JP2015518324A - Codebook structure - Google Patents

Abstract

A method performed at a base station used in a wireless communication system is disclosed. This method includes having one, two, three, and four layer codebooks for four transmit antenna (4TX) transmission, each codebook including a plurality of precoding matrices, and a plurality of precoding. Precoding the data using one of the matrices and transmitting the precoded data to the user equipment, each of the 1-layer and 2-layer codebooks having a first A codebook and a second codebook are included, and each precoding matrix in the first codebook has a first index and a second index. Other devices, systems and methods are also disclosed.

Description

  This application was filed on March 7, 2013, US Provisional Application No. 61 / 767,896 entitled “Observations on Codebook Construction” and filed on March 8, 2013. US Provisional Application No. 61 / 775,058 entitled `` Observations on Codebook Construction '' and `` Enhancements to a Structured Codebook '' filed April 5, 2013 US Provisional Application No. 61 / 808,934 entitled “Enhancement to the 4 Transmit Antenna Precoding Codebook” filed April 29, 2013 Provisional Application No. 61 / 817,150 and US Provisional Application entitled “Enhancement to the 4 Transmit Antenna Precoding Codebook” filed on April 29, 2013 61 / 817,247 and May 2013 Claims all the benefits of US Provisional Application No. 61 / 821,989 entitled "Improvements to the 4 Transmit Antenna Precoding Codebook", filed on the 10th Is incorporated herein by reference.

  The present invention relates to the design of a precoding matrix, and more particularly to the design of a precoding matrix that derives a precoding matrix as the product of two matrices.

  Wireless communication systems require even higher spectral efficiency in order to accommodate higher throughput requirements within a limited frequency band. Multiple antennas or multiple-input and multiple-output (MIMO) systems, particularly closed-loop transmission techniques such as beam shaping and precoding, have been widely considered for improved spectral efficiency. . In the MIMO precoding scheme, data to be transmitted is divided into one or more streams, and these streams are mapped to one or more transmission layers, and the data in the layers is precoded before transmission. ) Or a precoding matrix. The number of transmission layers is called the transmission rank. The transmission rank can be optimally selected for a given channel realization, taking into account, for example, transmission power and overall channel statistics.

  In a precoding strategy based on a codebook, a given codebook consists of a transmitter or base station (BS) and all receivers, ie mobile stations (MS) or user equipment (UE). ) And made available. The receiver then selects from the codebook a precoder that maximizes its performance (eg, its data rate) and feeds back the precoder index. Precoder rank selection may also be included in the precoder selection algorithm. The rate of feedback can vary from short-term feedback, once every coherent time interval, to long-term feedback, once every few coherent time intervals.

  In many systems, the optimal precoder for two adjacent transmission blocks from the codebook is close in the set of all possible precoders with respect to the appropriate distance measure. Here, in an actual system, the channel does not change abruptly from one transmission block to an adjacent transmission block, so neighboring blocks can be in time or, for example, orthogonal frequency division multiplexing (OFDM). can be considered in frequency, such as over a set of tones in a multiplexing system. Thus, the precoders used in those blocks can be equal if the channel is fairly stationary and the codebook resolution is not very high. By increasing the resolution of the codebook or having a more dynamic channel, the precoders of adjacent blocks will be close, but no longer the same. The closeness between two precoders can be measured based on an appropriate distance index in space by all such precoders. Some examples of differential, dual and multi-resolution codebooks are disclosed in US Pat.

  A codebook that is efficient, that is, low feedback overhead, easy to store and search, and efficient across both uniform linear array (ULA) and cross-polarization configurations In order to obtain, we consider a precoding codebook design for a four transmit antenna (TX) MIMO downlink channel, for both uniform linear array (ULA) and cross-polarized antenna configurations. Details of the preferred codebook structure will be described. Several people have proposed codebook designs for specific antenna configurations (Non-Patent Document 1). The basic properties of the spatial correlation matrix we use are not used in the prior art. The codebook structure here is derived using the basic characteristics of the spatial correlation matrix under ULA and cross-polarized antenna configurations. Each precoding codeword is derived as the product of two matrices, which makes them efficient, lower feedback overhead for a given performance level, and more for a given feedback overhead. Achieve excellent performance.

US Patent Application Publication No. 2008/0232501 (M. A. Khojastepour et al., "STATIC AND DIFFERENTIAL PRECODING CODEBOOK FOR MIMO SYSTEMS") US Patent Application Publication No. 2009/0274225 (M. A. Khojastepour et al., "MULTI-RESOLUTION PRECODING CODEBOOK")

3GPP TS 36.213 V10.8.0 (2012-12), 3rd Generation Partnership Project; Technical Specification Group Radio Access Network; Evolved Universal Terrestrial Radio Access (E-UTRA); Physical layer procedures (Release 10), http: //www.3gpp .org / A. Forenza, D. Love and R. Heath, "Simplified Spatial Correlation Models for Clustered MIMO Channels With Different Array Configurations," IEEE Trans. Veh. Tech., July 2007 S. Loyka, "Channel capacity of MIMO architecture using the exponential correlation model," IEEE Commun. Letters, 2001 Ericsson, ST-Ericsson, "Design and Evaluation of 4 TX Precoder Codebooks for CSI Feedback," 3GPP TSG RAN WG1 R1-104847 62, Madrid, August 2010 D. Love, R. Heath and T. Strohmer, "Grassmannian beamforming for multiple-input multiple-output wireless systems," IEEE Trans. Inf. Theory, Oct. 2003 NEC Group, "DL MU-MIMO Enhancement Schemes," 3GPP TSG RAN WG1 R1-130364 NEC Group, "MU-MIMO: CQI Computation and PMI Selection," 3GPP TSG RAN WG1 R1-103832 NEC Group, "DL MU-MIMO enhancement via Residual Error Norm feedback," 3GPP TSG RAN WG1 R1-113874

  The object of the present invention is to provide a codebook for an efficient precoding codeword that requires lower feedback overhead for a given performance level and achieves better performance for a given feedback overhead. Is to provide.

  One aspect of the invention includes a method performed in a base station used in a wireless communication system. This method includes having one, two, three, and four layer codebooks for four transmit antenna (4TX) transmission, each codebook including a plurality of precoding matrices, and a plurality of precoding. Precoding the data using one of the matrices and transmitting the precoded data to the user equipment, each of the 1-layer and 2-layer codebooks having a first A codebook and a second codebook are included, and each precoding matrix in the first codebook has a first index and a second index.

  Another aspect of the present invention includes a method performed in a user equipment used in a wireless communication system. The method includes receiving precoded data from a base station, and each of a 1-layer, 2-layer, 3-layer, and 4-layer codebook for four transmit antenna (4TX) transmission includes a plurality of pre-coded data. Each of the one-layer and two-layer codebooks includes a first codebook and a second codebook, and each precoding matrix in the first codebook has a first index And a second index.

