CN101321013B - System and method for transmission of data - Google Patents

Abstract

The invention discloses a data transmission system and a data transmission method, wherein the data transmission system comprises a data transmitting device, for predesigning the precoding codebook set and transmitting the codebook set to the data receiving device based on a non-linear receiver, and selecting the corresponding codebook for data transmission after receiving the codebook keynumber feedback by the data receiving device; and a data receiving device, for performing the channel estimation after receiving the codebook set from the data transmitting device, selecting the precoding codebook set for transmitting the data according to the result of the channel estimation, and feedbacking the corresponding codebook of the precoding codebook to the data transmitting device. The device has better diversity and the multiplexing gain than the linear receiving under almost the same complexity and the linear receiving.

Description

Data transmission system and data transmission method

Technical field

The present invention relates to the communications field, relate more specifically to a kind of data transmission system and data transmission method.

Background technology

Because traditional SISO of comparing of MIMO (multiple-input and multiple-output) (the single output of single input), the huge advantage aspect channel capacity and anti-wireless fading channel is acknowledged as the key technology of next generation wireless communication.Many antennas can bring huge diversity gain, spatial multiplexing gain and array gain.To the application of many antennas mainly aspect three: the one, aspect Space Time Coding, set about, to improve diversity gain and coding gain, such as STBC (Space Time Coding), SFBC (space-frequency coding) encoding scheme etc.; The 2nd, consider from improving the capacity aspect, increase spatial multiplexing gain to improve the capacity of system, as adopting VBLAST (Vertical Bell laboratory layering space-time) etc.; The 3rd, consider from the aerial array aspect, improve array gain to anti-fading, such as Beamforming.

Now know, when transmit leg is known channel information, by regulating transmit leg, can further improve performance, be i.e. closed loop (Close-Loop) design.For closed loop design, also be divided into different design standards: as, have plenty of the capacity of raising system, have plenty of the BER of raising system etc.In the situation that allows based on feedback link to exist, better performance when closed loop design can reach than open loop.

But, because the anxiety of Radio Resource, so be the problem that needs are considered for the design of feedback quantity.Feed back huge information and seem unrealistic.In the process of research closed loop, have the research based on channel Complete Information amount (Perfect CSIT), have the research based on channel statistical value feedback quantity.These all need receiver real-time tracking channel, and feedback pretreatment parameter at random is to transmit leg.Recently, the design to code book occurs, and drawn a lot of achievements.

At present the research of code book mainly concentrated on two large classes: the vector codebooks design and the empty time-code code book design that improves the diversity spatial multiplexing gain that improve the Beamforming of array gain.

In the empty time-code code book design of the raising system diversity spatial multiplexing gain that exists at present, many code book designs that is based on linear reception programme (high ML receives except complexity) of usefulness.Such as the code book design based on ZF or MMSE receiver.These designs mainly are to utilize side's property such as space of the right singular matrix of channel, certain design standard (as the maximization mean square error the MMSE trace of a matrix, maximization Shannon capacity etc.) on, ask the maximization of certain subspace distance guassmann stream shape, or utilize the Lloyd algorithm directly at space search.

But system adopts linear reception programme, and performance can be restricted to a certain extent.In the situation without the quantification restriction, nonlinear precoding mode is better than linear reception programme.Non-linear receiver noted here is not the ML receiver, but adopts the non-linear reception of MMSE-VBLAST scheme.So be necessary to study the code book design of non-linear receiver.

Summary of the invention

One or more problems in view of the above the invention provides a kind of data transmission system data transmission method of unifying.

Data transmission system according to the present invention comprises: data sending device, be used for designing in advance the precoding codebook collection and the code book collection being sent to data sink based on non-linear receiver, and behind the code book index that receives data sink feedback number, select corresponding code book to carry out data and send; And data sink, be used for behind the code book collection that receives from data sending device, carrying out channel estimating, the precoding codebook of selecting the transmission of data to use according to channel estimation results, and the code book index that selected precoding codebook is corresponding number feeds back to data sending device.

