TW201707394A - Method and apparatus for hybrid beamforming - Google Patents

Abstract

Embodiments of the present disclosure provide a method and apparatus for hybrid beamforming in a base station as well as a method and apparatus for hybrid beamforming in a mobile terminal. The method for hybrid beamforming in a base station comprising: calculating a wideband analog beamforming matrix based on long-term-level estimation of a physical channel; quantizing the wideband analog beamforming matrix to obtain a quantized wideband analog beamforming matrix; applying the quantized wideband analog beamforming matrix to the physical channel to obtain an effective channel of the physical channel; calculating a sub-band digital beamforming matrix based on short-term-level estimation of the effective channel; and performing hybrid beamforming on downlink signals with the sub-band digital beamforming matrix and the quantized wideband analog beamforming matrix. The hybrid beamforming solution according to the embodiments of the present disclosure makes massive MIMO systems more practical and cost-effective to deploy.

Description

Hybrid beamforming method and device

Embodiments of the present invention generally relate to wireless communication technologies and, more particularly, to hybrid beamforming methods and apparatus.

Large-scale multiple-input multiple-output (MIMO) or large-scale antenna systems have been widely recognized as key implementation technologies for 5G communication systems due to their enormous advantages in energy efficiency, spectrum efficiency, robustness, and reliability. The basic premise of massive MIMO is that the number of base station antennas is much larger than the number of single antenna terminals. In theory, a massive MIMO system with full digital beamforming (BF) can produce optimal performance. However, the price paid is hardware complexity and cost (number of RF channels), as well as the complexity of signal processing and the rapid increase in energy consumption. Therefore, when the number of antennas is very large, it may not be feasible to implement the same number of radio frequency channels. Therefore, how to implement massive MIMO with a limited number of RF channels is a key technical issue.

In order to reduce the number of required RF channels, dynamic antenna switching techniques are an option, but because of the inability to take full advantage of the extra antennas, this scheme provides limited array gain and poor performance in the associated channels. In order to solve this problem, consider using analog waves of active antennas. Beam shaping. In general, analog beamforming controls the phase of the signal on each antenna through the network of analog phase shifters. The performance of analog beamforming is usually second best due to hardware limitations in amplitude and phase control. In order to realize multi-data flow precoding with larger beamforming gain, a hybrid analog and digital beamforming (hereinafter referred to as hybrid beamforming) strategy has recently been proposed. However, at present, hybrid beamforming algorithms are still used, for example, in flexible user scheduling, training signal design, and channel information acquisition for FDD (frequency division multiplexing) systems or TDD systems with or without accurate antenna calibration. Far from being mature.

Embodiments of the present invention provide a method and apparatus for hybrid beamforming in a base station and a method and apparatus for hybrid beamforming in a mobile terminal to address or at least partially alleviate the above problems in the prior art .

In a first aspect, embodiments of the present invention provide a method of hybrid beamforming in a base station. The method includes: calculating a broadband analog beamforming matrix based on long-term estimation of a physical channel; quantizing the broadband analog beamforming matrix to obtain a quantized broadband analog beamforming matrix; to the physical channel Applying the quantized broadband analog beamforming matrix to obtain an equivalent channel of the physical channel; calculating a subband digital beamforming matrix based on short-term estimation of the equivalent channel; and utilizing the sub- A hybrid beamforming is performed on the downlink signal with a digital beamforming matrix and the quantized broadband analog beamforming matrix.

In one embodiment, quantifying the broadband analog beamforming matrix comprises: normalizing the amplitude of each non-zero element in the broadband analog beamforming matrix; and for each of the non-zero elements A phase search is performed element by element in a predetermined set of phases to select a phase that maximizes the capacity of the equivalent channel.

In one embodiment, the method further comprises performing sub-band user scheduling based on a short-term estimate of the equivalent channel.

In one embodiment, the method further comprises: transmitting, by the horizontal linear array and the vertical linear array of the base station, a first training signal and a second training signal to the mobile terminal, respectively, so that the mobile terminal is respectively based on the a first training signal and the second training signal to estimate a horizontal subchannel covariance matrix associated with the horizontal linear array and a vertical subchannel covariance matrix associated with the vertical linear array; The mobile terminal receives an estimated feedback of the horizontal subchannel covariance matrix and an estimated feedback of the vertical subchannel covariance matrix; and constructs the horizontal subchannel covariance based on the feedback The Kroneck product of the matrix and the vertical subchannel covariance matrix is used as the broadband channel covariance matrix of the physical channel.

In a second aspect, embodiments of the present invention provide a method of hybrid beamforming in a base station. The method includes: calculating a broadband analog beamforming matrix based on long-term estimation of a physical channel; applying the broadband analog beamforming matrix to the physical channel to obtain an equivalent channel of the physical channel; Short-term estimation of the equivalent channel, calculating the sub-band digital beamforming matrix and performing sub-band user scheduling; The wideband analog beamforming matrix and the subband digital beamforming matrix are used for hybrid beamforming of downlink signals for scheduled users.

In one embodiment, the method further includes: transmitting, by the horizontal linear array and the vertical linear array of the base station, a first training signal and a second training signal to the mobile terminal, respectively, so that the mobile terminal is respectively based on the first Training a signal and the second training signal to estimate a horizontal subchannel covariance matrix associated with the horizontal linear array and a vertical subchannel covariance matrix associated with the vertical linear array; respectively from the action The terminal receives an estimated feedback of the horizontal subchannel covariance matrix and an estimated feedback of the vertical subchannel covariance matrix; and constructs the horizontal subchannel covariance matrix based on the feedback The Kronek product of the vertical subchannel covariance matrix is used as the broadband channel covariance matrix of the physical channel.

In a synergistic aspect, embodiments of the present invention provide a method for hybrid beamforming in a mobile terminal. The method includes estimating a horizontal subchannel covariance matrix associated with the horizontal linear array based on a first training signal received from a horizontal linear array of antennas of a base station; based on an antenna from the base station a second training signal received by the linear array, estimating a vertical subchannel covariance matrix associated with the vertical linear array; and feeding back to the base station an estimate of the horizontal subchannel covariance matrix, and Estimation of the vertical subchannel covariance matrix.

In one embodiment, the method further comprises: associating the horizontal subchannel covariance matrix with the first transmission correlation coefficient, and The vertical subchannel covariance matrix is associated with a second transmission correlation coefficient; and wherein an estimate of the horizontal subchannel covariance matrix is fed back to the base station, and an estimate of the vertical subchannel covariance matrix is included Transmitting, to the base station, the amplitude and phase of the first transmission correlation coefficient and the amplitude and phase of the second transmission correlation coefficient.