  Yet another aspect of the present invention includes a base station used in a wireless communication system. The base station includes a transmitter that transmits precoded data to a user equipment, and each of a 1-layer, 2-layer, 3-layer, and 4-layer codebook for transmitting four transmit antennas (4TX) Includes a plurality of precoding matrices, each of the 1-layer and 2-layer codebooks having a first codebook and a second codebook, wherein each precoding matrix in the first codebook is: It has a first index and a second index.

  Yet another aspect of the present invention includes a user equipment used in a wireless communication system. This user apparatus has a receiver that receives precoded data from a base station, and each of a 1-layer, 2-layer, 3-layer, and 4-layer codebook for transmitting four transmit antennas (4TX) is plural. Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook, and each precoding matrix in the first codebook has a first codebook And a second index.

  Yet another aspect of the present invention comprises a 1-layer, 2-layer, 3-layer, and 4-layer codebook for four transmit antenna (4TX) transmission, each codebook including a plurality of precoding matrices, A base station that pre-codes data using one of the pre-coding matrices and a user equipment that receives the pre-coded data from the base station, each of the 1-layer and 2-layer codebooks: A wireless communication system includes a first codebook and a second codebook, and each precoding matrix in the first codebook has a first index and a second index.

  Yet another aspect of the invention includes a method performed in a wireless communication system. This method includes precoding data and transmitting the precoded data from the base station to the user equipment, and includes one layer, two layers, three layers for transmitting four transmit antennas (4TX). Each of the layer and 4-layer codebooks includes a plurality of precoding matrices, each of the 1-layer and 2-layer codebooks having a first codebook and a second codebook, Each precoding matrix in the book has a first index and a second index.

  The first index may be for a plurality of subbands and the second index may be for each subband.

  The second codebook may comprise a legacy codebook or a householder codebook.

  Each of the 3 layer and 4 layer codebooks may have a legacy codebook or a householder codebook.

Each precoding matrix W in the first codebook may satisfy W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is the inner codebook. C ^{(1) is} chosen and the second matrix W ⁽²⁾ is chosen from the outer codebook.

Yet another aspect of the invention includes a method performed in a base station used in a wireless communication system. The method includes having a codebook including a plurality of precoding matrices, precoding data using one of the plurality of precoding matrices, and transmitting the precoded data to a user equipment. And each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is derived from the ^first codebook C ⁽¹⁾ The second matrix W ⁽²⁾ is selected from the second codebook.

Yet another aspect of the present invention includes a method performed in a user equipment used in a wireless communication system. The method includes receiving data precoded using one of a plurality of precoding matrices from a base station, wherein the codebook includes a plurality of precoding matrices, and each precoding matrix W is , W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ and the second matrix W ⁽²⁾ is Selected from the second codebook.

Yet another aspect of the present invention includes a base station used in a wireless communication system. The base station has a transmitter that transmits data precoded using one of a plurality of precoding matrices to a user equipment, the codebook includes a plurality of precoding matrices, and each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ and the second matrix W ⁽²⁾ Is selected from the second codebook.

Yet another aspect of the present invention includes a user equipment used in a wireless communication system. The user apparatus has a receiver that receives data precoded using one of a plurality of precoding matrices from a base station, and the codebook includes a plurality of precoding matrices, and each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ and the second matrix W ⁽²⁾ Is selected from the second codebook.

Yet another aspect of the present invention provides a base station having a codebook including a plurality of precoding matrices, precoding data using one of the plurality of precoding matrices, and precoding data from the base station. Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is the first codebook C ⁽ The second matrix W ⁽²⁾ chosen from ¹⁾ contains the wireless communication system chosen from the second codebook.

Yet another aspect of the invention includes a method performed in a wireless communication system. The method includes precoding data using one of a plurality of precoding matrices, and transmitting the precoded data from a base station to a user equipment, wherein the codebook includes a plurality of precoding matrices. Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾ , where the first matrix W ⁽¹⁾ is derived from the ^first codebook C ^(1). The second matrix W ⁽²⁾ is selected from the second codebook.

Fig. 2 shows a downlink multi-user MIMO system with N _T transmit antennas at the transmitter and N _R receive antennas at the receiver. A gain vector 3-bit codebook referring to g = [a ₁ , a ₂ , a ₃ , a ₄ ] is shown as a gain vector. The in-phase terms in the 8-PSK alphabet for rank 1 are shown. The in-phase terms in the 16-PSK alphabet for rank 2 are shown. Fig. 5 shows another in-phase term in the 16-PSK alphabet for rank 2. The in-phase terms in the 8-PSK alphabet for rank 2 are shown. The in-phase terms in the 8-PSK alphabet for rank 1 are shown. Fig. 5 shows another in-phase term in the 8-PSK alphabet for rank 1. The in-phase terms in the 24-PSK alphabet for rank 2 are shown. The in-phase terms in the 24-PSK alphabet for rank 2 are shown. The in-phase terms in the 12-PSK alphabet for rank 2 are shown. The in-phase terms in the 16-PSK alphabet for rank 2 are shown. The in-phase terms in the 16-PSK alphabet for rank 2 are shown.

FIG. 1 shows a downlink multi-user MIMO system with N _T transmit antennas at the BS and N _R receive antennas at the UE. A multi-antenna communication system 100 comprising a multi-level precoding codebook is shown schematically in FIG. The transmitter 110 includes t transmission antennas 111.1 to 111.n. t through the fading channel 130 and r receive antennas 121.1 to 121.12 connected to the receiver 120. to r. Channel estimator 125 provides an estimate of channel 130 to receiver 120. The channel estimate is also quantized and provided to transmitter 110 via quantization rate control feedback channel 135.

  In systems using beamforming techniques such as MIMO systems, a beamforming matrix (also called precoding matrix, precoder, codeword or precoding codeword) generated in response to sensed channel conditions is first received at the receiver. And quantized and then provided to the source transmitter (eg, via feedback). A conventional way to reduce the overhead associated with this feedback is to provide a matrix codebook at each of the transmitter and receiver. Each codebook has multiple or sets of potential beamforming matrices that can be used depending on the channel conditions sensed at the receiver. When the receiver has identified the appropriate matrix codebook, the receiver has one or more indices pointing to the appropriate codeword in the codebook stored at the transmitter (instead of the actual matrix entity). I will give you feedback.

Chapter I Example 1
Section 1 Uniform Linear Array In the following, we may assume co-polarized antennas in close proximity, unless otherwise noted.

We have the following findings from observations on uniform linear array (ULA) transmit antenna configurations. Consider a system with N co-polarized transmit antennas, and let C denote the transmit spatial correlation matrix. Let's define J to be a matrix with 0 everywhere except on the opposite corner element. That is, J = [J _{m, n} ]. here,

It is.

  Vector

If so, it is called Hermitian. We propose the following set of properties: The first finding relates to the spatial correlation matrix of the ULA transmission antenna arrangement, and it has a wide generality (see Non-Patent Document 2).