Wherein, data sending device comprises: the code book design module is used for according to non-linear Receiver Design code book collection; And data transmission blocks, be used for selecting corresponding precoding codebook to carry out data and sending according to the code book index of data sink feedback number.

Wherein, the code book design module comprises: the singular value decomposition module is used for matrix $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Carry out singular value decomposition, obtain singular value Σ; Conversion module is used for matrix is carried out bilateral displacement and Given conversion, obtains the upper triangular matrix R that equates with the diagonal element of needs, $Q_0^T\mathrm{\Σ}{P}_0=R,$ Wherein, Σ is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value; And the code book generation module, be used for combination $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value decomposition U _BΣ V _B ^*, decompose in conjunction with GMD $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]=Q_B{\mathrm{RP}}_B*,$ Can obtain P _B=V _BP ₀, with P _BAs precoding codebook, wherein, P _BFor $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ The right matrix that obtains after decomposing of GMD.

Wherein, the precoding codebook concentrated of precoding codebook is mutually orthogonal.Be multi-input multi-output system according to data transmission system of the present invention.

Data transmission method according to the present invention may further comprise the steps: S202, and data sending device designs in advance the precoding codebook collection and the code book collection is sent to data sink based on non-linear receiver; S204, data sink carries out channel estimating behind the code book collection that receives from data sending device, the precoding codebook of selecting the transmission of data to use according to channel estimation results, and the code book index that selected precoding codebook is corresponding number feeds back to data sending device; And S206, behind the code book index that receives data sink feedback number, data sending device is selected corresponding code book to carry out data to send.

Wherein, step S202 comprises: S2022, and to matrix $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Carry out singular value decomposition, obtain singular value Σ; S2024 carries out bilateral displacement and Given conversion to matrix, obtains the upper triangular matrix R that equates with the diagonal element of needs, $Q_0^T\mathrm{\Σ}{P}_0=R,$ Wherein, Σ is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value; And S2026, in conjunction with $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value decomposition U _BΣ V _B ^*, decompose in conjunction with GMD $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]=Q_B{\mathrm{RP}}_B*,$ Can obtain P _B=V _BP ₀, with P _BAs precoding codebook, wherein, P _BFor $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ The right matrix that obtains after decomposing of GMD.

Wherein, the precoding codebook concentrated of precoding codebook is mutually orthogonal.Be used for multi-input multi-output system according to data transmission method of the present invention.

The present invention is with non-linear standard design code book out, receives in the much the same situation in complexity and linearity, can reach than linearity to receive better diversity and spatial multiplexing gain.

Description of drawings

Accompanying drawing described herein is used to provide a further understanding of the present invention, consists of the application's a part, and illustrative examples of the present invention and explanation thereof are used for explaining the present invention, do not consist of improper restriction of the present invention.In the accompanying drawings:

Fig. 1 is the block diagram of MIMO feedback pre-coding system;

Fig. 2 is the flow chart according to the data transmission method of the embodiment of the invention; And

Fig. 3 is the block diagram according to the data transmission system of the embodiment of the invention.

Embodiment

The below's minute a plurality of parts describe the code book design based on non-linear reception programme among the present invention in detail.

1.MIMO system model

Consider that a transmitting antenna is M _t, reception antenna is M _rThe non-selective channel of frequency, the baseband equivalence model after the sampling is:

$\stackrel{\→}{y}=\mathrm{HF}\stackrel{\→}{x}+\stackrel{\→}{z}---\left(1\right)$

$\stackrel{\→}{x}\∈C^{L\×1}$ Be the information symbol that sends, $F\∈C^{M_t\×L}$ Be Linear precoding matrix, $\stackrel{\→}{y}\∈C^{M_r\×1}$ Be the signal that receives, $H\∈C^{M_r\×M_t}$ Be the channel matrix of K for order.Suppose $E(\stackrel{\→}{x}\stackrel{\→}{x}*)={\mathrm{\σ}}_x^2{I}_L,$ $\stackrel{\→}{z}\∈N(0,{\mathrm{\σ}}_z^2{I}_M)$ Be the multiple Gaussian noise of Cyclic Symmetry.

For simply, suppose that mimo channel is uncorrelated fading channel, in same data block, channel response is constant, and in different masses, channel response is for independently.

