CN102025404B - Codebook generation method and device - Google Patents

Abstract

The invention provides a codebook generation method and device. The method comprises the following steps: acquiring an initial codebook set and feeding back; and determining to modify the initial codebook at real time according to channel performance information obtained from estimation. Seen from the codebook modification-based adaptive codebook generation method that on the basis of not changing the existing codebook basis matrix, a group of new codebook modification matrixes are generated according to the performance statistical property of a system, feedback expenditure is considered to modify the initial codebook set, and codebook accuracy is adjusted timely according to the real-time change of the channel, so that the codebook capacity is expanded and the codebook accuracy is improved under the condition of increasing little feedback expenditure, thereby more effectively improving precoding performance in the case of limited feedback.

Description

A kind of method for generating codebooks and device

Technical field

The present invention relates to multiple-input and multiple-output (MIMO, Multiple Input Multiple Output) technology, under espespecially a kind of Limited Feedback condition, the method for generating codebooks based on precoding and device.

Background technology

In mimo system, introduce precoding and can obtain larger power system capacity and the availability of frequency spectrum of Geng Gao.But the realization of precoding is all generally condition based on transmitting terminal known channel state information (CSI, Chancel StateInformation).At symmetrical time division duplex (TDD, Time Double Division) in system, reception and transmitting hocket, now receiving terminal is also transmitting terminal, if the transformation time receiving and launch is less than the coherence time of channel, the channel condition information of estimating in the time receiving can be used for the preliminary treatment of next x time to transmitted signal.But, when forward link in system and reverse link when asymmetric as Frequency Division Duplexing (FDD) (FDD, Frequency Double Division) system, just need to quantize channel parameter or beam forming vector etc., to adapt to the band-limited actual conditions of feedback channel.

In order to reduce feedback information volume, relatively one of effective method is precoding (or beam forming) scheme adopting based on matrix (or vector) code book, is called limited feedback precoding method.Limited feedback precoding method roughly comprises: first design one group of pre-coding matrix or beam vector of optimizing according to the statistical property of channel, form a code book, this code book is all known at transmitting terminal and receiving terminal; Receiving terminal selects a code word in code book to be used as transmitting terminal pre-coding matrix or beam forming vector according to the channel matrix of estimating and certain performance criteria, and the sequence number in code book feeds back to transmitting terminal by it.

The design of code book is a very crucial link in the design of MIMO pre-coding system, and a lot of documents have also provided method for designing and the various code word Criterion of Selecting of various precoding codebooks.In 3G standard evolution process, the transmission precoding MIMO technology based on code book has been adopted as alternative.

In code book design, need two topmost problems that solve to be: how quantized channel space, the i.e. problem of design codebooks collection; And how recipient is according to proximity, selects the problem of current code book.In Zhe Liangge question essence, complement each other, having designed code book collection receiving terminal could be according to proximity, selects the current code book that needs feedback; And had just proximity, just can design optimum code book collection.

Solve above two problems, from the angle of vector quantization, key point is exactly to design suitable quantitative criteria.The same with the Precoding Design based on Complete Information, the Precoding Design based on partial information also needs to have design standard.Corresponding with it, design standard mainly contains two large classes, volumetric(al) standards and the error rate (BER, Bit Error Rate) standard.But no matter any standard, between the expense of feeding back and performance, all can there is a contradiction, therefore, on code book design and thus the impact of the mode of the mode on precoding and feedback be the hot issue that research institution is studied for this field always.

Summary of the invention

In view of this, main purpose of the present invention is to provide a kind of method for generating codebooks, can the in the situation that of Limited Feedback, effectively improve the performance of precoding.

Another object of the present invention is to provide a kind of code book generating apparatus, can the in the situation that of Limited Feedback, effectively improve the performance of precoding.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of method for generating codebooks, comprising:

Obtain initial codebook set feedback;

According to the channel performance information of estimating to obtain, determine initial codebook to revise in real time.