  Knowledge 1 The matrix C is a Hermitian Toeplitz matrix. That is, C is

Meet.

  Lemma 1 The eigenspace of any Hermitian Toeplitz matrix can be completely described by Hermitian vectors. In other words, assuming a Hermitian Toeplitz matrix A, x is the eigenvector of A such that Ax = λx. Then

It becomes.

  Lemma 2 Assume λ is the eigenvalue of Hermitian Toeplitz matrix A with algebraic redundancy of 1. Then, if x is an eigenvector such that Ax = λx, for some δ∈ [0,2π),

It becomes.

  A simplified model for the correlation matrix is an exponential correlation model (3), which is further explained in the appendix,

Given in. Here, ρ∈ [0,1] and θ∈ [0,2π).

2.1 4TX ULA
In this section, we consider the case of N = 4 co-polarized transmit antennas. First, without loss of generality, we can give the following structure for each eigenvector x of the spatial correlation matrix C:

Recalling that matrix C must be a Hermitian Toeplitz matrix and applying Lemma 1 and 2, we have

You can derive what you have to do.

  next,

Consider any two eigenvectors of the form (8) given by Then, the sufficient condition to make the orthogonality between these two eigenvectors real is:

Is to ensure. this is,

And can be simplified.

  We do not need equation (11), but scalar

You will find it useful for all possible values of.

Section 2 Polarized Settings Assume that the transmitter has 2N cross-polarized antennas, each consisting of a pair of N co-polarized antennas. Then, the correlation matrix of each one of these two co-polarization sets is represented as C, which is Hermitian and Toeplitz. Overall 2N x 2N correlation matrix

Can be written.

It can be seen that it has the form In addition, the matrix

It is. Here (.) ^T represents a transposition operation, and the phase term exp (jβ) can be ignored without loss of optimality. The two eigenvalues are 1 ± | α |

Also models the correlation matrix of two transmission ULAs.

Section 3 Codebook Structure We will now proceed to detail the codebook using the knowledge developed in sections 1 and 2. We detail in particular a subset of codebooks suitable for closely spaced 4TX ULA and cross-polarized antenna configurations, in addition to other configurations. We first consider a rank 1 codebook consisting of a set of 4 × 1 vectors. First, general structure without losing generality

think of. We will define three component codebooks from which the rank 1 codebook will be formed. The first one is called the gain vector codebook

Is expressed as P ₂ and P ₃ . Gain vector codebook

think of. In order to cover closely spaced 4TX ULAs, we need enough vectors in a rank 1 codebook with a structure of the form (8). With reference to g = [a ₁ , a ₂ , a ₃ , a ₄ ] as the gain vector, the 3-bit codebook of the gain vector in FIG. 2 is given. Here, for a settable scalar ψ> 0,

It is. Note that the gain vectors corresponding to the indices 0, 1, and 2 are suitable for the case of 4TX ULAs that are closely spaced according to the constraints in equation (8). The gain vector corresponding to index 0 is suitable for 4TX cross polarization. On the other hand, those corresponding to indexes 3 and 4 deal with a scenario referred to herein as the power imbalance case (see Appendix 8). Index 7 indicates reuse of the existing default codebook, and indexes 5 and 6 are included to simply provide other choices.

Next, in order to quantize the phase, we introduce _two phase codebooks P ₂ and P ₃ . Vector x

The constraint condition θ ₄ = θ ₂ + θ ₃ in equation (8) is assumed to be actual.

We select θ ₂ using code book P ₂ and select θ ₃ using code book P ₃ . A simple way to build these two codebooks is through [0, 2π] uniform quantization, using a given number of bits for each codebook. In this option of selecting the phase, if the gain vector corresponding to index 0 in the gain codebook of FIG. 2 is selected, the resulting vector is the general eigenvector of the correlation matrix of the 4TX cross polarization. It can be seen that it fits the structure.

  Next, we consider a rank 2 codebook consisting of a set of semi-unitary 4 × 2 matrices. From the findings in Section 1, we have the following structure

A subset of such a matrix with can be defined.

Note that this structure is compliant with 4TX ULA (see section 1) and using an exponential correlation model (as discussed in section 6) a ₁ = a _{1 ′} = a ₂ = a _{2 ′} It also conforms to the construction of the first two dominant eigenvectors of a 4TX ULA. This is also suitable for a 4TX cross polarization configuration (as discussed in Section 2). In addition we have the following structure according to 4TX ULA

Can also be included.

Section 4 Codebook structure in product form We next describe two codebook structures in which each codeword is derived as a matrix product, based on the principles outlined in sections 1 and 2 . In each case, we use the codebook designed in Non-Patent Document 4 as a basis and extend this codebook, following the principles outlined in Sections 1 and 2.

For n = 0,... 15, _let w _n = [1 exp (j2πn / 16)] ^T. We call this codebook the first embodiment and call its inner (broadband) codebook

It is defined as The rank 1 outer codebook is

Is defined. The outer rank 2 codebook is

Is defined.

For one configurable scalar ψ _q > 0, select one of the scalars a _q , b _q

.

In each feedback interval to select one rank-2 codeword for each subband, we first select one common matrix from the inner (wideband) codebook C ⁽¹⁾ ,

Denote the (final) codebook corresponding to rank 2, including all possible such final precoder selections. Similar procedures and notation for other ranks and inner precoders

Adopted for other choices.

Note that for a settable scalar {θ _q ∈ [0, 1]}, one selection of γ _q and d _q is to set d _q = γ _q / 2 = θ _q . Under this selection, we then set a triplet

Explain how to decide. From the description in Appendix 7 using an exponential correlation model, a good choice is to assume that (unquantized) θ is uniformly distributed in [0,1).

Say. Thus, a good strategy to obtain a finite set Θ = {θ} is to go through a uniform quantization of [0,1) using a given number of bits. For example, for 2 bits, Θ = {0, 1/4, 1/2, 3/4}.

Considering the choice of a _q , b _q , one possibility is to associate them with the variables p, q described in Appendix 7. Therefore, for the correlation magnitude parameter ρ = | a |

A finite set of values that can be obtained can be selected. For example, we can assume the set {1/2, 2/3, 3/4, 1} for the correlation magnitude parameter ρ = | a |. Next, using the formula in Appendix 7, we

To obtain a result equal to. Next, the triplet set {θ _q , a _q , b _q } is a Cartesian product.

It can be seen that is a size 16 or equivalently 4 bits. Another example is a set using Θ = {0, 1/5, 2/5, 3/5, 4/5} and only three values {2/3, 3/4, 1} for the correlation magnitude parameter.

May be used to obtain a Cartesian product of size 15. Another example is that the Cartesian product size is 16, Θ = {0, 1/8, 2/8,..., 7/8} and only two values {3/4, 1 for the correlation scale parameter. } Using}

It is obtained by using.