The signal noise ratio of transmitting terminal (Signal Noise Ratio is called for short SNR) is defined as:

Suppose

Be Gaussian random vector, then the Shannon capacity of mimo channel is:

$C={\log }_2\frac{|{\mathrm{\σ}}_z^{2}I+{\mathrm{\σ}}_x^2\mathrm{HFF}*H*|}{\left|{\mathrm{\σ}}_z^2{I}_L\right|}---\left(3\right)$

If CSIT and CSIR can get, then by selecting pre-coding matrix F energy maximized system capacity.Make the singular value decomposition (SVD) of H be H=U Λ V ^*, wherein, Λ is K * K diagonal matrix,

Non-zero singular value for H.Then select F to be:

F＝ΛΦ ^1/2 (4)

Wherein, Φ is diagonal matrix, its diagonal entry θ _kBe the water injection rate to the K sub-channels:

${\mathrm{\θ}}_k\left(u\right)={(u-\frac{\∂}{{\mathrm{\λ}}_{H.k}^2})}^{+};---\left(5\right)$

Select u to satisfy ${\mathrm{\σ}}_x^2\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{\mathrm{\θ}}_k\left(u\right)={\mathrm{\ρ\σ}}_z^2,$ ρ is the SNR of transmitting terminal.This moment, the heap(ed) capacity of system was:

$C_{\mathrm{IT}}=\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{\log }_2(1+{\mathrm{\λ}}_{H,k}^2\frac{{\mathrm{\θ}}_k}{\mathrm{\α}})---\left(6\right)$

Y.Jiang has proposed a kind of based on channel unified channel decomposing (Uniform ChannelDecomposition, abbreviation UCD) combined transceiving processing mode, adopt the receive mode of UCD-VBLAST can make system both reach the capacity of (6) formula, can obtain very high diversity gain again, be a kind of extraordinary scheme.The present invention studies a kind of design of code book on the basis of this transmitting-receiving processing scheme, reduce the feedback quantity of system.The below does a brief introduction to the UCD-VBLAST scheme.

2.UCD-VBLAST

The closed expression of A.MMSE-VBLAST

Concerning the detection scheme of MMSE-VBLAST, the ZF vector of every layer data is:

${\stackrel{\→}{w}}_i={(\underset{j=1}{\overset{i}{\mathrm{\Σ}}}{\stackrel{\→}{h}}_j{\stackrel{\→}{h}}_j*+\mathrm{\αI})}^{-1}{\stackrel{\→}{h}}_i,i=1,...,M_t---\left(7\right)$

Every layer SINR can be calculated as after detecting:

${\mathrm{\ρ}}_i=\frac{{|{\stackrel{\→}{h}}_i*{\stackrel{\→}{w}}_i|}^2{\mathrm{\σ}}_x^2}{{\stackrel{\→}{w}}_i*(\underset{j=1}{\overset{i-1}{\mathrm{\Σ}}}{\mathrm{\σ}}_x^2{\stackrel{\→}{h}}_i{\stackrel{\→}{h}}_j*+{\mathrm{\σ}}_z^{2}I){\stackrel{\→}{w}}_i};---\left(8\right)$

Now study the closed expression of w, structural matrix $H_a={\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}{I}_{M_t}\end{array}\right]}_{(M_r+M_t)\×M_t},$ Then carry out QR and decompose, obtain:

$H_a=Q_{H_a}{R}_{H_a}=\left[\begin{array}{c}{Q}_{H_a}^u\\ Q_{H_a}^l\end{array}\right]R_{H_a}---\left(9\right)$

Prove the ZF vector in (7)

Satisfy lower relation of plane:

${\stackrel{\→}{w}}_i=r_{\mathrm{Ha},\mathrm{ii}}^{-1}{\stackrel{\→}{q}}_{\mathrm{Ha},i},i=\mathrm{1,2},...,M_t---\left(10\right)$

Wherein, r _{Ha, ii}Be R _HaDiagonal element,

For

Column vector.Notice, $H=Q_{\mathrm{Ha}}^u{R}_{\mathrm{Ha}}$ The QR that is not H decomposes,

