quadratic programming, the principal eigenvector can be expressed as

[0228] the eigenvector of

[0229] and

[0230]

[0231] The performance evaluation of the example embodiments is performed using the following model:

[0232] – Access node: 8 transmit antennas;

[0233] – UE: 2 receive antennas;

[0234] – Channel model:

[0235] Scattering channel model,

[0236] Angle spread of 15 degrees, random center, uniform power distribution along the ring,

[0237] Access node - cross-polarized antennas, cross-polarization correlation γ = 0.3,

[0238] UE - antenna correlation 0;

[0239] – Conditional covariance generation: Monte Carlo;

[0240] – Measurement:

[0241] Rate R = log2det(I + W H HH H W):

[0242] – Given channel covariance R and codebook C, obtain the conditional covariance as:

[0243] Eigen - decomposition of R and non - zero eigen - modes

[0244] R = UΛU H ,

[0245] Generate N samples of {H} according to the following equation

[0246] H = UΛ 1 / 2 C,

[0247] where c in C ij is i.i.d. complex Gaussian with unit variance,

[0248] For each H, search for the PMI in codebook C based on a certain criterion, such as MIMO capacity.

[0249] Put the channel H into the set using the obtained matrix W.

[0250] For each set Obtain the conditional covariance by averaging over

[0251] Figure 9A Figure 9A Figure 900 shows a data graph of the downlink SNR and the sum - rate R, comparing the example technology with the channel information having only CSI feedback. As Figure 9A shown, the sum - rate R of only CSI feedback (line 905) and the rank - 1 channel (line 910) is lower than the example technology for most downlink SNR values. Most example technologies perform similar operations and lie within a region (e.g., region 950) for any given downlink SNR.

[0252] Figure 9B Figure 900 shows a detailed view of region 950. Figure 9B Region 950 is shown in more detail in Figure 9BAs shown, compared with the embodiment techniques that do not utilize conditional covariance, the embodiment techniques with conditional covariance processing provide a higher sum rate. R (e.g., the protruding portion 955).

[0253] Figure 10 An exemplary communication system 1000 is shown. Generally, the system 1000 enables multiple wireless or wired users to send and receive data and other content. The system 1000 may implement one or more channel access methods, such as code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), single-carrier FDMA (SC-FDMA), or non-orthogonal multiple access (NOMA).

[0254] In this example, the communication system 1000 includes electronic devices (EDs) 1010a to 1010c, radio access networks (RANs) 1020a to 1020b, a core network 1030, a public switched telephone network (PSTN) 1040, the Internet 1050, and other networks 1060. Although Figure 10 a certain number of these components or elements are shown, any number of these components or elements may be included in the system 1000.

[0255] The EDs 1010a to 1010c are used to operate or communicate in the system 1000. For example, the EDs 1010a to 1010c are used to send or receive signals via wireless or wired communication channels. Each of the EDs 1010a to 1010c represents any suitable end-user device and may include devices such as (or may be referred to as): user equipment (UE), wireless transmit / receive unit (WTRU), mobile station, fixed or mobile subscriber unit, cellular phone, personal digital assistant (PDA), smartphone, laptop computer, computer, touchpad, wireless sensor, or consumer electronic device.

[0256] Here, RANs 1020a to 1020b respectively include base stations 1070a to 1070b. Each of the base stations 1070a to 1070b is used for wireless connection with one or more of the EDs 1010a to 1010c so as to be able to access the core network 1030, PSTN 1040, Internet 1050, and / or other networks 1060. For example, the base stations 1070a to 1070b may include (or be) one or more of several well-known devices such as a base transceiver station (BTS), Node-B, evolved NodeB (eNodeB), next generation (NG) NodeB (gNB), home NodeB, home eNodeB, site controller, access point (AP), or wireless router. The EDs 1010a to 1010c are used for connection and communication with the Internet 1050 and may access the core network 1030, PSTN 1040, or other networks 1060.

[0257] In Figure 10 In the illustrated embodiment, base station 1070a forms part of RAN 1020a, and RAN 1020a may include other base stations, elements, and / or devices. In addition, base station 1070b forms part of RAN 1020b, and RAN 1020b may include other base stations, elements, or devices. The base stations 1070a to 1070b operate respectively to transmit or receive wireless signals within a specific geographical area or region (sometimes referred to as a "cell"). In some embodiments, multiple-input multiple-output (MIMO) technology may be employed, and each cell has multiple transceivers.