In one embodiment, the method further comprises: estimating an equivalent channel based on the third training signal received from the base station; and feeding back an estimate of the equivalent channel to the base station.

In a fourth aspect, embodiments of the present invention provide an apparatus for hybrid beamforming in a base station. The apparatus includes a long-term estimation unit configured to calculate a broadband analog beamforming matrix based on a long-term estimation of a physical channel, and a quantization unit configured to quantize the broadband analog beamforming matrix to obtain a a quantized broadband analog beamforming matrix; an equivalent channel acquiring unit configured to apply the quantized broadband analog beamforming matrix to the physical channel to obtain an equivalent channel of the physical channel; short-term estimation a unit configured to calculate a subband digital beamforming matrix based on a short-term estimation of the equivalent channel; and a hybrid beamforming unit configured to utilize the sub-band digital beamforming matrix and the A quantized broadband analog beamforming matrix that performs hybrid beamforming on the downlink signal.

In a fifth aspect, an embodiment of the present invention provides an apparatus for hybrid beamforming in a base station. The apparatus includes a long-term estimating unit configured to calculate a broadband analog beamforming matrix based on a long-term estimation of a physical channel; an equivalent channel acquiring unit configured to be to the physical The channel applies the broadband analog beamforming matrix to obtain an equivalent channel of the physical channel; the short-term estimating unit is configured to calculate a sub-band digital beamforming matrix based on short-term estimation of the equivalent channel And performing subband user scheduling; and a hybrid beamforming unit configured to mix the downlink signals for the scheduled users using the broadband analog beamforming matrix and the subband digital beamforming matrix Beamforming.

In a sixth aspect, embodiments of the present invention provide an apparatus for hybrid beamforming in a mobile terminal. The apparatus includes a first estimating unit configured to estimate a horizontal subchannel covariance matrix associated with the horizontal linear array based on a first training signal received from a horizontal linear array of antennas of the base station; An estimating unit configured to estimate a vertical subchannel covariance matrix associated with the vertical linear array based on a second training signal received from a vertical linear array of antennas of the base station; and a feedback unit configured An estimate of the horizontal subchannel covariance matrix and an estimate of the vertical subchannel covariance matrix are fed back to the base station.

In one embodiment, the apparatus further comprises: an associating unit configured to associate the horizontal subchannel covariance matrix with the first transmission correlation coefficient, and to associate the vertical subchannel covariance matrix with the second Transmitting a correlation coefficient associated; and wherein feeding back an estimate of the horizontal subchannel covariance matrix to the base station and estimating the vertical subchannel covariance matrix comprises: feeding back the first to the base station The amplitude and phase of the correlation coefficient and the amplitude and phase of the second transmission correlation coefficient are transmitted.

In one embodiment, the apparatus further comprises: an equivalent channel estimation unit configured to estimate an equivalent channel based on the third training signal received from the base station; and the feedback unit is further configured to The base station returns an estimate of the equivalent channel.

The hybrid beamforming scheme according to an embodiment of the present invention makes deployment of a massive MIMO system more practical and cost effective. For example, the hybrid beamforming scheme according to the embodiment of the present invention may achieve at least one of the following beneficial effects: by quantifying the broadband analog beamforming matrix, the requirements of the existing hardware can be better satisfied, thereby reducing the hard Complexity and cost of implementation; frequency selective gain can be achieved by performing subband user scheduling based on short-term estimation of equivalent channels; instead of pairing wideband channels by using two low-dimensional covariance matrix feedbacks The feedback of the covariance matrix can significantly reduce the feedback overhead and training signal overhead of the covariance matrix used for channel estimation.

300‧‧‧Method of hybrid beamforming in a base station

S310~S350‧‧‧Steps

500‧‧‧Method of hybrid beamforming in a base station

S510~S540‧‧‧Steps

600‧‧‧Methods for hybrid beamforming in mobile terminals

S610~S630‧‧‧Steps

700‧‧‧A device for hybrid beamforming in a base station

710‧‧ long-term estimation unit

720‧‧‧Quantification unit

730‧‧‧ equivalent channel acquisition unit

740‧‧‧Short time estimation unit

750‧‧‧Mixed beamforming unit

800‧‧‧A device for hybrid beamforming in a base station

810‧‧‧Long-term estimation unit

820‧‧‧ equivalent channel acquisition unit

830‧‧‧Short time estimation unit

840‧‧‧Mixed beamforming unit

900‧‧‧A device for hybrid beamforming in mobile terminals

910‧‧‧First Estimation Unit

920‧‧‧Second estimation unit

930‧‧‧Return unit

The features, advantages, and other aspects of the various embodiments of the present invention will become more apparent from the aspects of the invention. In the drawings: Figure 1 shows a block diagram of a hybrid analog and digital beamforming architecture in which embodiments of the present invention may be implemented; Figure 2 illustrates another embodiment in which embodiments of the present invention may be implemented A block diagram of a mixed analog and digital beamforming architecture; FIG. 3 illustrates a first aspect of a base station in accordance with an embodiment of the present invention. A flowchart of a method of performing hybrid beamforming; FIG. 4 is a schematic diagram showing a uniform planar array of antennas in which embodiments of the present invention may be implemented; FIG. 5 illustrates a second aspect of an embodiment of the present invention A flowchart of a method for hybrid beamforming in a base station; FIG. 6 is a flow chart showing a method for hybrid beamforming in a mobile terminal in accordance with an embodiment of the present invention; FIG. EMBODIMENT OF THE INVENTION A block diagram of a device for performing hybrid beamforming in a base station in a fourth aspect; FIG. 8 is a block diagram showing an apparatus for performing hybrid beamforming in a base station according to a fifth aspect of the present invention. And FIG. 9 shows a block diagram of an apparatus for hybrid beamforming in a mobile terminal in accordance with a fifth aspect of an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the drawings. While the preferred embodiment of the invention has been shown in the drawings Rather, these embodiments are provided so that this disclosure will be thorough and complete.