Also non-row orthogonal matrix.By top conversion, just the Nonlinear M MSE-VBLAST testing process of complexity can be converted into a linearity test process.Calculate when receiving vector (matrix), can avoid the matrix inversion process of MMSE complexity.Receiving matrix linear after the equivalence is:

$W_{\mathrm{equal}}=Q_{\mathrm{Ha}}^u{R}_{\mathrm{Ha}-\mathrm{diag}}^{-1}---\left(11\right)$

R _Ha-diagBe diagonal matrix, element value is corresponding R _HaThe value of the diagonal entry of matrix, its inverse matrix are exactly that each element is got inverse.

(10) are brought into (8), can obtain lower relation of plane:

$\mathrm{\α}(1+{\mathrm{\ρ}}_i)=r_{\mathrm{Ha},\mathrm{ii}}^2,i=\mathrm{1,2},...,M_t---\left(12\right)$

In addition, not during water filling, MMSE-VBLAST is without capacitance loss, namely at transmit leg

$\underset{i=1}{\overset{M_t}{\mathrm{\Σ}}}{\log }_2(1+{\mathrm{\ρ}}_i)={\log }_2|H*H{\mathrm{\α}}^{-1}+I|---\left(13\right)$

B.GMD (geometric mean decomposition)

Can be seen that by formula (12) that affect the MMSE-VBLAST algorithm performance mainly is smallest tier SNR during layering detects, namely the upper triangular matrix R in (9) formula _HaThe minimum value of diagonal entry.When open-loop design, MMSE-VBLAST can not change R _HaValue.But, can design a kind of pre-coding matrix F, so that the maximization of the minimum value of the diagonal entry of the R that the QR of equivalent channel matrix HF decomposes? the GMD resolution of matrix addresses this problem.

At first see following optimization problem:

$\underset{Q,P}{\max }\min \{r_{\mathrm{ii}}:1\≤i\≤K\}$

Constraints: QRP*=H, Q*Q=I, P*P=I,

r _Ij=0, to i〉j, R ∈ R ^{K * K}(14)

Here, K is the order of H.

The geometric mean decomposition of matrix (Geometric mean decomposition is called for short GMD) is the globally optimal solution of optimization problem in (14).

The GMD of matrix decomposes as follows:

If matrix H ∈ is C ^{M * n}Order be K, then H can be decomposed into following form H=QRP ^*, wherein, P and Q are the row orthogonal matrix, and R is upper triangular matrix, and its diagonal entry is the geometric average of its positive singular value, that is:

$r_{\mathrm{ii}}=\stackrel{\‾}{\mathrm{\σ}}={\left(\underset{{\mathrm{\σ}}_j>0}{\mathrm{\Π}}{\mathrm{\σ}}_j\right)}^{1/K},1\≤i\≤K---\left(15\right)$

Can find out that the cornerwise value of R equates, for the long-pending geometric mean of non-zero singular value, so method is called GMD.

C.UCD-VBLAST

The GMD that MMSE-VBLAST is used channel matrix H decomposes, and has proposed the MMSE-VBLAST processing method based on precoding of a kind of UCD-VBLAST of crying.

Decompose in conjunction with the MMSE-VBLAST that introduces previously and GMD, the precoding F that revises in (4) is (16) formula:

F＝ΛΦ ^1/2Ω ^* (16)

Wherein, Ω ∈ C ^{L * K}Be unitary matrix, Ω * Ω=I for fear of capacitance loss, generally selects L〉K.With (16) substitution (3) as can be known, it still can make the Shannon capacity maximum (during the transmitting terminal water filling) of system.

Associating (16) and (1), the pseudo channel that obtains this mimo system is:

G＝HF＝UΛφ ^1/2Ω ^*＝UΣΩ ^* (17)

Wherein, Σ=Λ Φ ^1/2For diagonal element is

Diagonal matrix.