Described channel performance is Signal to Interference plus Noise Ratio SINR; Described estimation channel performance information is:

For each code stream, its SINR is for below shown in formula:

${\mathrm{SINR}}_i\left(n\right)=\frac{{\left|\right|H_i{V}_i\left(n\right)\left|\right|}^2}{{\mathrm{\σ}}_i^2+\underset{k=1,k\≠i}{\overset{K}{\mathrm{\Σ}}}{\left|\right|H_i{V}_k\left|\right|}^2+{\left|\right|H_i{V}_i(:)\left|\right|}^2};$

Wherein, SINR _i(n) SINR of n code stream of i user of expression, H _iand σ _i ²be respectively i user's channel matrix and noise power, V _i(n) be the precoding codebook vector for n code stream of user i, V _i(:) is the precoding codebook vector of user i except n code stream, V _kfor the precoding codebook vector of interference user, the number that K is all interference users.

Described decision is modified in real time to initial codebook:

Described SINR value between any two code streams does not meet | SINR _i(n)-SINR _i(m) | when≤ε, wherein, ε is a default smaller value, and the energy difference between any two code streams that represent to allow, adopts correction matrix P, according to the following formula described initial codebook set is revised:

$\tilde{V}={\mathrm{VP}}^k$ k＝0，1，2，…N _t-1，

Wherein, V is original initial codebook set, and P is correction matrix, and k is modifying factor, N _tfor number of transmit antennas,

for revised new codebook set;

Correction matrix $P=\left[\begin{array}{ccccc}0& 0& \mathrm{\Λ}& 0& 1\\ 1& 0& \mathrm{\Λ}& 0& 0\\ 0& 1& 0& 0& 0\\ \mathrm{\Λ}& \mathrm{\Λ}& \mathrm{\Λ}& \mathrm{\Λ}& \mathrm{\Λ}\\ 0& 0& \mathrm{\Λ}& 1& 0\end{array}\right],$ The order of wherein supposing channel matrix is n.

Described decision is modified in real time to initial codebook:

Described SINR value between any two code streams does not meet | SINR _i(n)-SINR _i(m) | when≤ε, wherein, ε is a default smaller value, the energy difference between any two code streams that represent to allow, the SINR on Dui Ge road carries out balance.

The SINR on described Dui Ge road carries out balance and comprises:

The situation that is 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{cc}\sin \left(\mathrm{\&theta;}\right)& 0\\ 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right],$ Wherein θ is a selectable angle, and span is 0≤θ < pi/2;

The situation that is greater than 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{ccccc}\sin \left(\mathrm{\&theta;}\right)& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& \sin \left(\mathrm{\&theta;}\right)+\mathrm{\&Delta;}& \mathrm{\&Lambda;}& 0& 0\\ 0& 0& \mathrm{\&Lambda;}& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right],$

Wherein, increment factor Δ meets $0<\mathrm{\&Delta;}<(\frac{1}{\sin \left(\mathrm{\&theta;}\right)}-\sin \left(\mathrm{\&theta;}\right)),$ Be an adjustable factors, θ is a selectable angle, and span is 0≤θ < pi/2;

V is original initial codebook set,

for revised new codebook set.

Described channel performance is signal to noise ratio snr.

Described decision is modified in real time to initial codebook:

Described SNR value between any two code streams does not meet | SNR _i(n)-SNR _i(m) | when≤ε ', wherein, ε ' is a default smaller value, and the energy difference between any two code streams that represent to allow, adopts correction matrix P, according to the following formula described initial codebook set is revised:

$\tilde{V}={\mathrm{VP}}^k$ k＝0，1，2，…N _t-1，

Wherein, V is original initial codebook set, and P is correction matrix, and k is modifying factor, N _tfor number of transmit antennas,

for revised new codebook set;

Correction matrix $P=\left[\begin{array}{ccccc}0& 0& \mathrm{\&Lambda;}& 0& 1\\ 1& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& 1& 0& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 1& 0\end{array}\right],$ The order of wherein supposing channel matrix is n.

Described decision is modified in real time to initial codebook:

Described SINR value between any two code streams does not meet | SINR _i(n)-SINR _i(m) | when≤ε, wherein, ε is a default smaller value, the energy difference between any two code streams that represent to allow, the SINR on Dui Ge road carries out balance;

The SINR on described Dui Ge road carries out balance and comprises:

The situation that is 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{cc}\sin \left(\mathrm{\&theta;}\right)& 0\\ 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right],$ Wherein θ is a selectable angle, and span is 0≤θ < pi/2;

The situation that is greater than 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{ccccc}\sin \left(\mathrm{\&theta;}\right)& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& \sin \left(\mathrm{\&theta;}\right)+\mathrm{\&Delta;}& \mathrm{\&Lambda;}& 0& 0\\ 0& 0& \mathrm{\&Lambda;}& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right],$

Wherein, increment factor Δ meets $0<\mathrm{\&Delta;}<(\frac{1}{\sin \left(\mathrm{\&theta;}\right)}-\sin \left(\mathrm{\&theta;}\right)),$ Be an adjustable factors, θ is a selectable angle, and span is 0≤θ < pi/2;

V is original initial codebook set,

for revised new codebook set.