  We next consider an alternative codebook, hereinafter referred to as the second embodiment. The codeword of this codebook is also derived in product form. Inner wideband codebook

It is. Rank 1 outer codebook is

The rank 2 outer codebook is defined as

Is defined. In either case, the rank 3 and rank 4 codebooks can be fixed to the legacy (Halth holder) rank 3 and rank 4 codebooks. In addition, the entire legacy codebook can be included as a subset.

  The first embodiment has preferable properties that are lacking in the second embodiment. Its properties are for each rank k and the inner precoder

Assuming that each of the precoder matrices can be selected with equal probability, the expected value of each row norm square of the selected precoder matrix (ie, the sum of the magnitude squares of the elements of the row) is equal. This property is beneficial for operating the power amplifier (ie, controlling power amplifier backoff) and utilizing available transmit power.

4.1 Embedding in Codebooks Larger Note that channel matrix realization depends on both spatial correlation matrix and short-term (also known as “fast”) fading. In some scenarios, such as when the co-polarized antennas are widely spaced, there can be significant variations in the observed channel matrix due to fast fading. Therefore, a good codebook needs to accommodate such significant variations in the observed channel matrix due to fast fading, and this can also affect other judgments such as minimum chordal distance. The codebook needs to include codewords designed using criteria (Non-Patent Document 5). A useful way to deal with such a case would be to embed a codebook obtained using the principles described above into a larger codebook as a subset.

Section 5 Conclusion We have detailed the codebook structure and presented two embodiments that conform to the matrix product format. This structure is motivated by the basic nature of the spatial correlation matrix and makes possible codebook optimization.

Section 6 Appendix: 4TX ULA Using Exponential Correlation Model
Next, we have a correlation matrix

Consider the case of specialization. Note that the matrix C is Hermitian Toeplitz and is also completely characterized by one complex scalar. Therefore, the eigenvector can be expected to have more structures in addition to the one held by the eigenvector of a general Hermitian Toeplitz matrix. We will also utilize this additional structure described below. In this case, the matrix J is

Can be written.

  We first consider the case of | a | <1. In this case, the eigenvector of an arbitrary matrix of the form shown in Expression (24) has the following properties. Given an arbitrary matrix C of the form shown in equation (24),

Indicates four real number eigenvalues. Then

Is a diagonal matrix of the form The matrix S is

For some positive real scalars p, q, r, s such that

Have The matrix H is a 4 × 4 real Hadamard matrix. That is, the columns of H are orthogonal to each other, and all the elements belong to the set {± 1}. Therefore, since each column of E must satisfy the condition shown in the equation (5), each column of H = [h ₁ ,... H ₄ ] must satisfy the following condition.

Also, since E must be a unitary matrix, H must also satisfy the following additional conditions:

An important example of H is shown below.

Using H given above and recalling ρ = | a |, we can derive an expression that yields the scalars p, q, r, s as follows: First,

Get. Substituting this into equation (26), after several operations,

Get. Note that in the special case where ρ = 0, the correlation matrix C is reduced to the identity matrix, so that we can arbitrarily choose p, q, r, s (according to the respective norm constraints). Can do. Furthermore, for ρ <1, we

Can be determined.

  On the other hand, when ρ = 1, the matrix C is

Is a rank 1 matrix given by Then the eigenvector of C corresponding to one nonzero eigenvalue of C is

Can be shown. The choice of r, s can be arbitrary (according to norm constraints) since the associated eigenvalue is zero.

Section 7 Appendix: Adapting Power Imbalance A more general model for the spatial correlation of cross-polarized antenna configurations is shown below. Consider a transmitter with 2N cross-polarized antennas, each consisting of a pair of pairs, each with N co-polarized antennas. The correlation matrix of each of these two co-polarization sets is then represented by a matrix C that is Hermitian and Toeplitz. Overall 2N x 2N correlation matrix

Can be written.

Can be shown. Matrix

Can represent a 2 × 2 semi-definite matrix except for differences in scaling factors. Therefore, the 2 × 2 unitary matrix formed by the two eigenvectors can be any 2 × 2 unitary matrix. Next, in order to design a codebook suitable for such a scenario, we consider the first embodiment shown in Section 4 and extend its outer codebook C ⁽²⁾ as follows: To do.

  Rank 1 outer codebook is now

Is defined. The outer rank 2 codebook is now

Is defined.

  Similarly, for the second embodiment, the rank 1 outer codebook is

And the rank 2 outer codebook is

Is defined.

  Here, the code book defined above is a transmitter having four geographically spaced co-polarized antennas, each consisting of a pair of two co-polarized antennas for each location. It is also suitable for this case. Thus, each one of these two co-polarization sets is given by C, which is Hermitian Toeplitz. 4x4 correlation matrix as a whole

Indicates the Kronecker product, and d ₁ , d ₂ ∈ [0, 1] are normalized gain terms reflecting different average propagation path gains from the two locations.

Chapter II Example 2
Codebook structure in product form:
Next, based on the principle derived from the above, we describe a structured codebook structure in which each codeword is derived as a matrix product.

For n = 0,..., N−1, _let w _n = [1 exp (j2πn / N)] ^T. We define the inner (broadband) codebook as follows: First, for some positive integer K, J, L,

Define Here, K is called a step, J is called a width, and L is called an extent. These parameters are typically

Selected to meet. Next we have the inner (broadband) codebook

That is, D (q2) is a vector whose main diagonal is a vector

Is a diagonal matrix containing

It is.

  For a given q2, q1, and

We can introduce an overlap between {W ⁽¹⁾ (k, q1)} for successive choices of k. In particular,

By guarantee, W ⁽¹⁾ several columns of (k, q1) is ^{W (1) (k + 1} , q1) can be seen to be the same as those in the. This is a useful property to have in the inner wideband codebook because the correlation in time or frequency changes gradually. However, for some m

It is not necessary to hold when To introduce such overlap between different inner codewords in such cases, we first

It can be ensured that W ⁽¹⁾ (k, q1) and the sequence of W ⁽¹⁾ (k + 1, q1 ′) overlap.

  Rank 1 outer codebook is

Can be defined as We say here that W ^(2,1) (r, s) for a given r, s can be any one of the four shown vectors. Here, e _i represents a J × 1 column selection vector for selecting the i-th column in the J × J unit matrix. To limit size, only certain combinations (r, s), referred to herein as “feasible” combinations, will be allowed. Here, r = s can be a feasible combination. For any subband, the final codeword of rank 1 is

Formed by selecting the inner codeword D (q2) W ⁽¹⁾ (k, q1) from C ⁽¹⁾ together with the outer codeword W ^(2,1) (r, s) from q2) Obtain the final codeword for that subband as W ⁽¹⁾ (k, q1) W ^(2,1) (r, s). The inner codeword selection may be common across all subbands.