According to above-mentioned method, structural matrix:

$G_a=\left[\begin{array}{c}\mathrm{HF}\\ \sqrt{a}{I}_L\end{array}\right]=\left[\begin{array}{c}\mathrm{U\Σ}{\mathrm{\Ω}}^{*}\\ \sqrt{a}{I}_L\end{array}\right]=\left[\begin{array}{c}\mathrm{U\Σ}:0_{K\×(L-K)}{{\mathrm{\Ω}}_0}^{*}\\ \sqrt{a}{I}_L\end{array}\right]=\left[\begin{array}{c}I\\ {\mathrm{\Ω}}_0\end{array}\right]\left[\begin{array}{c}\mathrm{U\Σ}:0_{K\×(L-K)}\\ \sqrt{a}{I}_L\end{array}\right]{{\mathrm{\Ω}}_0}^{*}---\left(18\right)\begin{array}{}\end{array}$

Wherein, Ω ₀Be K * K unitary matrix, its front K classifies Ω as.

GMD decomposes:

$J=\left[\begin{array}{c}\mathrm{U\Σ}:0_{K\×(L-K)}\\ \sqrt{a}{I}_L\end{array}\right]=Q_J{R}_J{P}_J---\left(19\right)$

So make Ω ₀=P _J ^*, so just can use the MMSE-VBLAST method, and every layer SNR is identical, also be the optimal solution under (14) condition.

More than be the main thought of UCD-VBLAST, a kind of algorithm that reduces operand is:

Definition

Be diagonal matrix, its diagonal element is:

${\widetilde{\mathrm{\σ}}}_i=\sqrt{{\mathrm{\σ}}_i^2+\mathrm{\α}},1\≤i\≤K---\left(20\right)$

${\widetilde{\mathrm{\σ}}}_i=\sqrt{\mathrm{\α}},K+1\≤i\≤L---\left(21\right)$

Matrix

GMD be decomposed into:

$\widetilde{\mathrm{\Σ}}=\mathrm{QRP}*---\left(22\right)$

Pre-coding matrix can directly be calculated as:

F＝V[Φ ^1/2：0 _K×(L-K)]P (23)

Wherein, [Φ ^1/2: 0 _{K * (L-K)}] be K * L matrix.

$Q_G^u=U[{\mathrm{\Φ}}^{1/2}:0_{K\×(L-K)}]\widetilde{\mathrm{\Σ}}Q---\left(24\right)$

To sum up can obtain one about the computational process of UCD-VBLAST:

The first step: the SVD decomposing H of calculating H=U Λ V ^*

Second step: calculate Φ ^1/2

The 3rd step: calculate Σ=Λ Φ ^1/2

The 4th step: obtain

, right

Do the GMD decomposition and obtain Q, P;

The 5th step: calculate pre-coding matrix F according to formula (22);

The 6th step: calculate receiving end processing array Q according to formula (23) _G

The 7th step: detect.

3. code book design

In above-mentioned feedback, need feeding back precoding matrix F, feedback overhead is larger, when particularly antenna number increases.For save radio resources, be necessary it is carried out the design of code book.

The Limited Feedback amount of supposing channel is N (bit), and then system only needs 2 ^NIndividual pre-coding matrix T (code book), codebook set can be expressed as:

T _set：＝{T ₁，..，T _N}； (26)

When communication occured, the recipient at first estimated current channel status, then according to certain standard, selected optimum code book in codebook set, and the sequence number that will choose at last code book sends transmit leg to.Two subject matters that exist in the middle of this are: design codebooks collection T _SetProblem and recipient's typical problem of how selecting optimum code book.

Some rules of design codebooks are as follows:

Orthogonality; Namely

$T_i^H{T}_i=I,i=1,...,N;---\left(27\right)$

Based on orthogonality, different design standards is arranged; For the receiving algorithm of MMSE-VBLAST, can select the SNR of each layer as one of standards of measurement.

In addition, can see, when N=∞, can not directly select F in (16) formula as code book, because F does not satisfy orthogonality.

At this moment, employing quadrature principle is the convenience in order to design, and can bring preferably numerical stability, and this is very important when designing in system.Because the problem of constraint, the loss that inevitably can bring capacity and performance, article will be done certain comparison with rate distortion function to the loss of performance in the back.