Described channel performance is error rate BER;

Described decision is modified in real time to initial codebook:

When the described system BER estimating higher than set in advance threshold value time, adopt following formula described initial codebook set is revised: $\tilde{V}={\mathrm{\&Phi;}}^{k}V$ k＝0，1，2，…N _t-1；

Wherein, V is original initial codebook set, and Φ is correction matrix, and k is modifying factor, N _tfor number of transmit antennas,

for revised new codebook set, correction matrix Φ is shown below:

$\mathrm{\&Phi;}=\left[\begin{array}{cccc}{e}^{\mathrm{j\&pi;}\frac{1}{N_t^2}}& 0& \mathrm{\&Lambda;}& 0\\ 0& e^{\mathrm{j\&pi;}\frac{2}{N_t^2}}& \mathrm{\&Lambda;}& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& 0\\ 0& 0& \mathrm{\&Lambda;}& e^{\mathrm{j\&pi;}\frac{n}{N_t^2}}\end{array}\right];$

Wherein, the order that n is channel, N _tfor number of transmit antennas.

A kind of code book generating apparatus, comprises initial codebook set generation unit, estimation unit, comparing unit and amending unit, wherein,

Initial codebook set generation unit, for obtaining initial codebook set feedback, and exports to amending unit by the initial codebook set obtaining;

Estimation unit, for estimating channel performance information, and exports to comparing unit;

Comparing unit, for according to the channel performance information from estimation unit, in the time that the difference between any two code streams that estimation obtains does not meet default non-correction conditions, the correction result notice sending to amending unit is for showing to need correction; In the time that the difference between any two code streams that estimation obtains meets default non-correction conditions, the correction result notice sending to amending unit is for showing not need correction;

Amending unit, for receiving the correction result notice from comparing unit, when correction result notice is when showing to revise, to revising from the initial codebook set of initial codebook set generation unit, and exports revised codebook set; When correction result notice is when showing not need to revise, will export as codebook set from the initial codebook set of initial codebook set generation unit

The technical scheme providing from the invention described above can be found out, comprises and obtains initial codebook set feedback, according to the channel performance information of estimating to obtain, determines initial codebook to revise in real time.From the present invention is based on the adaptive codebook generation method of code book correction, do not changing on the basis of existing code book basis matrix, according to the performance statistics characteristic of system, generate one group of new code book correction matrix, and consider original initial codebook set to revise in conjunction with feedback overhead, adjusted in time the precision of code book according to the real-time change of channel, in the situation that feedback overhead increase is very little, expand the capacity of code book, improve the precision of code book, thereby the in the situation that of Limited Feedback, more effectively improve the performance of precoding.

Brief description of the drawings

Fig. 1 is the flow chart of method for generating codebooks of the present invention;

Fig. 2 is the composition structural representation of code book generating apparatus of the present invention.

Embodiment

Fig. 1 is the flow chart of method for generating codebooks of the present invention, as shown in Figure 1, comprises the following steps:

Step 100: obtain initial codebook set feedback.

In this step, be initial codebook matrix set with the Basic codebook that each large standard (as 3GPP LTE, IEEE802.16m etc.) proposes at present.Roughly comprise:

First, utilize existing singular value decomposition (SVD) technology to estimating that at receiving terminal the channel condition information obtaining decomposes, decomposition method is as shown in formula (1):

H＝U∑V ^H (1)

Wherein, U and V are orthogonal matrixes, and V is initial codebook set, ∑=diag (λ ₁, λ ₂λ _n) be the diagonal matrix being formed by singular value.

Then, the antenna configuration of supporting according to system and the number of code stream, quantize its call number of rear feedback to the V matrix in formula (1), and the selection that code book integrates, repeats no more as foundation here with the codebook set of existing document or standard.