[0258] The base stations 1070a to 1070b communicate with one or more of the EDs 1010a to 1010c using a wireless communication link via one or more air interfaces 1090. The air interface 1090 may utilize any suitable radio access technology.

[0259] It is envisioned that system 1000 may use a multi-channel access function, including the schemes described above. In a particular embodiment, the base stations and EDs implement 5G new radio (NR), LTE, LTE-A, or LTE-B. Of course, other multiple access schemes and wireless protocols may also be utilized.

[0260] RANs 1020a to 1020b communicate with the core network 1030 to provide voice, data, applications, Voice over Internet Protocol (VoIP), or other services to the EDs 1010a to 1010c. It should be understood that the RANs 1020a to 1020b and / or the core network 1030 may communicate directly or indirectly with one or more other RANs (not shown). The core network 1030 may also be used as a gateway access to other networks such as the PSTN 1040, the Internet 1050, and other networks 1060. Additionally, some or all of the EDs 1010a to 1010c may include functionality to communicate with different wireless networks over different wireless links using different wireless technologies and / or protocols. The EDs may communicate with a service provider or a switch (not shown) and the Internet 1050 via a wired communication channel instead of (or in addition to) wireless communication.

[0261] Although Figure 10 an example of a communication system is shown, various changes may be made to Figure 10 it. For example, the communication system 1000 may include any number of EDs, base stations, networks, or other components in any suitable configuration.

[0262] Figure 11A and Figure 11B illustrate exemplary devices in which the methods and teachings provided by the present disclosure may be implemented. In particular, Figure 11A an exemplary ED 1110 is shown, Figure 11B and an exemplary base station 1170 is shown. These components may be used in the system 1000 or any other suitable system.

[0263] As Figure 11A shown, the ED 1110 includes at least one processing unit 1100. The processing unit 1100 implements various processing operations of the ED 1110. For example, the processing unit 1100 may perform signal encoding, data processing, power control, input / output processing, or any other function that enables the ED 1110 to operate in the system 1000. The processing unit 1100 also supports the methods and teachings described in more detail above. Each processing unit 1100 includes any suitable processing or computing device for performing one or more operations. Each processing unit 1100 may include, for example, a microprocessor, a microcontroller, a digital signal processor, a field programmable gate array, or an application specific integrated circuit.

[0264] ED 1110 also includes at least one transceiver 1102. The transceiver 1102 is used to modulate data or other content for transmission via at least one antenna or a Network Interface Controller (NIC) 1104. The transceiver 1102 is also used to demodulate data or other content received by at least one antenna 1104. Each transceiver 1102 includes any suitable structure for generating signals for wireless or wired transmission or for processing signals received wirelessly or via a wired connection. Each antenna 1104 includes any suitable structure for transmitting or receiving wireless or wired signals. One or more transceivers 1102 may be used for ED 1110, and one or more antennas 1104 may be used for ED 1110. Although the transceiver 1102 is shown as a single functional unit, at least one transmitter and at least one separate receiver may also be used.

[0265] ED 1110 also includes one or more input / output devices 1106 or interfaces (e.g., a wired interface to the Internet 1050). The input / output devices 1106 facilitate interaction with users or other devices in the network (network communication). Each input / output device 1106 includes any suitable structure for providing information to the user or receiving information from the user, such as a speaker, a microphone, a keypad, a keyboard, a display, or a touch screen, including network interface communication.

[0266] In addition, ED 1110 includes at least one memory 1108. The memory 1108 stores instructions and data used, generated, or collected by ED 1110. For example, the memory 1108 may store software instructions or firmware instructions executed by the processing unit 1100 and data for reducing or eliminating interference in incoming signals. Each memory 1108 includes any suitable volatile or non-volatile storage and retrieval device. Any suitable type of memory may be used, e.g., random access memory (RAM), read only memory (ROM), hard disk, optical disk, subscriber identity module (SIM) card, memory stick, secure digital (SD) memory card, etc.

[0267] As Figure 11BAs shown, base station 1170 includes at least one processing unit 1150, at least one transceiver 1152 (including the functions of a transmitter and a receiver), one or more antennas 1156, at least one memory 1158, and one or more input / output devices 1166 or interfaces. A scheduler understood by those skilled in the art is coupled to the processing unit 1150. The scheduler may be included within the base station 1170 or operate independently of the base station 1170. The processing unit 1150 implements various processing operations of the base station 1170, such as signal encoding, data processing, power control, input / output processing, or any other function. The processing unit 1150 may also support the methods and teachings described in more detail above. Each processing unit 1150 includes any suitable processing or computing device for performing one or more operations. For example, each processing unit 1150 may include a microprocessor, a microcontroller, a digital signal processor, a field programmable gate array, or an application specific integrated circuit, etc.