1 shows a block diagram of a hybrid analog and digital beamforming architecture 100 in which embodiments of the present invention may be implemented. As shown in Figure 1, the base station is equipped with N _t antennas to serve S single antenna users. N ( N Each of the N _t ) RF chains is connected to all N _t antennas. First, the S data flows S ₀ (t) ... S _S-1 (t) are beamformed in the digital domain to generate N data streams shaped by the digital beam. Then, the N data streams are converted from the frequency domain to the time domain by an Inverse Digital Fourier Transform (IDFT) and input to the N RF channels to convert from the digital domain to the analog domain, thereby generating N analog data flows. . Then, N analogy beamforming process data and generate N _T streams in the analog domain. Then, N _T streams each of the transmission antenna is mapped to a base station, which in turn is transmitted to the UE (User Equipment) 1 ...... UE S.

2 shows a block diagram of another hybrid analog and digital beamforming architecture 200 in which embodiments of the present invention may be implemented. The hybrid beamforming process in Figure 2 is similar to that in Figure 1. However, in the architecture 200 of Figure 2, N ( N Each of the N _t ) radio frequency channels is only connected to a portion of the antenna, that is, to N _t /N antennas. Thus, architecture 200 is less complex than architecture 100 in FIG.

A hybrid beamforming method and apparatus according to an embodiment of the present invention will be described in detail below with reference to FIGS. 3 through 8.

In a first aspect, embodiments of the present invention propose a method of hybrid beamforming in a base station. 3 shows a flow diagram of a method 300 of performing hybrid beamforming in a base station in accordance with a first aspect of an embodiment of the present invention.

The method 300 begins in step 310. In step S310, a broadband analog beamforming matrix T is calculated based on long-term estimation of the physical channel. Then, in step S320, the broadband analog beamforming matrix T is quantized to obtain a quantized broadband analog beamforming matrix. . Subsequently, in step S330, the quantized broadband analog beamforming matrix is applied to the physical channel. To obtain the equivalent channel of the physical channel. Next, in step S340, the subband digital beamforming matrix W ( b ) is calculated based on the short time estimation of the equivalent channel. Finally, the subband digital beamforming matrix W ( b ) and the quantized broadband analog beamforming matrix are utilized. , performing hybrid beamforming on the downlink signal. Thus, with the hybrid beamforming scheme according to an embodiment of the invention, the downlink beamforming matrix F ( b ) on subband b is a quantized broadband analog beamforming matrix The product of the subband digital beamforming matrix W ( b ), namely: Where b =1, 2, ..., B and B are the number of subbands. It can be seen from the above formula (1) that in order to obtain the downlink beamforming matrix F ( b ) on the subband b , it is necessary to calculate the quantized broadband analog beamforming matrix. And subband digital beam shaping matrix W ( b ). Therefore, how to calculate a quantized broadband analog beamforming matrix will be discussed in detail later. And sub-band digital beamforming matrix W (b).

As described above, the quantized broadband analog beamforming matrix It is obtained by quantizing the broadband analog beamforming matrix T , so first, how to obtain the broadband analog beamforming matrix T will be described.

The wideband analog beamforming matrix T is at a wide frequency level and is designed based on channel statistics such as a wideband channel covariance matrix. Assumed K S users are scheduled at the same time. H ( b ) denotes the transmission channel matrix of K × N _t on the b- th subband: Where h _k ( b ) is the N _t ×1 downlink channel of the kth user on subband b .

For the hybrid beamforming architecture 100 shown in FIG. 1, considering the maximization of the reduced-dimensional equivalent channel capacity obtained by shaping the analog beam, there are: among them, Is a corresponding N _t × N _t transmission channel common variance matrix obtained by summing the broadband common-commutation matrix of k scheduled users, namely: The k- th user wideband channel common variance matrix is represented by: Therefore, the analog beamforming matrix T is represented by the following equation: T = [ u ₁ u ₂ ... u _N ] (6) where T is the dimension N _t × N , u ₁ u ₂ ... u _N is the largest with N Characteristic value corresponding Feature vector. As can be seen from equation (6), the analog beamforming should include beams pointing to all k scheduled users.

Next, traditional precoding algorithms such as zero-forcing algorithms can be utilized, based on equivalent channels. A N × K digital beam shaping matrix W ( b ) on the subband b is obtained.

Furthermore, for the hybrid beamforming architecture 200 shown in FIG. 2, since each RF channel is only connected to a part of the antennas, that is, N _t /N antennas, T is a block diagonal matrix of the following form: Wherein, g _k for the first N _t / N × 1 vector. Thus, architecture 200 can be viewed as a special case of architecture 100 resulting from setting the non-blocking diagonal elements to zero. For architecture 200, the key is how to design g _k in equation (7). To this end, the present invention proposes obtaining these vectors from the corresponding elements of the analog beamforming matrix shown in equation (6) for architecture 100. Therefore, we can get: g _k = t _k /∥ t _k ∥, t _k = u _k (( k -1) N _t / N +1: kN _t / N ) (8)

Further, the generation of the digital beamforming matrix W ( b ) on the subband b is the same as that of the hybrid architecture 100, that is, the equivalent channel can be utilized. Traditional precoding algorithms, such as zero-forcing algorithms.

From the above analysis, in order to obtain the beamforming matrix T analog and digital beamforming matrix W (b), the base station needs to know all of the user's broadband channel covariance matrix R _k and vector dimensionality reduction equivalent channel . Therefore, how to estimate them within a hybrid architecture becomes a key issue. A discussion will be made below for a TDD system with a calibrated antenna, and an FDD system and a TDD system without a calibrated antenna, respectively.

TDD system with calibrated antenna

In a TDD system, if the antenna is accurately calibrated, the channel reciprocity can be utilized to estimate the downstream channel based on the uplink training signal.

In this case, the estimation of channel status information consists of two parts. Part of it is the estimation of the channel vector h _k ( b ) and the broadband channel covariance matrix, which can be completed at the long-term level considering the slow variation characteristics of the broadband channel covariance matrix and the reduction of training overhead. Covariance matrices deduced analog beamforming matrix T based broadband channel. The other part is the equivalent channel vector with the derived analog beamforming matrix T Estimate. The estimate is short-lived and at the sub-band level to obtain a frequency selective gain.

Embodiments of the present invention propose the following process to obtain a broadband channel covariance matrix and a dimensionality reduction equivalent channel for all users.

Long-term estimation of common vector of channel vector and broadband channel

Step 1: Based on the uplink training signal, estimate the channel vector h _k ( b ) on all subbands for all users.