Step 101: according to the channel performance information of estimating to obtain, determine initial codebook to revise in real time.

In this step, estimate that the channel performance obtaining is the statistic property of system to be estimated as upgrading granularity or cycle taking the time/frequency source block in concrete system, such as to Signal to Interference plus Noise Ratio (SINR, Signal-to-Interference-Pulse-Noise Ratio) statistic property estimate, to signal to noise ratio (SNR), the estimation of the statistic properties such as BER etc.Wherein, can adopt existing algorithm to the estimation of the statistic property such as SNR, BER, repeat no more here.The present invention also proposes a kind of statistic property to SINR and estimates.

The estimation of SINR is carried out for each code stream, at receiving terminal, its SINR is estimated, because the precoding vector of each code stream is all different, therefore, can adopt formula (2) to calculate the SINR of each code stream:

${\mathrm{SINR}}_i\left(n\right)=\frac{{\left|\right|H_i{V}_i\left(n\right)\left|\right|}^2}{{\mathrm{\&sigma;}}_i^2+\underset{k=1,k\&NotEqual;i}{\overset{K}{\mathrm{\&Sigma;}}}{\left|\right|H_i{V}_k\left|\right|}^2+{\left|\right|H_i{V}_i(:)\left|\right|}^2}---\left(2\right)$

Wherein, SINR _i(n) SINR of n code stream of i user of expression, H _iand σ _i ²be respectively i user's channel matrix and noise power, V _i(n) be the precoding codebook vector for n code stream of user i, V _i(:) is the precoding codebook vector of user i except n code stream.V _kfor the precoding codebook vector of interference user, the number that K is all interference users.

In the time that the performance statistics result being obtained by formula (2) estimation is undesirable, adopt the SINR of formula (2) to user i _i(n) estimation result, any two code streams (the SINR value as n and m) should meet the condition as shown in formula (3):

|SINR _i(n)-SINR _i(m)|≤ε (3)

Wherein, ε is a default smaller value, the energy difference between any two code streams that represent to allow.

In the time that the SINR value between any two code streams that estimation obtains does not meet formula (3), the present invention proposes to introduce code book correction matrix basic codebook matrix (being initial codebook set) is revised, and specific implementation can adopt following two kinds of methods.

Method one: adopt correction matrix to initial codebook set revise, modification method is as shown in formula (4):

$\tilde{V}={\mathrm{VP}}^k$ k＝0，1，2，…N _t-1 (4)

Wherein, V is original initial codebook set, and P is correction matrix, and k is modifying factor, N _tfor number of transmit antennas, for revised new codebook set.

The order of supposing channel matrix is n, and correction matrix P can be the square formation of one n × n, as shown in formula (5):

$P=\left[\begin{array}{ccccc}0& 0& \mathrm{\&Lambda;}& 0& 1\\ 1& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& 1& 0& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 1& 0\end{array}\right]---\left(5\right)$

When modifying factor k is initial, be 0, in the time of the discontented pedal system demand (not meeting formula (3)) of the SINR estimating for the first time, k increases progressively 1, otherwise k continues to remain 0, and the rest may be inferred.Can find out, whether now, only needing 1 bit of increase to feed back k value needs to increase progressively.Along with the judgement of each cycle adds up or keeps, make all code streams on all available antennas, to be rotated and to obtain best spatial sub-channel.The present invention is based on the method for designing of the codebook matrix correction of Signal to Interference plus Noise Ratio, in the situation that only increasing 1bit feedback overhead, ensured that code stream abundant search in spatial domain is applicable to the transmission subchannel of self.

Method two: the SINR on Dui Ge road carries out balance dwindles the difference of the SINR between each code stream as far as possible, and the present invention carries out balanced simple thinking to different code streams, has realized the balance of the error rate between code stream by simple method.Concrete grammar is as follows:

The situation that is 2 for order, modification method is as shown in formula (6):

$\tilde{V}=V\left[\begin{array}{cc}\sin \left(\mathrm{\&theta;}\right)& 0\\ 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right]---\left(6\right)$

In formula (6), θ is a selectable angle, and span is 0≤θ < pi/2, determine, and conditional number is larger according to the conditional number of the channel of H matrix, and value is more close to pi/2.V is original initial codebook set,

for revised new codebook set.

