CN103684700A - 3D (three-dimensional) MU-MIMO (multiple user-multiple input multiple output) precoding method based on orthogonal joint codebook set - Google Patents

Abstract

The invention discloses a 3D (three-dimensional) MU-MIMO (multiple user-multiple input multiple output) precoding method based on an orthogonal joint codebook set, and relates to the technical field of mobile communication. The method comprises the following steps of enabling a base station terminal to adopt a uniform surface antenna matrix, and simultaneously store the orthogonal joint codebook set together with a user terminal; according to a 3D MIMO channel matrix, enabling the user terminal to select the optimum precoding vector in the horizontal dimension and the vertical dimension; feeding the corresponding number of the precoding vector back to the base station terminal, enabling the base station terminal to form a 3D precoding matrix, and sending signals to a plurality of users to carry out the 3D precoding processing. The method has the advantages that the orthogonal joint codebook set covers larger space, and the orthogonality is formed between the codebooks, so the matching of the 3D MIMO channel is more accurate and comprehensive, and the interference of MU channel sharing is effectively inhibited; meanwhile, by selecting the optimum precoding matrix, under the condition of not increasing the feedback amount, the channel information in the horizontal dimension and the vertical dimension are comprehensively utilized, and the integral property of the system is improved.

Description

A kind of 3D MU-MIMO method for precoding based on quadrature associating code book collection

Technical field

The present invention relates to communication technical field, mainly for LTE-Advanced system 3D MU-MIMO, propose a kind of method for precoding.

Background technology

As LTE-Advanced and LTE air interface core technology-MIMO technology, can be in the situation that not increasing bandwidth and transmitting power, improve power system capacity, mainly that binding time territory and spatial domain are processed received signal effectively by being distributed in many antennas in space, utilize multipath transmisstion characteristic and the random fading characteristic of channel simultaneously, effectively improved transmission quality and promoted transmission rate.

Yet in actual working environment, the embodiment of MIMO technical advantage greatly depends on mimo channel model characteristics.Therefore, set up and effectively can reflect that the channel model of true MIMO operational environment just seems very important.Existing mimo channel model is based upon in two dimension (2Dimension, 2D) plane mostly.If merely from Electromagnetic Wave Propagation mechanism, this channel model imperfection, because this model is only considered the electromagnetic horizontal transmission in space, so 2D MIMO does not make full use of space resources.For this reason, academia comprises that industrial quarters carried out the mimo channel modeling work based on 3D in a large number.3D mimo channel model requires to consider the electromagnetic horizontal and vertical propagation in space simultaneously, requires to consider azimuth and the angle of pitch simultaneously, and therefore, channel model is more accurate.Consider the advantage of 3D MIMO technology, industrial quarters comprises that the companies such as high pass, A Lang, DOCOMO, Nuo Xi have set up advanced air interface (the Advanced Radio Interface Technologies for4G Systems towards 4G specially, ARTIST4G) working group, this working group points out: the 3D MIMO technology based on active array will improve the key technology of Cell Edge User throughput and data transmission rate as future mobile communication system.Meanwhile, 3GPP will have started the technical specification standard operation of LTE-Advanced R12 since the end of the year in 2012, the formal core using 3D MIMO technology as LTE-Advanced R12 technical specification.Taking this as an opportunity, attracted domestic and international a large amount of scholar, engineer to join in 3D MIMO key technology and standardized research work.

In LTE-Advanced system, the precoding technique based on 3D channel comprises choosing of code book design and code book etc.Common precoding codebook comprises DFT(discrete Fourier transform) code book, random code book, Jim Glassman code book etc.Selection for optimum precoding vectors is the performance index criterion based on different.Such as, minimum singular value criterion, heap(ed) capacity criterion etc.In pre-coding system based on code book, first the code book having designed is stored in to transmitting terminal and receiving terminal, receiving terminal receives signal, by channel estimating, obtain current C SI, then utilize CSI, in stored local code book, based on precoding Criterion of Selecting, select optimum precoding vectors, and the corresponding call number of optimum precoding vectors is fed back to base station end, and base station end recovers optimum precoding vectors by this call number, and generating pre-coding matrix carries out precoding processing to input data.

In 3D MIMO multi-user system, along with increasing of number of users, user distribution is zones of different in community, comprises center of housing estate and cell edge.Traditional precoding technique has only utilized the channel information of horizontal dimension, and cannot to user, distinguish in vertical dimension, and this has also just caused serious inter-user interference.For example, the 3D precoding based on DFT code book, horizontal dimensions all adopts identical DFT code book with the code book of vertical dimensions, and the optimum precoding vectors of horizontal dimension and the optimum precoding vectors of vertical dimension are by choosing in this DFT code book.In 3D pre-coding scheme, horizontal dimension all adopts identical DFT code book with vertical dimension base, there is no Consideration of Three-dimensional spatial character completely, thereby causes the inhibition of CCI limited in one's ability.Therefore,, in order to promote capacity and the user performance of center of housing estate and Cell Edge User simultaneously, demand combined level dimension and vertical dimension channel information, to realize 3D MIMO precoding.