Embodiments of the present invention propose a method for multi-analog beamforming reception of orthogonal uplink training signals for estimating channel vectors. Specifically, training signals of different users are transmitted on different subcarriers. It is assumed that the uplink training signal s _k of the user k for the b- th sub-band channel estimation is transmitted on the bth (k) th subcarrier. N _t / N different analog beamforming matrices are used in N _t / N OFDM symbols for reception, respectively. use A received signal vector of N x1 on the kth subcarrier in the i- th OFDM symbol. In N _t / N OFDM symbols, the received signal on the b (k) th subcarrier is represented by the following equation: The above formula can be further expressed as: Analog beamforming matrix T ₁ T ₂ ... The selection needs to meet the following conditions: combination matrix [ T ₁ T ₂ ... The rank of n ] is N _t , and the channel vector h _k ( b ) can be estimated according to equation (10).

Step 2: Estimate the broadband channel co-variance matrix R _k based on the channel vector h _k ( b ) according to equation (5).

Step 3: Based on R _k , the analog beamforming matrix T is derived from equation (6) or (7). After that, the analog beamforming matrix T needs to be quantized, which will be described later.

Short-term estimation of equivalent channel vectors with derived analog beamforming

Step 4: Set the analog beamforming matrix to the form derived in step 3 and send the uplink training signal to the equivalent channel vector from all users. Make an estimate.

Step 5: Based on the equivalent channel Perform digital beamforming and user scheduling.

FDD system or TDD system without accurate calibrated antenna

For FDD systems or TDD systems that do not have a precisely calibrated antenna, channel reciprocity is not well maintained. Therefore, it is necessary to estimate and feedback the downlink channel common variance matrix and the equivalent channel vector.

In this case, channel state information estimation and feedback consists of two parts. Part of it is the long-term estimation and feedback of the common-variation matrix of broadband channels. The analog beamforming matrix is derived based on the feedback of the broadband common-variation matrix. The other part is the short-term estimation and feedback of the equivalent channel vector with the derived analog beamforming.

Embodiments of the present invention propose the following process to obtain a broadband channel covariance matrix and a dimensionality reduction equivalent channel for all users.

Long-term estimation and feedback of a common-variation matrix of broadband channels

Due to the practical limitations of geometric size, massive MIMO systems typically employ a Uniform Planar Array (UPA) of antennas, as shown in Figure 4. Uniform planar array, N _t transmit antennas 4 includes M _c M _r rows by column antennas, i.e., N _t = M _r M _c. It is well known that the channel vector of N _t ×1 can be approximated as the horizontal subchannel vector of M _r ×1 and the kronecker product of the vertical subchannel vector of M _c ×1, namely: The corresponding channel common variance matrix can be obtained by:

It can be seen that the channel covariance matrix R of the uniform planar array can be approximated by the subchannel covariance matrix R _r associated with the horizontal linear array and the Krone of the subchannel covariance matrix R _c associated with the vertical linear array. Keji.

Therefore, the feedback of the total channel common variance matrix R of N _t × N _t can be replaced by the feedback of two low-dimensional covariance matrix: one is the covariance matrix R of the horizontal linear array M _r × M _{r r} , the other is the covariance matrix R _c of M _c × M _c of the vertical linear array. In this way, the feedback overhead of the covariance matrix for channel estimation and the overhead of the training signal can be significantly reduced.

Step 1: Each user estimates the horizontal channel vector on all subbands based on the first downlink training signal with a specially designed analog beamforming. Each user then quantizes and feeds back the horizontal broadband covariance matrix.

In order to estimate the horizontal channel vector, the base station sets the first initial analog beamforming matrix to the following form:

Where e _i is a basic vector, except that the i- th element is 1 in e _i and all other elements are zero.

Then, the base station selects the i-th array antenna level (i.e., selects the antenna in the i-th row in FIG. 4) and M _r N RF channels. Subsequently, using the base station in the above formula (13) a first initial analog beamforming matrix, in the selected i-th horizontal antenna array transmits a training signal for the downlink channel vector estimate level (also referred to as a first lower Line training signal). By using the first initial analog beamforming matrix in the above equation (13), the conventional training signal design and channel estimation method for the full digital beamforming method can be reused.

As a specific example, it is assumed that the antenna array in Fig. 4 includes 4 rows x 8 columns of antennas. The base station first selects 8 antennas and 8 RF channels in the 1st row. Subsequently, the base station transmits the first downlink training signal on the 8 antennas of the first row and silences the antennas of the other rows using the first initial analog beamforming matrix in the above equation (13). Next, the user estimates a horizontal channel vector corresponding to the 8 antennas of the 1st row based on the first downlink training signal, thereby obtaining an 8×1 horizontal channel vector. Then, the user calculates a covariance matrix corresponding to the 8 antennas of the 1st row based on the 8×1 horizontal channel vector, thereby obtaining an 8×8 covariance matrix. Similarly, the base station and the user perform the above processes for the 8 antennas of the 2nd to 4th rows, respectively, thereby finally obtaining 4 8×8 covariance matrix. Then, by averaging the four 8×8 covariance matrices, a common variance number matrix R _r for the four horizontal antenna arrays can be obtained.

Based on the estimated horizontal channel vectors on all subbands, the horizontal subchannel covariance matrix can be estimated by each user according to equation (5). Quantization and feedback of the horizontal subchannel common variance matrix can reuse existing A method for linear arrays.

In one embodiment, the total number of horizontal sub-channel matrix R _r with the first correlation coefficients associated with transmission variation. Taking a single-polarized linear array as an example, the horizontal sub-channel common variance matrix R _r can be approximated as: Where ρ = αe ^jθ is the transmission correlation coefficient. Thus, the feedback of the horizontal subchannel common variogram matrix R _r can be simplified to the feedback of the amplitude α and the phase θ of ρ .

Step 2: Each user estimates the vertical channel vectors on all subbands based on a second downlink training signal with a specially designed analog beamforming. Each user then quantizes and feeds back the vertical wideband covariance matrix.

In order to estimate the vertical channel vector, the base station sets the second initial analog beamforming matrix to the following form.

As can be seen from equation (15), the second initial analog beamforming matrix includes a unit matrix of M _c × M _c , and the remaining elements are all zero.

Then, in addition to the base station, the i- th vertical antenna array is selected (ie, the antenna of the ith column in FIG. 4 is selected) and M _c N -channels, and using the second initial analog beamforming matrix in equation (15) to transmit the downlink training signal (also referred to as the second downlink training signal) for vertical channel vector estimation, the vertical channel vector The feedback of the estimated and vertical subchannel covariance matrix R _c is similar to that in step 1.