In the design of finite rate code book, should be so that the subspace distance of different code books be maximum.

This is a quantification problem in essence.But for the distance of How to choose subspace, must decide according to different design criterions.As, if select MMSE mark criterion, then select to make the matrix 2 norm subspace distances of two code books maximum; If select MMSE determinant criterion, then selection makes their Fubini-Study subspace distance maximum etc.

4. the code book based on UCD-VBLAST designs

UCD-VBLAST is a kind of detection scheme based on the spatial reuse form, and the single channel wave beam that is different from based on the design of finite rate code book forms, and the code book design not only will be carried out the vector quantization design, must consider the orthogonality between the matrix code book column vector.Fig. 1 has shown the general frame of reponse system:

Select the standard of precoding codebook matrix F to be based on the non-linear receiver (MMSE-VBLAST) that channel UCD decomposes.According to different pre-coding matrix Criterion of Selectings, can carry out in the steps below respectively the code book design.

Design criterion: when satisfying the last MMSE-VBLAST of employing detection, minimum SNR is maximum.Should satisfy following condition:

$\underset{T\∈T_{\mathrm{set}}}{\max }\min \{r_{\mathrm{ii}}:1\≤i\≤K\}$

Constraints: $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]T=QR,T*T=I,Q*Q=I$

r _Ij=0, to i〉j, R ∈ R ^{K * K}(29)

With reference to figure 2, the data transmission method according to the embodiment of the invention is described.As shown in Figure 2, this data transmission method may further comprise the steps: S202, and data sending device designs in advance the precoding codebook collection and the code book collection is sent to data sink based on non-linear receiver; S204, data sink carries out channel estimating behind the code book collection that receives from data sending device, the precoding codebook of selecting the transmission of data to use according to channel estimation results, and the code book index that selected precoding codebook is corresponding number feeds back to data sending device; And S206, behind the code book index that receives data sink feedback number, data sending device is selected corresponding code book to carry out data to send.

Wherein, step S202 comprises: S2022, and to matrix $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Carry out singular value decomposition, obtain Σ; S2024 does bilateral displacement and Given conversion to singular value matrix, the upper triangular matrix R that the diagonal element that obtains needing equates, that is: $Q_0^T\mathrm{\Σ}{P}_0=R,$ Wherein, Σ is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value; S2026, in conjunction with $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value decomposition U _BΣ V _B ^*Decompose with GMD $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]=Q_B{\mathrm{RP}}_B*,$ Can obtain P _B=V _BP ₀, the solution of T optimum is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ The right matrix P that obtains after decomposing of GMD, so T _Opt=V _BP ₀

With reference to figure 3, the data transmission system that is used for realizing method shown in Figure 2 is described.As shown in Figure 3, this data transmission system comprises: data sending device 302, be used for designing in advance the precoding codebook collection and the code book collection being sent to data sink based on non-linear receiver, and behind the code book index that receives data sink feedback number, select corresponding code book to carry out data and send; And data sink 304, be used for behind the code book collection that receives from data sending device, carrying out channel estimating, the precoding codebook of selecting the transmission of data to use according to channel estimation results, and the code book index that selected precoding codebook is corresponding number feeds back to data sending device.

Wherein, data sending device 302 comprises: code book design module 3022 is used for according to non-linear Receiver Design code book collection; And data transmission blocks 3024, be used for selecting corresponding precoding codebook to carry out data and sending according to the code book index of data sink feedback number.

Wherein, code book design module 3022 comprises: the singular value decomposition module is used for matrix $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Carry out singular value decomposition, obtain singular value Σ; Conversion module is used for matrix is carried out bilateral displacement and Given conversion, obtains the upper triangular matrix R that equates with the diagonal element of needs, $Q_0^T\mathrm{\Σ}{P}_0=R,$ Wherein, Σ is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value; And the code book generation module, be used for combination $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value decomposition U _BΣ V _B ^*, decompose in conjunction with GMD $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]=Q_B{\mathrm{RP}}_B*,$ Can obtain P _B=V _BP ₀, with P _BAs precoding codebook, wherein, P _BFor $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ The right matrix that obtains after decomposing of GMD.Wherein, the precoding codebook concentrated of precoding codebook is mutually orthogonal.