The situation that is greater than 2 for order, modification method is as shown in formula (7):

$\tilde{V}=V\left[\begin{array}{ccccc}\sin \left(\mathrm{\&theta;}\right)& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& \sin \left(\mathrm{\&theta;}\right)+\mathrm{\&Delta;}& \mathrm{\&Lambda;}& 0& 0\\ 0& 0& \mathrm{\&Lambda;}& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right]---\left(7\right)$

Wherein, increment factor Δ meets $0<\mathrm{\&Delta;}<(\frac{1}{\sin \left(\mathrm{\&theta;}\right)}-\sin \left(\mathrm{\&theta;}\right)),$ It is an adjustable factors.θ is a selectable angle, and span is 0≤θ < pi/2, determine, and conditional number is larger according to the conditional number of the channel of H matrix, and value is more close to pi/2.V is original initial codebook set,

for revised new codebook set.

From the present invention is based on the adaptive codebook generation method of code book correction, do not changing on the basis of existing code book basis matrix, according to the performance statistics characteristic of system, generate one group of new code book correction matrix, and consider original initial codebook set to revise in conjunction with feedback overhead, adjusted in time the precision of code book according to the real-time change of channel, in the situation that feedback overhead increase is very little, expand the capacity of code book, improve the precision of code book, thereby the in the situation that of Limited Feedback, more effectively improve the performance of precoding.

In addition, in the time adopting SINR to estimate that convergence rate estimation complexity slow or SINR is larger, can adopt prior art to estimate every data flow SNR.

If the estimation in this step is the estimation for SNR statistic property, so, after adopting prior art to estimate every data flow SNR, any two code streams (SNR as n and m) should meet formula (8):

|SNR _i(n)-SNR _i(m)|≤ε′ (8)

Wherein ε ' is default smaller value, the energy difference between any two code streams that represent to allow.In the time estimating that SNR value between any two code streams of obtaining does not meet formula (8), still can adopt the described method of said method one and method two to initial codebook set revise.

If the estimation in this step is the estimation for BER statistic property, so, can only obtain the situation of BER for single data stream or other, in the time adopting the system BER that goes out of existing method statistic higher than a certain threshold value setting in advance, adopt following methods to initial codebook set revise, as formula (9):

$\tilde{V}={\mathrm{\&Phi;}}^{k}V$ k＝0，1，2，…N _t-1 (9)

Wherein, V is original initial codebook set, and Φ is correction matrix, and k is modifying factor, N _tfor number of transmit antennas.

for revised new codebook set, correction matrix Φ is as shown in formula (10) formula:

$\mathrm{\&Phi;}=\left[\begin{array}{cccc}{e}^{\mathrm{j\&pi;}\frac{1}{N_t^2}}& 0& \mathrm{\&Lambda;}& 0\\ 0& e^{\mathrm{j\&pi;}\frac{2}{N_t^2}}& \mathrm{\&Lambda;}& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& 0\\ 0& 0& \mathrm{\&Lambda;}& e^{\mathrm{j\&pi;}\frac{n}{N_t^2}}\end{array}\right]---\left(10\right)$

Wherein, the order that n is channel, N _tfor number of transmit antennas.

For the inventive method, a kind of code book generating apparatus is also provided, comprise initial codebook set generation unit,, estimation unit, comparing unit and amending unit, wherein,

Initial codebook set generation unit, for obtaining initial codebook set feedback, and exports to amending unit by the initial codebook set obtaining.

Estimation unit, for estimating channel performance information, and exports to comparing unit.

Comparing unit, for according to the channel performance information from estimation unit, in the time that the difference between any two code streams that estimation obtains does not meet default non-correction conditions, the correction result notice sending to amending unit is for showing to need correction; In the time that the difference between any two code streams that estimation obtains meets default non-correction conditions, the correction result notice sending to amending unit is for showing not need correction.

Amending unit, for receiving the correction result notice from comparing unit, when correction result notice is when showing to revise, to revising from the initial codebook set of initial codebook set generation unit, and exports revised codebook set; When correction result notice is when showing not need to revise, will export as codebook set from the initial codebook set of initial codebook set generation unit.