Also, similarly, in one embodiment, the total number of vertical sub-channel R _c matrix associated with a second transmission coefficient of variation is associated. Thus, the feedback of the vertical subchannel covariance matrix R _c can be simplified to the feedback of the amplitude and phase of the second transmission correlation coefficient.

Step 3: According to equation (12), the base station obtains the broadband common-covariance matrix of all users based on the feedback of the horizontal and vertical broadband common-variation matrix in steps 1 and 2. The base station then derives an analog beamforming matrix T from equation (6) or (7).

Short-term estimation and feedback of equivalent channel vectors with derived analog beamforming

Step 4: The base station sets the analog beamforming matrix to the form derived in step 3, and transmits a downlink training signal (also referred to as a third downlink training signal) for the equivalent channel estimation.

Step 5: Each user estimates, quantifies, and rewards the equivalent channel vector . At this point, traditional channel state information feedback methods can be used, such as scalar quantization, self-adjusting codebooks, and the like.

Step 6: The base station is based on the equivalent channel from all users in step 5. Feedback, performing digital beamforming and user scheduling.

For steps 5 and 6, take the self-adjusting codebook as an example, the final codeword Is the covariance matrix R _eff and predefined codeword W (e.g., the DFT vector) product channels, namely:

The user can estimate the common-variation matrix of the broadband equivalent channel based on the estimated equivalent channel.

due to

Therefore, the base station can also derive the common-variation matrix of the wide-band equivalent channel based on the knowledge of the analog beamforming and the broadband common-variation matrix derived in step 3.

Correspondingly, since the user and the base station can respectively obtain the covariance matrix, it is not necessary to feed back the covariance matrix in the form of a self-adjusting codebook. The user only needs to select and feedback the best codeword W in equation (16).

The above describes how to calculate the broadband analog beamforming matrix T and the subband digital beamforming matrix W ( b ) through a specific embodiment. In the following, how to quantize the broadband analog beamforming matrix by quantizing the broadband analog beamforming matrix T will be described in detail. .

It can be seen from the equations (6) and (8) described above that each element in the analog beamforming matrix T obtained based on the channel estimation is an arbitrary complex number whose phase and amplitude are random. However, analog beamforming is implemented using a phase shifter that requires each element in the analog beamforming matrix T to be constant modulus and phase selected in a predetermined set of phases. Therefore, considering the complexity and cost of hardware implementation, it is necessary to transform the analog beamforming matrix T into a matrix that conforms to existing hardware requirements, and to make the loss as small as possible. In this context, this transformation is also referred to as quantification of the analog beamforming matrix T for hardware impairments. An intuitive approach is to separately quantify each non-zero element in the analog beamforming matrix T. However, this approach may not be optimal considering the maximum channel capacity. To this end, for the analog beamforming matrix T in equations (6) and (8), the embodiment of the present invention proposes a method of performing phase searching according to the channel capacity maximization criterion, and the specific process is as follows.

As shown in equation (3), an optimized analog beamforming matrix with finite phase resolution (ie, a quantized analog beamforming matrix) can be represented by the following equation according to the equivalent channel capacity maximization criterion: among them, Represents the determinant of the matrix in the brackets, Representation Largest matrix . matrix It is called a candidate matrix, in which the non-zero elements have an amplitude of 1 and the phase is selected from a predetermined set of phases. In order to find an optimized analog beamforming Need to search all candidate matrices. However, since the number of analog beamforming candidate matrices is for architecture 100 For architecture 200 Therefore, the computational complexity of accomplishing such a comprehensive search is very high for practical use in large antenna systems. Therefore, it is important and necessary to significantly reduce the sub-optimal method for computational complexity of analog beamforming matrix searches. To this end, embodiments of the present invention propose a method of performing phase search in accordance with channel capacity maximization criteria.

First, the amplitude of analog beamforming matrix T of formula (6) or (7) of each element were normalized, and its phase is compared with the phase of the predetermined phase set, select its nearest A phase value that forms an initialized analog beamforming matrix and is used as a seed for optimization. As a specific example, the predetermined phase set includes 16 phase values obtained by dividing 16 degrees into 360 degrees. It should be understood that the selection of the predetermined set of phases depends on the requirements of the hardware used for analog beamforming, i.e., any suitable predetermined set of phases can be selected depending on the particular hardware.

Next, you need to perform several iterations. In each iterative operation, according to the channel capacity maximization criterion, the phase search is performed one by one for each element in the above-mentioned initialized analog beamforming matrix, so that each iterative operation is determined. An element in the middle. Specifically, in each iterative operation, only the phase of one element in the initialized analog beamforming matrix is changed, while the other elements remain unchanged, and finally, for one element, one of the predetermined phase sets can be selected. In (19) The largest phase value. Thus, the number of analog beamforming matrix for candidates architecture 100 may be reduced significantly to qdNN _t, architecture 200 may be significantly reduced for the qdN _t, where q represents the number of repeated operations. Usually, the decimal value q is sufficient to satisfy the convergence of the search. For example, in the latter analogy, the value of q is 4.

It should be understood that the above-described quantization process for matrix T is presented for illustrative purposes only. Different quantization processes may be employed for different hardware, and the scope of the invention is not limited in this respect.

It can be understood from the above description that the proposed analog beamforming is based on the broadband common-covariance moment of all candidate users served in each cell. The beam is indicated to all of the candidate users. Digital beamforming is then performed based on the estimation of the reduced-dimensional equivalent channel after the analog beam shaping. The traditional precoding algorithm such as ZF (zero-forcing) can be reused for digital beamforming design. The design regarding user scheduling will be described below.

First, an embodiment of the present invention proposes a joint analog beamforming and broadband user scheduling scheme (hereinafter referred to as a joint scheduling scheme). Considering broadband user scheduling, analog beamforming can be designed in conjunction with user scheduling based on criteria for maximum weighting and capacity. Based on the sum of the wideband channel covariance matrices of the scheduled users, the analog beamforming matrix is derived from equations (4), (5), and (6) described above.

In the joint scheduling scheme, since the design of the analog beamforming depends on the user to be scheduled, the channel vector h _k ( b ) of all S users on all B subbands should be in each scheduling subframe ( 1 k S ;1 b B) is estimated.

In the joint scheduling scheme, since user scheduling is considered in analog beamforming, there is a relationship between analog beamforming and user scheduling. The wide-band channel common-variation matrix has a slow-changing characteristic, and user scheduling is a fast-changing process. Once the results of the user scheduling change, the analog beamforming matrix also needs to change accordingly. Since the measurement of analog beamforming (channel estimation) requires a lot of overhead, it is desirable that the analog beamforming matrix is a slowly varying amount. In other words, the analog beamforming matrix preferably does not change once the scheduled user changes. To this end, embodiments of the present invention also propose a separate analog beamforming and subband user scheduling. Scheme (hereinafter referred to as the separation scheduling scheme).