In sum, the present invention is with non-linear standard design code book out, receives in the much the same situation in complexity and linearity, can reach than linearity to receive better diversity and spatial multiplexing gain.

The above is the preferred embodiments of the present invention only, is not limited to the present invention, and for a person skilled in the art, the present invention can have various modifications and variations.Within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (9)

1. data transmission system is characterized in that comprising:

Data sending device, be used for designing in advance the precoding codebook collection and described code book collection being sent to data sink based on non-linear receiver, and behind the code book index that receives described data sink feedback number, selecting corresponding code book to carry out data sends, wherein, described non-linear receiver is the non-linear receiver that adopts the MMSE-VBLAST scheme; And

Described data sink, be used for behind the code book collection that receives from described data sending device, carrying out channel estimating, the precoding codebook of selecting the transmission of data to use according to channel estimation results, and the code book index that selected precoding codebook is corresponding number feeds back to described data sending device.

2. data transmission system according to claim 1 is characterized in that, described data sending device comprises:

The code book design module is used for according to non-linear Receiver Design code book collection; And

Data transmission blocks is used for selecting corresponding precoding codebook to carry out data and sending according to the code book index of described data sink feedback number.

3. data transmission system according to claim 2 is characterized in that, described code book design module comprises:

The singular value decomposition module is used for matrix $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Carry out singular value decomposition, obtain the singular value ∑, wherein, H is for using channel matrix;

Conversion module is used for described matrix is carried out bilateral displacement and Given conversion, obtains the upper triangular matrix R that equates with the diagonal element of needs,

Wherein, ∑ is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value, P and Q are the row orthogonal matrix, R is upper triangular matrix; And,

The code book generation module is used for combination $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value decomposition U _B∑ V _B ^*, decompose in conjunction with GMD $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]=Q_B{{\mathrm{RP}}_B}^{*},$ Can obtain P _B=V _BP ₀, with P _BAs precoding codebook, wherein, P _BFor $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ The right matrix that obtains after decomposing of GMD.

4. data transmission system according to claim 3 is characterized in that, the precoding codebook that described precoding codebook is concentrated is mutually orthogonal.

5. each described data transmission system in 4 according to claim 1 is characterized in that described data transmission system is multi-input multi-output system.

6. a data transmission method is characterized in that, may further comprise the steps:

S202, data sending device designs in advance the precoding codebook collection and described code book collection is sent to data sink based on non-linear receiver, and wherein, described non-linear receiver is the non-linear receiver that adopts the MMSE-VBLAST scheme;

S204, described data sink carries out channel estimating behind the code book collection that receives from described data sending device, the precoding codebook of selecting the transmission of data to use according to channel estimation results, and the code book index that selected precoding codebook is corresponding number feeds back to described data sending device; And

S206, behind the code book index that receives described data sink feedback number, described data sending device is selected corresponding code book to carry out data to send.

7. data transmission method according to claim 6 is characterized in that, described step S202 comprises:

S2022 is to matrix $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Carry out singular value decomposition, obtain the singular value ∑, wherein, H is for using channel matrix;

S2024 carries out bilateral displacement and Given conversion to described matrix, obtains the upper triangular matrix R that equates with the diagonal element of needs,

Wherein, ∑ is $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value, P and Q are the row orthogonal matrix, R is upper triangular matrix; And,

S2026, in conjunction with $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ Singular value decomposition U _B∑ V _B ^*, decompose in conjunction with GMD $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]=Q_B{{\mathrm{RP}}_B}^{*},$ Can obtain P _B=V _BP ₀, with P _BAs precoding codebook, wherein, P _BFor $\left[\begin{array}{c}H\\ \sqrt{\mathrm{\α}}I\end{array}\right]$ The right matrix that obtains after decomposing of GMD.

8. data transmission method according to claim 7 is characterized in that, the precoding codebook that described precoding codebook is concentrated is mutually orthogonal.

9. each described data transmission method in 8 according to claim 6 is characterized in that described data transmission method is used for multi-input multi-output system.

Images