  Next, to extend the selectability in each subband, we can move D (q2) to the outer codebook. In other words, the inner wideband codebook

And the outer subband rank 1 codebook for a feasible combination (r, s, q2)

Can be defined as In all the cases described above, the outer codebook can be made dependent on the selection of the inner codeword. In other words, each feasible combination of (r, s) in equation (2-4) and (r, s, q2) in equation (2-5) itself may be a function of inner codeword selection. it can. In other words, two different inner codewords can select codewords from the outer codebook with different possible combinations. In each case, the set of feasible combinations for each selection of inner codewords is predefined and known to all users and base stations.

  Next, consider the case of rank 2. The first possibility is to maintain the inner codebook defined in equation (2-2) with the following outer subband rank 2 codebook that is invariant to inner codeword selection.

Only certain combinations of (r, s) may be allowed to limit size. Note that since the set of allowed combinations is common across all selections of inner codewords, for each allowed (r, s), D (q2) W ⁽¹⁾ ( The columns of k, q1) W ^(2,2) (r, s) must be orthogonal to each other. Note that r = s is one such selection that ensures orthogonality for each selection of inner codeword D (q2) W ⁽¹⁾ (k, q1).

In order to extend the set of possible final rank 2 codewords without undue overhead, we can generate allowed combinations depending on the choice of inner codewords. In particular, an outer subband codebook can be defined which depends on the inner codeword selection D (q2) W ^{(1) (} identified by the indices q1, q2, k ⁾ . This outer subband codebook is

A code word represented by W ^(2,2) (r, s, k, q1, q2) is included. Here, r, sε {1,..., J}. The phase θ (r, s, k, q1, q2) along with the allowed combination (r, s) is the resulting final codeword D (q2) W ⁽¹⁾ (k, q1) W ^{(2 , 2) It} must be ensured that the two columns (r, s, k, q1, q2) are orthogonal. Note that due to the structure of our inner codebook, it is sufficient that this phase term is a function of only r, s and q1, so that we write the phase term as θ (r, s, q1). be able to. To allow more choices in the outer codebook, D (q2) can be moved to the outer codebook, similar to that made in the rank 1 case. In other words, for a feasible combination (r, s, q2)

An inner wideband codebook with an outer subband rank 2 codebook with codewords of the form

Can be defined as

  For further expansion of the set of rank 2 codewords, we can otherwise ensure orthogonality between the resulting final codeword sequences. Assuming that the inner codebook is defined as in equation (2-2) (because we can apply a given step after a simple change,

And skips the case where D (q2) has been moved to the outer codebook). Next, assume that the rank-2 outer codebook depends on the selection of the inner codeword and includes a codeword of the form shown in equation (2-7). Furthermore, regarding the selection of the inner codeword D (q2) W ⁽¹⁾ (k, q1),

A codeword of the form Preferably, such an operator is such that if the vector x has a constant magnitude property that all its elements have the same magnitude,

Also has such a property. An example of such an operator is HH (x, t), which is tε {2,3,4} and any unit norm whose first element is real and strictly less than 1 -Householder conversion for vector x

Gives the t th column of the 4 × 4 unitary matrix obtained via. For our structure, x = D (q2) W ⁽¹⁾ (k, q1) e (r, s) satisfies two conditions necessary to define the householder transformation. Also, if the vector x has a constant magnitude characteristic, HH (x, t) also has that characteristic.

  Other examples of such operators are

Is to set. Note that when x has a constant magnitude characteristic, a diagonal matrix D (x) is constructed so that its non-zero entries have unit magnitude (magnitude),

And a constant magnitude characteristic with respect to the vector D (x) Px.

  Note that different sets of inner codewords (ie inner codebooks) for rank 1, rank 2 and other ranks are defined, although all of these have the general structure shown in equation (2-2). can do. In this way, an inner codebook specific to the rank can be defined. Recall that the outer codebook for each subband can already be a function of the inner codeword. In the code book described above, de-duplication may be performed if necessary. In particular, if there are any two inner codewords that yield an equivalent set of rank r codewords per final subband for any rank r, then one of these two inner codewords. Only must be kept in the inner codebook for that rank r. Where one of the two final codewords is identical to the other, except for a difference by right-multiplication with a matrix that is column permutation and / or diagonal and whose non-zero entry has unit size. For example, these two final codewords are equivalent.

  Note that having a larger rank 2 inner codebook than the rank 1 codebook is beneficial for MU-MIMO. Since only one inner codeword needs to be reported for all subbands, a larger inner codebook allows for better quantization resolution without excessively increasing feedback. In general, users under MU-MIMO transmission are served with a lower rank than reported, so better resolution for rank 2 and higher benefits for SU-MIMO. It not only brings benefits, but also benefits MU-MIMO. In this case, better resolution will ensure that the column subset extracted from the user's reported precoder is also efficient, i.e., allows MU-MIMO gain with sufficient accuracy. Let's go.

Chapter III Example 3
In the 11th edition (Release 11 (Rel-11)) LTE cellular network, the network can semi-statically configure multiple CSI processes for the same user. Rel-12 and beyond users are required to support both legacy 4TX codebooks as well as improved 4TX codebooks. As alluded to above, these two codebooks can be viewed as two subsets (components) of a larger codebook. Furthermore, for each CSI process, a separate codebook subset limitation can be applied. A useful inevitable conclusion from these two findings is that the network can configure the codebook separately for each CSI process (in terms of targeted users) in a semi-static form. That is, for each CSI process of a given user, the network can configure the element (ie, legacy or improved) codebook that the user uses. In addition, further codebook subset restrictions can be applied on a per CSI process basis, assuming element codebook selection for those processes. In order to reduce the signaling overhead, we propose to apply only one latter codebook subset limitation per CSI process. As a result, even if the CSI process (or equivalently the mode defined for that CSI process) requires the user to report the precoding matrix (ie PMI) for each subband, All such reported matrices must adhere to the constructed (common) subset limitation for that process.

Codebook structure in product form We first present a general codebook structure in which each codeword is derived as a matrix product. For convenience, we ignore the 1/2 normalization factor. For n = 0,..., N−1, W _n = [1 exp (j2πn / N)] ^T represents a 2 × 1 beam vector, and the inner (wideband) codebook is

It is defined as Where {a _k } is a real-valued scalar,

It is. Note that the (angle) separation between any two adjacent beam vectors in a particular inner codeword A (q) W ⁽¹⁾ (k) is 2π / N, so that N and J are Both determine the angular range (angular span) of the phase term of each inner codeword. Intuitively, a larger angular range makes it possible to generate an appropriate codebook for fading scenarios with less correlation. On the other hand, the scalar {d _p } has two inner code words A (q) W ⁽¹⁾ (k) and A (q ′) W ⁽¹⁾ (for a certain q, q′ε {1,. assists in controlling the separation between the phase terms in any two beam vectors associated with k). Intuitively, such a small separation would be beneficial for taking advantage of correlations in time and frequency.