[0268] Each transceiver 1152 includes any suitable structure for generating signals for wireless or wired transmission to one or more EDs or other devices. Each transceiver 1152 also includes any suitable structure for processing signals received from one or more EDs or other devices via wireless or wired means. Although the transmitter and the receiver are shown in combination as the transceiver 1152, the transmitter and the receiver may be separate components. Each antenna 1156 includes any suitable structure for transmitting or receiving wireless or wired signals. Although the common antenna 1156 shown here is coupled to the transceiver 1152, if configured as separate components, one or more antennas 1156 may be coupled to the transceiver 1152, allowing the separate antennas 1156 to be coupled to the transmitter and the receiver. Each memory 1158 includes any suitable volatile or non-volatile storage and retrieval device. Each input / output device 1166 facilitates interaction with users or other devices in the network (network communication). Each input / output device 1166 includes any suitable structure for providing information to the user or receiving information from the user (including network interface communication).

[0269] Figure 12is a block diagram of a computing system 1200 that can be used to implement the devices and methods disclosed herein. For example, the computing system can be any entity in a UE, access network (AN), mobility management (MM), session management (SM), user plane gateway (UPGW), or access stratum (AS). A particular device may utilize all of the components shown or only a subset of these components, and the degree of integration between devices may vary. Additionally, a device can include multiple instances of components, e.g., multiple processing units, processors, memories, transmitters, receivers. The computing system 1200 includes a processing unit 1202. The processing unit includes a central processing unit (CPU) 1214, a memory 1208, and may also include a mass storage device 1204, a video adapter 1210, and an I / O interface 1212 connected to a bus 1220.

[0270] The bus 1220 can be one or more of several types of bus architectures, including a memory bus or memory controller, a peripheral bus, or a video bus. The CPU 1214 can include any type of electronic data processor. The memory 1208 can include any type of non-transitory system memory, e.g., static random access memory (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), read-only memory (ROM), or a combination thereof. In one embodiment, the memory 1208 can include ROM used at power-on and DRAM for program and data storage during program execution.

[0271] The mass storage 1204 can include any type of non-transitory storage device for storing data, programs, and other information and making these data, programs, and other information accessible via the bus 1220. The mass storage 1204 can include, for example, one or more of a solid state drive, a hard disk drive, a disk drive, or an optical drive.

[0272] Video adapter 1210 and I / O interface 1212 provide interfaces to couple external input and output devices to processing unit 1202. As shown, examples of input and output devices include display 1218 coupled to video adapter 1210 and mouse / keyboard / printer 1216 coupled to I / O interface 1212. Other devices may be coupled to processing unit 1202 and may utilize additional or fewer interface cards. For example, a serial interface such as Universal Serial Bus (USB) (not shown) may be used to provide an interface to external devices.

[0273] Processing unit 1202 also includes one or more network interfaces 1206, which may include a wired link such as an Ethernet cable or a wireless link for accessing nodes or different networks. Network interface 1206 allows processing unit 1202 to communicate with remote units over a network. For example, network interface 1206 may provide wireless communication via one or more transmitters / transmitting antennas and one or more receivers / receiving antennas. In one embodiment, processing unit 1202 is coupled to a local area network 1222 or a wide area network for processing data and communicating with remote devices such as other processing units, the Internet, or remote storage facilities.

[0274] It should be understood that one or more steps of the embodiment methods provided herein may be performed by corresponding units or modules. For example, a signal may be transmitted by a transmitting unit or module. A signal may be received by a receiving unit or module. A signal may be processed by a processing unit or module. Other steps may be performed by the following: an estimating unit or module, a projecting unit or module, a selecting unit or module, a combining unit or module, a precoding unit or module, a processing unit or module, a filtering unit or module, a denoising unit or module, a determining and decomposing unit or module, or a communication unit or module. Each unit or module may be hardware, software, or a combination thereof. For example, one or more of these units or modules may be integrated circuits, such as a field programmable gate array (FPGA) or an application-specific integrated circuit (ASIC).