The above, be only preferred embodiment of the present invention, is not intended to limit protection scope of the present invention, all any amendments of doing within the spirit and principles in the present invention, is equal to and replaces and improvement etc., within all should being included in protection scope of the present invention.

Claims (8)

1. a method for generating codebooks, is characterized in that, comprising:

Obtain initial codebook set feedback;

According to the channel performance information of estimating to obtain, determine initial codebook to revise in real time;

Wherein, described channel performance is Signal to Interference plus Noise Ratio SINR; Described estimation channel performance information is:

For each code stream, its SINR is for below shown in formula:

${\mathrm{SINR}}_i\left(n\right)=\frac{{\left|\right|H_i{V}_i\left(n\right)\left|\right|}^2}{{\mathrm{\&sigma;}}_i^2+\underset{k=1,k\&NotEqual;i}{\overset{K}{\mathrm{\&Sigma;}}}{\left|\right|H_i{V}_k\left|\right|}^2+{\left|\right|H_i{V}_i(:)\left|\right|}^2};$

Wherein, SINR _i(n) SINR of n code stream of i user of expression, H _iwith be respectively i user's channel matrix and noise power, V _i(n) be the precoding codebook vector for n code stream of user i, V _i(:) is the precoding codebook vector of user i except n code stream, V _kfor the precoding codebook vector of interference user, the number that K is all interference users.

2. method for generating codebooks according to claim 1, is characterized in that, described decision is modified in real time to initial codebook:

Described SINR value between any two code streams does not meet | SINR _i(n)-SINR _i(m) | when≤ε, wherein, described m is m code stream, and ε is a default smaller value, and the energy difference between any two code streams that represent to allow, adopts correction matrix P, according to the following formula described initial codebook set is revised:

$\tilde{V}={\mathrm{VP}}^k,k=\mathrm{0,1,2},...N_t-1,$

Wherein, V is original initial codebook set, and P is correction matrix, and k is modifying factor, N _tfor number of transmit antennas, for revised new codebook set;

Correction matrix $P=\left[\begin{array}{ccccc}0& 0& \mathrm{\&Lambda;}& 0& 1\\ 1& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& 1& 0& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 1& 0\end{array}\right],$ The order of wherein supposing channel matrix is n.

3. method for generating codebooks according to claim 1, is characterized in that, described decision is modified in real time to initial codebook:

Described SINR value between any two code streams does not meet | SINR _i(n)-SINR _i(m) | when≤ε, wherein, described m is m code stream, and ε is a default smaller value, the energy difference between any two code streams that represent to allow, the SINR on Dui Ge road carries out balance.

4. method for generating codebooks according to claim 3, is characterized in that, the SINR on described Dui Ge road carries out balance and comprises:

The situation that is 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{cc}\sin \left(\mathrm{\&theta;}\right)& 0\\ 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right];$

The situation that is greater than 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{ccccc}\sin \left(\mathrm{\&theta;}\right)& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& \sin \left(\mathrm{\&theta;}\right)\mathrm{\&Delta;}& \mathrm{\&Lambda;}& 0& 0\\ 0& 0& \mathrm{\&Lambda;}& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right],$

Wherein, increment factor △ meets

be an adjustable factors, θ is a selectable angle, and span is 0< θ < pi/2;

V is original initial codebook set,

for revised new codebook set.

5. method for generating codebooks according to claim 1, is characterized in that, described channel performance is signal to noise ratio snr, and described decision is modified in real time to initial codebook:

Described SNR value between any two code streams does not meet | SNR _i(n)-SNR _i(m) | when≤ε ', wherein, described m is m code stream, and ε ' is a default smaller value, and the energy difference between any two code streams that represent to allow, adopts correction matrix P, according to the following formula described initial codebook set is revised:

$\tilde{V}={\mathrm{VP}}^k,k=\mathrm{0,1,2},...N_t-1,$

Wherein, V is original initial codebook set, and P is correction matrix, and k is modifying factor, N _tfor number of transmit antennas, for revised new codebook set;

Correction matrix $P=\left[\begin{array}{ccccc}0& 0& \mathrm{\&Lambda;}& 0& 1\\ 1& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& 1& 0& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 1& 0\end{array}\right],$ The order of wherein supposing channel matrix is n.