In the separation in the schedule, based on all the S wideband channel candidate users and covariance matrices, designed for indicating beam S for all user analog beamforming, namely: Sub-band user scheduling is then performed based on a short-term estimate of the equivalent channel after the analog beam shaping. Thus, analog beamforming and user scheduling are performed separately, thereby enabling subband user scheduling to obtain frequency selective gain.

In the separate scheduling scheme, the analog beamforming matrix changes when a user's channel common variance matrix changes, regardless of user scheduling within each subframe. Thus, the update period of the analog beamforming matrix is slower than the aforementioned joint scheduling scheme. Another advantage of this separate scheduling scheme is that only the equivalent channel vector T ^H h _k ( b ) is required within each subframe. k S ;1 b B ) Estimation is made without estimating the channel vector h _k ( b ) as in the joint scheduling scheme.

A performance comparison between the two user scheduling schemes is shown in Table I below. Analog parameters and assumptions are summarized in Table II. It can be seen from the comparison that for 64 transmit antennas and 16 RF channels, the split scheduling scheme is 20% higher than the cell edge gain of the joint scheduling scheme.

In a second aspect, embodiments of the present invention also provide a method of hybrid beamforming in a base station. FIG. 5 illustrates a flow diagram of a method 500 of hybrid beamforming in a base station in accordance with a second aspect of an embodiment of the present invention. The method 500 begins at step 510. At step S510, a broadband analog beamforming matrix is calculated based on long-term estimation of the physical channel. Then, in step S520, the broadband analog beamforming matrix is applied to the physical channel to obtain an equivalent channel of the physical channel. Then, in step S530, based on the short-term estimation of the equivalent channel, the sub-band digital beamforming matrix is calculated and sub-band user scheduling is performed. Then, in step S540, the downlink signal for the scheduled user is subjected to hybrid beamforming using the broadband analog beamforming matrix and the subband digital beamforming matrix.

In one embodiment, the method 500 further includes transmitting, by the horizontal linear array and the vertical linear array of the base station, a first training signal and a second training signal to the mobile terminal, respectively, so that the mobile terminal is respectively based on the first Training a signal and the second training signal to estimate a horizontal subchannel covariance matrix associated with the horizontal linear array and a vertical subchannel covariance matrix associated with the vertical linear array; respectively from the action The terminal receives an estimate of the horizontal subchannel covariance matrix Feedback, and estimated feedback of the vertical subchannel covariance matrix; and Kroneck based on the feedback to construct the horizontal subchannel covariance matrix and the vertical subchannel covariance matrix Product, as the broadband channel covariance matrix of the physical channel.

It should be understood that the above description of user scheduling, training signal design, channel estimation, and feedback of channel estimation results as described for the hybrid beamforming method of the first aspect of the embodiments of the present invention is equally applicable to method 500. For the sake of brevity, I will not repeat them.

In a synergistic aspect, embodiments of the present invention also propose a method for hybrid beamforming in a mobile terminal. 6 shows a flow diagram of a method 600 for hybrid beamforming in a mobile terminal in accordance with an embodiment of the present invention. The method 600 begins at step 610. At step S610, a horizontal subchannel covariance matrix associated with the horizontal linear array is estimated based on the first training signal received from the horizontal linear array of antennas of the base station. Then, in step S620, a vertical subchannel covariance matrix associated with the vertical linear array is estimated based on a second training signal received from a vertical linear array of antennas of the base station. Then, in step S630, an estimate of the horizontal subchannel covariance matrix and an estimate of the vertical subchannel covariance matrix are fed back to the base station.

In one embodiment, the method 600 further includes: associating the horizontal subchannel covariance matrix with a first transmission correlation coefficient, and associating the vertical subchannel covariance matrix with a second transmission correlation coefficient; And wherein the base station is fed back to the horizontal subchannel co-variation moment The estimation of the array and the estimation of the vertical subchannel covariance matrix include: feeding back the amplitude and phase of the first transmission correlation coefficient and the amplitude and phase of the second transmission correlation coefficient to the base station.

In one embodiment, method 600 further includes estimating an equivalent channel based on a third training signal received from the base station; and feeding back an estimate of the equivalent channel to the base station.

In a fourth aspect, embodiments of the present invention also provide an apparatus for hybrid beamforming in a base station. 7 shows a block diagram of an apparatus 700 for hybrid beamforming in a base station in accordance with a fourth aspect of an embodiment of the present invention. As shown, the apparatus 700 includes a long time estimation unit 710 configured to calculate a broadband analog beamforming matrix based on long-term estimation of a physical channel, and a quantization unit 720 configured to shape the broadband analog beam The matrix is quantized to obtain a quantized broadband analog beamforming matrix; an equivalent channel acquisition unit 730 is configured to apply the quantized broadband analog beamforming matrix to the physical channel to obtain the physical channel Equivalent channel; short time estimation unit 740 configured to calculate a subband digital beamforming matrix based on short-term estimation of the equivalent channel; and hybrid beamforming unit 750 configured to utilize the sub-portion A hybrid beamforming is performed on the downlink signal with a digital beamforming matrix and the quantized broadband analog beamforming matrix.

In one embodiment, quantization unit 720 is further configured to: normalize the amplitude of each non-zero element in the wide-band analog beamforming matrix; and at the predetermined phase for each of the non-zero elements Performing a phase search element by element in the set to select the capacity of the equivalent channel The amount of phase that maximizes.

In one embodiment, apparatus 700 further includes a scheduling unit configured to perform sub-band user scheduling based on a short-term estimate of the equivalent channel.

In one embodiment, apparatus 700 further includes: a transmitting unit configured to transmit a first training signal and a second training signal to the mobile terminal through a horizontal linear array and a vertical linear array of the base station, respectively, to The mobile terminal estimates a horizontal subchannel covariance matrix associated with the horizontal linear array and a common subchannel covariation associated with the vertical linear array based on the first training signal and the second training signal, respectively a number matrix; a receiving unit configured to receive, from the mobile terminal, an estimated feedback of the horizontal subchannel covariance matrix, and an estimated feedback of the vertical subchannel covariance matrix, respectively; and a building unit And configuring, based on the feedback, a Kronecker product of the horizontal subchannel covariance matrix and the vertical subchannel covariance matrix to serve as a broadband channel covariance matrix of the physical channel.