[0275] Although the present disclosure and its advantages have been described in detail, it should be understood that various changes, substitutions, and modifications may be made herein without departing from the scope of the disclosure defined by the appended claims.

Claims

1. A method implemented by a node, characterized in that, The method includes: The node estimates first DL channel information of a first downlink (DL) channel based on a reference signal received on a first uplink (UL) channel, where the first UL channel is a subset of available UL channels and the first DL channel corresponds to the first UL channel; The node receives channel feedback of a set of DL channels; The node estimates second DL channel information based on a projection of the first DL channel information and the channel feedback; and The node communicates based on the second DL channel information.

2. The method according to claim 1, wherein The second DL channel information includes information derived from the first DL channel information and an orthogonal projection of the channel feedback.

3. The method according to claim 1 or 2, characterized in that, The estimating the second DL channel information includes: The node projects channel information derived from the channel feedback onto an orthogonal subspace of the first DL channel to generate the projection; The node selects one or more principal eigenvectors of the projection; and The node combines the first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

4. The method according to claim 1 or 2, characterized in that, The communicating based on the second DL channel information includes: The node precodes data based on the second DL channel information.

5. The method according to claim 1 or 2, characterized in that It further includes: After receiving the channel feedback, the node processes the channel feedback to generate processed channel feedback.

6. The method according to claim 5, characterized in that The estimating the second DL channel information includes: The node projects channel information derived from the processed channel feedback onto an orthogonal subspace of the first DL channel to generate the projection; The node selects one or more principal eigenvectors of the projection; and The node combines the first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

7. The method according to claim 5, characterized in that The processing the channel feedback includes: filtering or denoising the channel feedback using channel statistical information.

8. The method according to claim 7, characterized in that The filtering or denoising the channel feedback includes at least one of the following: performing minimum mean square error (MMSE) filtering on the channel feedback, projecting a principal subspace of the channel feedback, performing filtering on the channel feedback based on maximum fairness (MF) or maximum ratio combining (MRC), and performing quadratic programming on the channel feedback.

9. The method according to claim 5, characterized in that, The processing the channel feedback includes: The node determines a conditional covariance of the channel feedback; and The node filters or denoises the channel feedback based on the conditional covariance.

10. The method according to claim 9, wherein The filtering or denoising the channel feedback includes at least one of the following: performing MMSE filtering on the channel feedback, projecting a principal subspace of the channel feedback, performing filtering on the channel feedback based on MF or MRC, and performing quadratic programming on the channel feedback.

11. The method according to claim 1 or 2, characterized in that, The second DL channel information includes a DL precoding procedure.

12. A method implemented by a node, characterized in that, The method includes: The node estimates first DL channel information of a first downlink (DL) channel based on a reference signal received on a first uplink (UL) channel, where the first UL channel is a subset of available UL channels and the first DL channel is associated with the first UL channel; The node receives channel feedback of a DL channel set; The node performs eigen - decomposition on the first DL channel information to generate an eigen - decomposition of the first DL channel information; The node selects one or more first principal eigen - vectors of the eigen - decomposition of the first DL channel information as a partial precoding procedure; The node projects the channel information derived from the channel feedback onto an orthogonal subspace of the first DL channel; The node selects one or more second principal eigen - vectors of the channel information after the projection of the channel information as second DL channel information of a second DL channel, where the second DL channel is not associated with the first UL channel; and The node combines the partial precoding procedure and the one or more second principal eigen - vectors to form a precoding procedure.

13. The method according to claim 12, characterized in that, The DL channel set includes the first DL channel and the second DL channel.

14. The method according to claim 12 or 13, characterized in that, The reference signal includes a sounding reference signal SRS.

15. The method according to claim 12 or 13, characterized in that Further included is: After receiving the channel feedback, the node processes the channel feedback.

16. The method according to claim 15, wherein The processing of the channel feedback includes: filtering or denoising the channel feedback using channel statistical information.

17. The method according to claim 16, wherein Further included is: The node determines the conditional covariance of the channel feedback, where the filtering or denoising of the channel feedback is performed according to the conditional covariance.

18. The method according to claim 16, wherein The filtering or denoising of the channel feedback includes at least one of the following: performing minimum mean - square error MMSE filtering on the channel feedback, projecting the principal subspace of the channel feedback, performing filtering based on maximum fairness MF or maximum ratio combining MRC on the channel feedback, and performing quadratic programming on the channel feedback.