  Next, the rank 1 outer (subband) codebook is

Is defined. Here, e _i is the i-th J × 1 column selection vector (ie, the i-th column of the J × J unit matrix), and exp (jθ _{s, i} ) is a co-phaing term. is there. The (maximum) size of the rank 1 codebook is therefore JS. Smaller sizes can be obtained by selecting only a subset of all possible such vectors. Co-phase terms can be obtained by optimizing appropriate metrics such as the average chordal distance after limiting them in the M-PSK alphabet. Here, a positive integer M ≧ 1 is a design parameter. Optimization can be limited to ensure that the minimum angular separation is maintained between the cophase terms. For rank 2, the outer (subband) codebook is

Is defined. Note that different pairs (m, p) and (m ′, p ′) can have different numbers of co-phase terms. The cophase terms can be obtained by optimizing an appropriate metric such as the average chordal distance after limiting them in the M′-PSK alphabet. Here, the positive integer M ′ ≧ 1 is a design parameter and can be different from M. Optimization can be limited to ensure that the minimum angular separation is maintained between the cophase terms.

Next we propose two specific embodiments. Both embodiments have a 4-bit wideband codebook. Respect to the first embodiment, N = L = 8, J = 2, a k = k, k = 0, ..., 7 Use and _{d p = {0,1 / 16}} , the C ⁽¹⁾ To construct. Thus, each inner codeword is a 4 × 4 matrix. The corresponding subband codebook is 3 bits in size for both rank 1 and rank 2. The co-phase terms in the rank 1 codebook are placed in the 8-PSK alphabet, which is given in FIG. The notation adopted in the table in FIG. 3 is that if the entry corresponding to a certain (s, i) is t, θ _{s, i} = 2πt / M. Here, M = 8 for 8-PSK. For rank 2 codebooks, in addition to the co-phase terms given in FIG. 4A or 4B, we have (m, p) ε {(1,1), (2,2), (1,2)} Select. Alternatively, the cophase term can be selected as shown in FIG. Note that for beam combination (1,2) in FIG. 5, we use a further cophasing option.

For the second embodiment, N = 16, L = 8, J = 4, a _k = 2k, k = 0,..., 7 and d _p = {0, 1/32} and C ^{(1 )} . Thus, each inner codeword is a 4 × 8 matrix. The corresponding subband codebook is 4 bits in size for both rank 1 and rank 2. The co-phase term in the rank 1 codebook is placed in the 8-PSK alphabet, which is given in FIG. 6A or 6B. For the rank 2 codebook, in addition to the co-phase terms given in FIG. 7, we have (m, p) ε {(1,1), (2,2), (3,3), (4, 4), (1,2), (1,4), (2,3), (2,4)} are selected. Alternatively, we can select the co-phase term as shown in FIG. 8A or FIG. 8B. Alternatively, we can select the co-phase term as shown in FIG. 9 or FIG.

  The rank 3 and rank 4 codebooks may be fixed to the legacy (householder) rank 3 and rank 4 codebooks. All codeword matrices in the above-described codebook satisfy the constant magnitude characteristic.

Chapter IV Example 4
The user is configured to use the legacy 4TX codebook in one of the user's CSI processes, and that process (or equivalently the mode defined for that CSI process) gives the user per-subband Other related issues arise when requesting to report a precoding matrix. Here, when the user's preferred rank is 3 or 4, the size of the legacy codebook (4 bits for both rank 3 and rank 4) is too much for per-subband reporting. It may be. In other words, such users have experienced a good average SINR and will typically have been scheduled alone on their assigned resources, so there is no visible impact on performance. Thus, the feedback can be reduced. To achieve feedback reduction, the network defines subsampled versions of legacy codebooks for ranks 3 and 4 when the user's desired rank is 3 or 4, and these subsampling The user can be configured to report codewords from the version codebook. A subsampled rank 3 codebook is obtained by removing one or more codewords from the rank 3 legacy codebook, while one or more codewords from the rank 4 legacy codebook. Is removed to obtain a subsampling rank 4 codebook. These subsampled codebooks are defined by the network and communicated to all users in advance. Another way to provide more flexibility is to take advantage of codebook subset limitation. Here, it is assumed that the size of the rank 3 codebook (per subband) is limited to M codewords. The network then determines a subset containing no more than M codewords from the legacy rank 3 codebook (semi-static and possibly user specific) and communicates this subset to the user. be able to. The user then limits its search (for rank 3 codewords) to this subset on each subband. In order to report the preferred codeword to the user on each subband, the user employs lexicographic ordering (labeling). That is, the codeword with the smallest index in the indicated subset (as the original rank 3 legacy codebook) is assigned a new index 1 and within the indicated subset (original rank 3 legacy codebook). The codeword with the second smallest index (as codebook) is assigned the new index2. This process continues until all codewords in the subset are assigned a new index. Obviously, the new index ranges from 1 to M ′, where M ′ ≦ M. Also, since the subset is common across all subbands, the new index set is also common across all subbands, and therefore the new index set has to be determined only once by the user. The user then reports the new index of the selected precoder on each subband. The same procedure can be applied to rank 4 as well, where the value M can be different for rank 4 and rank 3.

  Finally, additional feedback can be incorporated into the CSI process (or a mode that is equivalently defined for that CSI process) to improve MU-MIMO performance. As detailed in our past work (Non-Patent Document 6), a user can also MU-CQI along with its single-user (SU) channel state information (CSI) report. Can be reported. This SU-CSI (including wideband or per subband PMI and per subband CQI) is calculated using the pilot and resource elements configured for that CSI process for interference measurements. Several ways to calculate these MU-CQIs are detailed in our past work (7), and one of these ways is (if so configured, sub- After estimating (or emulating) a co-scheduled set of co-scheduled interferers (based on the band), the user is given the SU-MIMO rules determined in the user's SU-CSI report. The MU-CQI is calculated using the PMI determined by use (hereinafter referred to as base PMI). Here, the set of commonly scheduled interfering PMIs (ie, transmission precoders assigned to other commonly scheduled users) on the subband assumed by the user is a function of the base PMI determined by the user. It is. Each set of commonly scheduled interfering PMIs that the user must assume can be configured by the network in a semi-static (and possibly user-specific) form. The size of the interfering PMI set (for each selection of base PMI) can be greater than one. To reduce overhead, the resulting MU-CQI calculated on a subband basis is 1 (as detailed in our work on broadband residual error norm feedback (1)). Can be combined as two (or at most two) wideband MU-CQIs, which are then reported. To further improve performance, multiple such sets of interfering PMIs can be configured (for each base PMI). The user may then report one (or at most two) wideband MU-CQI for each configured set of interfering PMIs and use differential feedback to reduce the feedback overhead. it can. Alternatively, the process described above can be repeated for several selections of base PMI, and the user can use one specific base PMI (using appropriate selection rules such as those that maximize the expected MU gain). And report it with the associated MU-CQI.

  We return to the following codebook structure.