In a fifth aspect, embodiments of the present invention also provide an apparatus for hybrid beamforming in a base station. 8 shows a block diagram of an apparatus 800 for performing hybrid beamforming in a base station in accordance with a fifth aspect of an embodiment of the present invention. As shown, the apparatus 800 includes a long-term estimation unit 810 configured to calculate a broadband analog beamforming matrix based on a long-term estimation of a physical channel; an equivalent channel acquisition unit 820 configured to the physical channel Applying the broadband analog beamforming matrix to obtain an equivalent channel of the physical channel; a short time estimating unit 830 configured to be based on Calculating a sub-band digital beamforming matrix and performing sub-band user scheduling for short-term estimation of the equivalent channel; and a hybrid beamforming unit 840 configured to utilize the broadband analog beamforming matrix and the sub- A digital beamforming matrix is used to perform hybrid beamforming on the downlink signals for the scheduled users.

In one embodiment, the apparatus 800 further includes: a training signal transmitting unit configured to transmit the first training signal and the second training signal to the mobile terminal through the horizontal linear array and the vertical linear array of the base station, respectively, so that The mobile terminal estimates a horizontal subchannel covariance matrix associated with the horizontal linear array and a vertical subchannel associated with the vertical linear array based on the first training signal and the second training signal, respectively a common variance matrix; a feedback receiving unit configured to receive an estimated feedback of the horizontal subchannel covariance matrix from the mobile terminal and an estimated feedback of the vertical subchannel covariance matrix, respectively; And a building unit configured to construct a Kronecker product of the horizontal subchannel covariance matrix and the vertical subchannel covariance matrix based on the feedback to serve as a broadband channel covariation of the physical channel Number matrix.

In a sixth aspect, embodiments of the present invention also provide an apparatus for hybrid beamforming in a mobile terminal. 9 is a block diagram of an apparatus 900 for performing hybrid beamforming in a mobile terminal in accordance with a sixth aspect of an embodiment of the present invention. As shown, apparatus 900 includes a first estimating unit 910 configured to estimate water associated with the horizontal linear array based on a first training signal received from a horizontal linear array of antennas of the base station a flat subchannel common variance matrix; a second estimating unit 920 configured to estimate a vertical subchannel associated with the vertical linear array based on a second training signal received from a vertical linear array of antennas of the base station a variance matrix; and a feedback unit 930 configured to feed back to the base station an estimate of the horizontal subchannel covariance matrix and an estimate of the vertical subchannel covariance matrix.

In an embodiment, the apparatus 900 further includes: an associating unit configured to associate the horizontal subchannel covariance matrix with the first transmission correlation coefficient, and the vertical subchannel covariance matrix and the second Transmitting a correlation coefficient associated; and wherein feeding back an estimate of the horizontal subchannel covariance matrix to the base station and estimating the vertical subchannel covariance matrix comprises: feeding back the first to the base station The amplitude and phase of the correlation coefficient and the amplitude and phase of the second transmission correlation coefficient are transmitted.

In one embodiment, apparatus 900 further includes an equivalent channel estimation unit configured to estimate an equivalent channel based on a third training signal received from the base station; and the feedback unit is further configured to The base station returns an estimate of the equivalent channel.

It should be understood that the units included in devices 700, 800, and 900 can be implemented in a variety of ways, including software, hardware, firmware, or any combination thereof. In one embodiment, one or more of the units may be implemented using software and/or firmware, such as machine executable instructions stored on a storage medium. In addition to or in lieu of machine-executable instructions, some or all of the units 700, 800, and/or 900 may be at least partially Or multiple hardware logic components to achieve. By way of example and not limitation, exemplary types of hardware logic elements that may be used include field effect programmable logic gate array (FPGA), special application integrated circuit (ASIC), application specific standard (ASSP), system single chip (SOC). , Complex Programmable Logic Devices (CPLDs), and more.

System level analogy results for the proposed scheme of the present invention will be described below. In this section, the performance of the proposed hybrid beamforming method for massive MIMO is verified on 19 sites/57 pentagon cells. Analog parameters and assumptions are summarized in Table II. Each base station has a monopole planar array of 8 rows by 8 columns of antennas and serves 10 single antenna users. Multi-user MIMO with separate analog beamforming and sub-band user scheduling (ie, the aforementioned separate scheduling scheme) is employed. The analogy results are shown in Tables III-IV.

Hybrid beamforming vs. digital beamforming

In Table III, hybrid architecture 100 achieves similar performance to digital beamforming with half of the RF channel. Further reducing the number of RF channels to a quarter, the performance loss is still limited to 9%. The hybrid architecture 200 has a greater performance penalty than full-bit beamforming, with a cell average loss of 29% and a cell edge loss of 38%, primarily due to less beamforming gain for simplified analog beamforming.

Hardware impairment

The hardware impairment of 4-bit phase resolution is shown in Table IV. With the proposed phase search method, the performance loss of the hybrid architecture 100 is limited to 4% in the ideal analog beamforming matrix (using the ideal analog beamforming matrix of equation (6) or (7)). Since the analog beamforming matrix is further optimized during the phase search according to the channel capacity maximization criterion, the hybrid architecture 200 achieves better performance, achieving gains of up to 15%.

Although the present invention has been described with reference to the specific embodiments thereof, it is obvious to those skilled in the art that the present invention is not limited to the details of the embodiments described above, and the present invention can be variously modified and modified. This is accomplished without departing from the scope of the invention. The present embodiments are to be considered in all respects as illustrative and not limiting, and the scope of the invention Variations are therefore included within the scope of the invention.