6. method for generating codebooks according to claim 5, is characterized in that, described decision is modified in real time to initial codebook:

Described SNR value between any two code streams does not meet | SNR _i(n)-SNR _i(m) | when≤ε ', the SNR on Dui Ge road carries out balance; Wherein, ε ' is a default smaller value, the energy difference between any two code streams that represent to allow;

The SNR on described Dui Ge road carries out balance and comprises:

The situation that is 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{cc}\sin \left(\mathrm{\&theta;}\right)& 0\\ 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right];$

The situation that is greater than 2 for order, modification method is shown below:

$\tilde{V}=V\left[\begin{array}{ccccc}\sin \left(\mathrm{\&theta;}\right)& 0& \mathrm{\&Lambda;}& 0& 0\\ 0& \sin \left(\mathrm{\&theta;}\right)\mathrm{\&Delta;}& \mathrm{\&Lambda;}& 0& 0\\ 0& 0& \mathrm{\&Lambda;}& 0& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}\\ 0& 0& \mathrm{\&Lambda;}& 0& \frac{1}{\sin \left(\mathrm{\&theta;}\right)}\end{array}\right],$

Wherein, increment factor △ meets be an adjustable factors, θ is a selectable angle, and span is 0< θ < pi/2;

V is original initial codebook set,

for revised new codebook set.

7. method for generating codebooks according to claim 1, is characterized in that, described channel performance is error rate BER;

Described decision is modified in real time to initial codebook:

When the described system BER estimating higher than set in advance threshold value time, adopt following formula described initial codebook set is revised: $\tilde{V}={\mathrm{\&Phi;}}^{k}V,k=\mathrm{0,1,2},...N_t-1;$

Wherein, V is original initial codebook set, and Φ is correction matrix, and k is modifying factor, N _tfor number of transmit antennas,

for revised new codebook set, correction matrix Φ is shown below:

$\mathrm{\&Phi;}=\left[\begin{array}{cccc}{e}^{\mathrm{j\&pi;}\frac{1}{{N_t}^2}}& 0& \mathrm{\&Lambda;}& 0\\ 0& e^{\mathrm{j\&pi;}\frac{2}{{N_t}^2}}& \mathrm{\&Lambda;}& 0\\ \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& \mathrm{\&Lambda;}& 0\\ 0& 0& \mathrm{\&Lambda;}& e^{\mathrm{j\&pi;}\frac{n}{{N_t}^2}}\end{array}\right];$

Wherein, the order that n is channel, N _tfor number of transmit antennas.

8. a code book generating apparatus, is characterized in that, comprises initial codebook set generation unit, estimation unit, comparing unit and amending unit, wherein,

Initial codebook set generation unit, for obtaining initial codebook set feedback, and exports to amending unit by the initial codebook set obtaining;

Estimation unit, for estimating channel performance information, and exports to comparing unit; Wherein, described channel performance is Signal to Interference plus Noise Ratio SINR; Described estimation channel performance information is:

For each code stream, its SINR is for below shown in formula:

${\mathrm{SINR}}_i\left(n\right)=\frac{{\left|\right|H_i{V}_i\left(n\right)\left|\right|}^2}{{\mathrm{\&sigma;}}_i^2+\underset{k=1,k\&NotEqual;i}{\overset{K}{\mathrm{\&Sigma;}}}{\left|\right|H_i{V}_k\left|\right|}^2+{\left|\right|H_i{V}_i(:)\left|\right|}^2};$

Wherein, SINR _i(n) SINR of n code stream of i user of expression, H _iwith

be respectively i user's channel matrix and noise power, V _i(n) be the precoding codebook vector for n code stream of user i, V _i(:) is the precoding codebook vector of user i except n code stream, V _kfor the precoding codebook vector of interference user, the number that K is all interference users;

Comparing unit, for according to the channel performance information from estimation unit, in the time that the difference between any two code streams that estimation obtains does not meet default non-correction conditions, the correction result notice sending to amending unit is for showing to need correction; In the time that the difference between any two code streams that estimation obtains meets default non-correction conditions, the correction result notice sending to amending unit is for showing not need correction;

Amending unit, for receiving the correction result notice from comparing unit, when correction result notice is when showing to revise, to revising from the initial codebook set of initial codebook set generation unit, and exports revised codebook set; When correction result notice is when showing not need to revise, will export as codebook set from the initial codebook set of initial codebook set generation unit.

Images