Codebook structure in product form We first show the construction of a general codebook in which each codeword is derived as a matrix product. For n = 0,..., N−1, _let w _n = [1 exp (j2πn / N)] ^T denote a 2 × 1 beam vector, and the inner (wideband) codebook

It is defined as Where {a _k } is a real-valued scalar,

It is. Note that the (angle) separation between any two adjacent beam vectors in a particular inner codeword A (q) W ⁽¹⁾ (k) is 2π / N, so that N (which is the granularity) (referred to as (qranularity)) and J (which is equal to the number of beam vectors per inner codeword) together determine the angular range of the phase term of each inner codeword. Intuitively, a larger angular range (which has a smaller N for a given J (ie, with a smaller granularity or with a larger 2π / N), or with a given N Can be achieved by having a larger J), making it possible to generate a suitable codebook even for fading scenarios with less correlation, and robust against timing alignment errors It will also provide (robustness). However, the cost of increasing J is the larger size of each outer subband codebook, while the smaller N prevents beam vector localization at a given inner codeword. As such, performance in cross-polarized configurations that are closely spaced can be degraded. On the other hand, the scalar {d _q } (referred to as the staggering factor) is the two inner codewords A (q) W ⁽¹⁾ (k) and A for some q, q′∈ {1,. (Q ′) W ⁽¹⁾ assists in controlling the separation between any two phase terms associated with (k). Intuitively, such a small separation would be beneficial for taking advantage of correlations in time and frequency.

  An extension to the inner codebook described above is to use two (or more) granularity sets. Here, each granularity can have its own set of wandering factors. This will typically increase the size of the wideband codebook, but can better satisfy different antenna configurations. Next we have such a composite inner (broadband) codebook for I choices of different granularities.

Is described. Note that J remains fixed over different granularities. In certain scenarios (with very low correlation), it may be advantageous to select at least one of the granularities so that two or more beam vectors in many of the associated inner codewords are orthogonal to each other. Let's go.

  Next, the rank 1 outer (subband) codebook is

Is defined. Here, e _i is the i-th J × 1 column selection vector (that is, the i-th column of the J × J unit matrix), and exp (jθ _{s, i} ) is a cophase term. The (maximum) size of the rank 1 codebook is therefore JS. Smaller sizes can be obtained by selecting only a subset of all possible such vectors. Co-phase terms can be obtained by optimizing appropriate metrics such as the average chordal distance after limiting them in the M-PSK alphabet. Here, a positive integer M ≧ 1 is a design parameter. Optimization can be limited to ensure that the minimum angular separation is maintained between the cophase terms. For rank 2, the outer (subband) codebook is

Is defined. Note that different pairs (m, p) and (m ′, p ′) can have different numbers of co-phase terms. Such co-phase terms can be obtained by optimizing appropriate metrics, such as the average chordal distance, after limiting them into the M′-PSK alphabet. Here, the positive integer M ′ ≧ 1 is a design parameter and can be different from M. Optimization can be limited to ensure that the minimum angular separation is maintained between the cophase terms.

  As described above, (1) we have the key structure that each eigenvector of the spatial correlation matrix must have under the ULA transmit antenna structure, and each eigenvector of the spatial correlation matrix under the cross-polarized transmit antenna configuration Identified key structure that must be present. (2) Next we realized the specified structure in at least one subset of the precoding codebook to ensure good performance. (3) We have also presented an embodiment that takes into account the identified structure and is also efficient.

  It should be understood that the foregoing is illustrative and illustrative in all respects and not restrictive, and that the scope of the invention disclosed herein is derived from the detailed description. It is not to be determined, but rather to be determined from the claims as interpreted according to the full breadth permitted by patent law. It should be understood that the embodiments shown and described are merely illustrative of the principles of the invention and that various modifications and changes can be made by those skilled in the art without departing from the scope and spirit of the invention. It is. Those skilled in the art could implement various other feature combinations without departing from the scope and spirit of the invention.

This application is filed on March 7, 2013, with US Provisional Application No. 61 / 774,275 entitled “Observations on Codebook Construction,” filed March 8, 2013, and “ US Provisional Application No. 61 / 775,058 entitled `` Observations on Codebook Construction '' and `` Enhancements to a Structured Codebook '' filed April 5, 2013 US Provisional Application No. 61 / 808,934 entitled “Enhancement to the 4 Transmit Antenna Precoding Codebook” filed April 29, 2013 Provisional Application No. 61 / 817,150 and US Provisional Application entitled “Enhancement to the 4 Transmit Antenna Precoding Codebook” filed on April 29, 2013 61 / 817,247 and May 2013 Claims all the benefits of US Provisional Application No. 61 / 821,989 entitled “Improvements to the 4 Transmit Precoding Codebook” filed on the 10th, The contents are incorporated herein by reference.

Claims (49)