Claims (15)

A method for hybrid beamforming in a base station, comprising: calculating a broadband analog beamforming matrix based on long-term estimation of a physical channel; and quantizing the broadband analog beamforming matrix to obtain a quantized broadband analog beam Forming a matrix; applying the quantized broadband analog beamforming matrix to the physical channel to obtain an equivalent channel of the physical channel; and calculating a subband digital beamforming matrix based on the short time estimation of the equivalent channel; And performing hybrid beamforming on the downlink signal by using the subband digital beamforming matrix and the quantized broadband analog beamforming matrix. The method of claim 1, wherein quantizing the broadband analog beamforming matrix comprises normalizing the amplitude of each non-zero element in the broadband analog beamforming matrix; and for each non-zero The element performs a phase search element by element in a predetermined set of phases to select a phase that maximizes the capacity of the equivalent channel. The method of claim 1, further comprising performing sub-band user scheduling based on the short-term estimation of the equivalent channel. The method of claim 1, further comprising: respectively transmitting a horizontal linear array and a vertical linear array of the base station The mobile terminal transmits a first training signal and a second training signal, so that the mobile terminal estimates a horizontal subchannel covariance matrix associated with the horizontal linear array and based on the first training signal and the second training signal, respectively. a vertical subchannel covariance matrix associated with the vertical linear array; receiving, from the mobile terminal, an estimated feedback of the horizontal subchannel covariance matrix, and an estimated feedback of the vertical subchannel covariance matrix; And constructing a Kroneck product of the horizontal subchannel covariance matrix and the vertical subchannel covariance matrix based on the feedback as a broadband channel covariance matrix of the physical channel. A method for performing hybrid beamforming in a base station, comprising: calculating a broadband analog beamforming matrix based on long-term estimation of a physical channel; applying the broadband analog beamforming matrix to the physical channel to obtain the physical channel An equivalent channel; calculating a subband digital beamforming matrix and performing subband user scheduling based on the short-term estimation of the equivalent channel; and utilizing the broadband analog beamforming matrix and the subband digital beamforming matrix, Hybrid beamforming for downlink signals of scheduled users. The method of claim 5, further comprising: transmitting the first training signal and the second training signal to the mobile terminal through the horizontal linear array and the vertical linear array of the base station, respectively, so that the action The terminal estimates a horizontal subchannel covariance matrix associated with the horizontal linear array and a vertical subchannel covariance matrix associated with the vertical linear array based on the first training signal and the second training signal, respectively; The mobile terminal receives the estimated feedback of the horizontal subchannel covariance matrix and the feedback of the vertical subchannel covariance matrix; and constructs the horizontal subchannel covariance matrix and the vertical based on the feedback The Kronecker product of the subchannel common variance matrix is used as the broadband channel covariance matrix of the physical channel. A method for hybrid beamforming in a mobile terminal, comprising: estimating a horizontal subchannel covariance matrix associated with the horizontal linear array based on a first training signal received from a horizontal linear array of antennas of the base station Estimating a vertical subchannel covariance matrix associated with the vertical linear array based on a second training signal received from a vertical linear array of antennas of the base station; and feeding back the horizontal subchannel covariance to the base station An estimate of the matrix and an estimate of the matrix of covariances of the vertical subchannels. The method of claim 7, further comprising: associating the horizontal subchannel covariance matrix with the first transmission correlation coefficient, and associating the vertical subchannel covariance matrix with the second transmission correlation coefficient; In which the base station is fed back to estimate the horizontal subchannel common variance matrix And estimating the vertical subchannel covariance matrix includes: feeding back the amplitude and phase of the first transmission correlation coefficient and the amplitude and phase of the second transmission correlation coefficient to the base station. The method of claim 7, further comprising: estimating an equivalent channel based on the third training signal received from the base station; and feeding back an estimate of the equivalent channel to the base station. An apparatus for performing hybrid beamforming in a base station, comprising: a long-term estimating unit configured to calculate a broadband analog beamforming matrix based on a long-term estimation of a physical channel; and a quantization unit configured to the broadband analog beam The shaping matrix is quantized to obtain a quantized broadband analog beamforming matrix; the equivalent channel acquiring unit is configured to apply the quantized broadband analog beamforming matrix to the physical channel to obtain the physical channel, etc. a valence channel; a short-term estimating unit configured to calculate a sub-band digital beamforming matrix based on a short-term estimation of the equivalent channel; and a hybrid beamforming unit configured to utilize the sub-band digital beamforming matrix And the quantized broadband analog beamforming matrix performs hybrid beamforming on the downlink signal. The apparatus of claim 10, wherein the quantizing unit is further configured to: amplitude of each non-zero element in the broadband analog beamforming matrix Normalization is performed; and for each of the non-zero elements, a phase search is performed element by element in a predetermined set of phases to select a phase that maximizes the capacity of the equivalent channel. The apparatus of claim 10, further comprising: a transmitting unit configured to transmit the first training signal and the second training signal to the mobile terminal through the horizontal linear array and the vertical linear array of the base station, respectively, so that the mobile terminal Estimating a horizontal subchannel covariance matrix associated with the horizontal linear array and a vertical subchannel covariance matrix associated with the vertical linear array based on the first training signal and the second training signal, respectively; receiving unit, Configuring to receive an estimate of the feedback of the horizontal subchannel covariance matrix from the mobile terminal and an estimate of the feedback of the vertical subchannel covariance matrix, respectively; and a building unit configured to construct based on the feedback The horizontal subchannel covariance matrix and the Kroneck product of the vertical subchannel covariance matrix are used as the broadband channel covariance matrix of the physical channel. An apparatus for performing hybrid beamforming in a base station, comprising: a long-term estimating unit configured to calculate a broadband analog beamforming matrix based on long-term estimation of a physical channel; an equivalent channel acquiring unit configured to The physical channel applies the broadband analog beamforming matrix to obtain an equivalent channel of the physical channel; the short-term estimating unit is configured to be based on a short-term estimation of the equivalent channel Calculating a subband digital beamforming matrix and performing subband user scheduling; and a hybrid beamforming unit configured to utilize the broadband analog beamforming matrix and the subband digital beamforming matrix for scheduling The user's downlink signal performs hybrid beamforming. An apparatus for hybrid beamforming in a mobile terminal, comprising: a first estimating unit configured to estimate an association with the horizontal linear array based on a first training signal received from a horizontal linear array of antennas of the base station a horizontal subchannel common variance matrix; a second estimating unit configured to estimate a vertical subchannel co-variation associated with the vertical linear array based on a second training signal received from a vertical linear array of antennas of the base station a number matrix; and a feedback unit configured to feed back to the base station an estimate of the horizontal subchannel covariance matrix and an estimate of the vertical subchannel covariance matrix. The apparatus of claim 16, further comprising: an associating unit configured to associate the horizontal subchannel covariance matrix with the first transmission correlation coefficient, and the vertical subchannel covariance matrix and the second transmission Correlating the correlation coefficient; and wherein the estimating the horizontal subchannel covariance matrix to the base station and the estimating the vertical subchannel covariance matrix comprises: feeding back the amplitude of the first transmission correlation coefficient to the base station And phase and the amplitude and phase of the second transmission correlation coefficient.