19. A node, characterized in that, Included is: A non - transitory memory storage including instructions; and One or more processors communicating with the memory storage, where the one or more processors execute the instructions to cause the node to perform the following operations: Estimate first DL channel information of a first downlink DL channel according to a reference signal received on a first uplink UL channel, where the first UL channel is a subset of the UL channels and the first DL channel corresponds to the first UL channel; Receive channel feedback of a DL channel set; Estimate second DL channel information according to the first DL channel information and the projection of the channel feedback; and Communicate according to the second DL channel information.

20. The node according to claim 19, characterized in that, The one or more processors further execute the instructions to cause the node to project the channel information derived from the channel feedback onto an orthogonal subspace of the first DL channel, thereby generating the projection; select one or more principal eigen - vectors of the projection; and combine the first DL channel information with the one or more principal eigen - vectors to form the second DL channel information.

21. The node according to claim 19 or 20, characterized in that, The one or more processors further execute the instructions to cause the node to precode data according to the second DL channel information.

22. The node according to claim 19 or 20, characterized in that, The one or more processors further execute the instructions to cause the node to process the channel feedback after receiving the channel feedback, thereby generating a processed channel feedback.

23. The node according to claim 22, wherein The one or more processors further execute the instructions to cause the node to project channel information derived from the processed channel feedback onto an orthogonal subspace of the first DL channel, thereby generating the projection; select one or more principal eigenvectors of the projection; and combine the first DL channel information with the one or more principal eigenvectors to form the second DL channel information.

24. The node according to claim 22, wherein The one or more processors further execute the instructions to cause the node to filter or denoise the channel feedback using channel statistics.

25. The node according to claim 24, characterized in that, Filter or denoise the channel feedback by at least one of the following: performing minimum mean square error (MMSE) filtering on the channel feedback, projecting the principal subspace of the channel feedback, performing filtering based on maximum fairness (MF) or maximum ratio combining (MRC) on the channel feedback, and performing quadratic programming on the channel feedback.

26. The node according to claim 22, characterized in that, The one or more processors further execute the instructions to cause the node to determine the conditional covariance of the channel feedback; and the node filters or denoises the channel feedback according to the conditional covariance.

27. The node according to claim 26, characterized in that, Filter or denoise the channel feedback by at least one of the following: performing MMSE filtering on the channel feedback, projecting the principal subspace of the channel feedback, performing filtering based on MF or MRC on the channel feedback, and performing quadratic programming on the channel feedback.

28. The node according to claim 19 or 20, characterized in that The second DL channel information includes a DL precoding procedure.

29. A node, characterized in that, Comprising: A non-transitory memory storage including instructions; and One or more processors in communication with the memory storage, wherein the one or more processors execute the instructions to cause the node to perform the following operations: Estimate first DL channel information of a first downlink (DL) channel according to a reference signal received on a first uplink (UL) channel, wherein the first UL channel is a subset of available UL channels, and the first DL channel is associated with the first UL channel; Receive channel feedback of a DL channel set; Perform eigen decomposition on the first DL channel information to generate an eigen decomposition of the first DL channel information; Select one or more first principal eigenvectors of the eigen decomposition of the first DL channel information as a partial precoding procedure; Project channel information derived from the channel feedback onto an orthogonal subspace of the first DL channel; Select one or more second principal eigenvectors of the channel information after the projection of the channel information as second DL channel information of a second DL channel, wherein the second DL channel is not associated with the first UL channel; and 30. The node according to claim 29, wherein, Combine the partial precoding procedure and the one or more second principal eigenvectors to form a precoding procedure.

31. The node according to claim 29 or 30, characterized in that, The DL channel set includes the first DL channel and the second DL channel.

32. The node according to claim 29 or 30, characterized in that, The reference signal includes a sounding reference signal (SRS). The one or more processors further execute the instructions to cause the node to process the channel feedback after receiving the channel feedback.

33. The node according to claim 32, wherein The one or more processors further execute the instructions to cause the node to determine a conditional covariance of the channel feedback, wherein the channel feedback is filtered or denoised based on the conditional covariance.

34. The node according to claim 32, characterized in that, Filtering or denoising the channel feedback by at least one of the following: performing minimum mean square error (MMSE) filtering on the channel feedback, projecting a principal subspace of the channel feedback, filtering the channel feedback based on maximum fairness (MF) or maximum ratio combining (MRC), and performing quadratic programming on the channel feedback.

Images