A method performed in a base station used in a wireless communication system, comprising:
Having a 1-layer, 2-layer, 3-layer and 4-layer codebook for four transmit antenna (4TX) transmission, each codebook including a plurality of precoding matrices;
Precoding data using one of the plurality of precoding matrices;
Transmitting the precoded data to a user equipment;
Have
Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook;
Each precoding matrix in the first codebook has a first index and a second index.   The method of claim 1, wherein the first index is for a plurality of subbands and the second index is for each subband.   The method of claim 1, wherein the second codebook comprises a legacy codebook or a householder codebook.   The method of claim 1, wherein each of the three-layer and four-layer codebooks comprises a legacy codebook or a householder codebook. Each precoding matrix W in the first codebook satisfies W = W ⁽¹⁾ W ⁽²⁾ ,
Here, the first matrix W ⁽¹⁾ is selected from the inner codebook C ⁽¹⁾ ,
The method of claim 1, wherein the second matrix W ⁽²⁾ is selected from an outer codebook. A method performed in a user equipment used in a wireless communication system, comprising:
Receiving precoded data from the base station;
Each of the 1-layer, 2-layer, 3-layer, and 4-layer codebooks for four transmit antenna (4TX) transmission includes a plurality of precoding matrices;
Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook;
Each precoding matrix in the first codebook has a first index and a second index.   The method of claim 6, wherein the first index is for a plurality of subbands and the second index is for each subband.   The method of claim 6, wherein the second codebook comprises a legacy codebook or a householder codebook.   The method of claim 6, wherein each of the three-layer and four-layer codebooks comprises a legacy codebook or a householder codebook. Each precoding matrix W in the first codebook satisfies W = W ⁽¹⁾ W ⁽²⁾ ,
Here, the first matrix W ⁽¹⁾ is selected from the inner codebook C ⁽¹⁾ ,
The method of claim 6, wherein the second matrix W ⁽²⁾ is selected from an outer codebook. A base station used in a wireless communication system,
A transmitter for transmitting precoded data to a user equipment;
Each of the 1-layer, 2-layer, 3-layer, and 4-layer codebooks for four transmit antenna (4TX) transmission includes a plurality of precoding matrices;
Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook;
Each precoding matrix in the first codebook has a first index and a second index. A user apparatus used in a wireless communication system,
A receiver for receiving precoded data from a base station;
Each of the 1-layer, 2-layer, 3-layer, and 4-layer codebooks for four transmit antenna (4TX) transmission includes a plurality of precoding matrices;
Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook;
Each precoding matrix in the first codebook has a first index and a second index, user equipment. 1 code, 2 layers, 3 layers and 4 layers codebook for 4 transmit antenna (4TX) transmission, each codebook containing a plurality of precoding matrices, and one of the plurality of precoding matrices is A base station to pre-code data using,
A user equipment receiving the precoded data from the base station;
Have
Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook;
Each wireless communication system, wherein each precoding matrix in the first codebook has a first index and a second index. A method performed in a wireless communication system, comprising:
Precoding the data,
Transmitting the precoded data from a base station to a user equipment;
Have
Each of the 1-layer, 2-layer, 3-layer, and 4-layer codebooks for four transmit antenna (4TX) transmission includes a plurality of precoding matrices;
Each of the 1-layer and 2-layer codebooks has a first codebook and a second codebook;
Each precoding matrix in the first codebook has a first index and a second index.   15. The method of claim 14, wherein the first index is for a plurality of subbands and the second index is for each subband.   15. The method of claim 14, wherein the second codebook comprises a legacy codebook or a householder codebook.   The method of claim 14, wherein each of the three-layer and four-layer codebooks comprises a legacy codebook or a householder codebook. Each precoding matrix W in the first codebook satisfies W = W ⁽¹⁾ W ⁽²⁾ ,
Here, the first matrix W ⁽¹⁾ is selected from the inner codebook C ⁽¹⁾ ,
The method of claim 14, wherein the second matrix W ⁽²⁾ is selected from an outer codebook. A method performed in a base station used in a wireless communication system, comprising:
Having a codebook containing multiple precoding matrices;
Precoding data using one of the plurality of precoding matrices;
Transmitting the precoded data to a user equipment;
Have
Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾
Here, the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ ,
The method, wherein the second matrix W ⁽²⁾ is selected from a second codebook.

And
J, K and L are positive integers,

20. The method of claim 19, wherein The second codebook for rank 1 transmission

Can be expressed as
Here, r, s = {1,..., J}, q2ε {1..., Q2}, γ _q2 = [0, 1] ∀q2,
21. The method of claim 20, wherein e _i represents a J × 1 column selection vector that selects the i th column of a J × J identity matrix.   The method according to claim 21, wherein parts of the feasible combinations (r, s, q2) are allowed. The second codebook for rank 2 transmission

Can be expressed as
Where r, s = {1,..., J},
21. The method of claim 20, wherein e _i represents a J × 1 column selection vector that selects the i th column of a J × J identity matrix.   24. The method of claim 23, wherein parts of possible combinations (r, s) are allowed.   24. The method of claim 23, wherein r = s. The second codebook for rank 2 transmission

Can be expressed as
Where r, sε {1,... J}, q2ε {1..., Q2},
21. The method of claim 20, wherein e _i represents a J × 1 column selection vector that selects the i th column of a J × J identity matrix.   27. A method according to claim 26, wherein parts of possible combinations (r, s, q2) are allowed.   21. The method of claim 20, wherein A (q1) = B (q1). 21. The method of claim 20, wherein ₂ [pi] ( _{dq + 1- dq} ) is small compared to 2 [pi] / N.   The method according to claim 19, wherein the codebook consists of a plurality of precoding matrices.   20. The method of claim 19, wherein the first matrix is common across multiple subbands and the second matrix is for each subband.   20. The method of claim 19, wherein the codebook is for four transmit antenna (4TX) transmission.   The method of claim 19, wherein J = 4. A method performed in a user equipment used in a wireless communication system, comprising:
Receiving data precoded using one of a plurality of precoding matrices from a base station;
A codebook includes the plurality of precoding matrices;
Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾
Here, the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ ,
The method, wherein the second matrix W ⁽²⁾ is selected from a second codebook. A base station used in a wireless communication system,
Having a transmitter for transmitting data precoded using one of a plurality of precoding matrices to a user equipment;
A codebook includes the plurality of precoding matrices;
Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾
Here, the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ ,
The second matrix W ⁽²⁾ is a base station selected from the second codebook. A user apparatus used in a wireless communication system,
Having a receiver for receiving data pre-coded using one of a plurality of pre-coding matrices from a base station;
A codebook includes the plurality of precoding matrices;
Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾
Here, the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ ,
The second matrix W ⁽²⁾ is a base station selected from the second codebook. A base station having a codebook including a plurality of precoding matrices and precoding data using one of the plurality of precoding matrices;
A user equipment receiving the precoded data from a base station;
Have
Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾
Here, the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ ,
The second matrix W ⁽²⁾ is a wireless communication system selected from the second codebook. A method performed in a wireless communication system, comprising:
Precoding data using one of a plurality of precoding matrices;
Transmitting the precoded data from a base station to a user equipment;
Have
A codebook includes the plurality of precoding matrices;
Each precoding matrix W satisfies W = W ⁽¹⁾ W ⁽²⁾
Here, the first matrix W ⁽¹⁾ is selected from the ^first codebook C ⁽¹⁾ ,
The method, wherein the second matrix W ⁽²⁾ is selected from a second codebook.

And
J, K and L are positive integers,

20. The method of claim 19, wherein The second codebook for rank 1 transmission

Can be expressed as
Where r, s = {1,..., J},
40. The method of claim 39, wherein e _i represents a J × 1 column selection vector that selects the i th column of the J × J identity matrix.   41. The method of claim 40, wherein parts of feasible combinations (r, s) are allowed.   41. The method of claim 40, wherein r = s. The second codebook for rank 2 transmission

Can be expressed as
Where r, s = {1,..., J},
40. The method of claim 39, wherein e _i represents a J × 1 column selection vector that selects the i th column of the J × J identity matrix.   44. The method of claim 43, wherein portions of feasible combinations (r, s, q2) are allowed. For each allowed (r, s), a column of D (q2) W ⁽¹⁾ (k, q1) W ^(2,2) (r, s) is a choice of each of the first matrices. 44. The method of claim 43, wherein each is orthogonal to each other.   44. The method of claim 43, wherein r = s. The second codebook for rank 2 transmission

Can be expressed as
Where r, s ∈ {1, ..., J},
40. The method of claim 39, wherein e _i represents a J × 1 column selection vector that selects the i th column of the J × J identity matrix. 48. The method of claim 47, wherein the two columns of D (q2) W ⁽¹⁾ (k, q1) W ^(2,2) (r, s, k, q1, q2) are orthogonal. 40. The method of claim 39, wherein ₂ [pi] ( _{dq + 1- dq} ) is small compared to 2 [pi] / N.

Images