Summary of the invention

The present invention proposes a kind of 3D MU-MIMO method for precoding based on quadrature associating code book collection, the construction method that comprises concrete quadrature associating code book collection.Base station end adopts uniform surface aerial array, stores this quadrature simultaneously combine code book collection with user side; User side carries out respectively choosing of optimum precoding vectors in horizontal dimensions and vertical dimensions, then the call number of the optimum precoding vectors of correspondence is fed back to base station end; Base station end forms optimum pre-coding matrix according to each user's optimum precoding vectors, thereby multi-user's transmitted signal is carried out to 3D precoding processing.

The technical scheme that the present invention solves the problems of the technologies described above is: design a kind of 3D MU-MIMO method for precoding based on quadrature associating code book collection, comprise step: base station end adopts surface antenna array structure, each root antenna port receives the signal message in two dimensional surface and processes signal message in three dimensions, structure quadrature associating code book collection is also stored in base station end and user side simultaneously, user side channel estimates to obtain 3D channel matrix, according to 3D channel matrix, in horizontal dimensions, from quadrature, combine code book with vertical dimensions respectively and concentrate the optimum precoding vectors of selection, and the sequence number in corresponding code book of the optimum precoding vectors in two dimensions is sent it back to base station end as feedback call number, base station end recovers optimum precoding vectors according to the call number receiving, generating pre-coding matrix carries out precoding processing to input data.

When according to DFT code book:

basis of formation code book F _basetime: according to formula: $T_i=\left(\begin{array}{cccc}1e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}1\frac{i}{N}}& e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}2\frac{i}{N}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}(N_t-1)\frac{i}{N}}\end{array}\right);i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,L;$ Determine basic code book F _base={ T ₁, T ₂..., T _lin element T _i.The structure of quadrature associating code book collection comprises: according to code book, to the level of coverage of channel space, code book complexity and the pre-coding matrix selected by the code book inhibition ability to CCI, choose basic code book F _base; By basic code book F _basecarry out orthogonal transform, obtain quadrature code book F' mutually orthogonal between each code word _base; Extract basic code book F _base={ T ₁, T ₂..., T _lin any one code word T _l(l=1,2 ... L), according to formula

find L corresponding with it vector by L vector, form quadrature sub-clustering code book

obtain thus quadrature sub-clustering code book collection Γ '={ F _{base_1}, F _{base_2}..., F _{base_L}, by this quadrature sub-clustering code book collection Γ ' and basic code book F _baseaccording to formula: Γ={ { F _base, F _{base_1}, { F _base, F _{base_2}..., { F _base, F _{base_L}combination, final quadrature associating code book collection Γ obtained.

Base station end generating pre-coding matrix is specially, and base station end recovers the optimum precoding vectors of all user's horizontal dimension and the optimum precoding vectors of vertical dimension according to the call number receiving, and generates 3D pre-coding matrix according to point multiplication operation.User side, based on maximizing minimum singular value criterion, according to 3D channel matrix, carries out respectively vertical dimensions and horizontal dimensions feedback quantity and calculates.Client feeds back call number comprises the optimum precoding vectors call number of horizontal dimension and the optimum precoding vectors call number of vertical dimension, and wherein, the optimum precoding vectors that the optimum precoding vectors call number of horizontal dimension correspondence is mated present level Vc SI is most at basic code book F _basesequence number, the optimum precoding vectors call number of vertical dimension is the corresponding quadrature sub-clustering of the optimum precoding vectors call number of horizontal dimension code book F _{base_l}.

Client feeds back is to the optimum precoding vectors call number of base station end, not only comprise the optimum precoding vectors call number of horizontal dimensions, comprised the optimum precoding vectors call number of vertical dimensions simultaneously, wherein, certain precoding vectors that the optimum precoding vectors call number of horizontal dimensions is optimum Match present level Vc SI is at basic code book F _basein sequence number; By the optimum precoding vectors call number of this horizontal dimension, determine corresponding with it quadrature sub-clustering code book F again _{base_l}as vertical dimension code book, thereby certain precoding vectors that the optimum precoding vectors call number of vertical dimension is the current vertical dimensions CSI of optimum Match is at this quadrature sub-clustering code book F _{base_l}in sequence number.

In the present invention, the design of quadrature associating code book collection not only combines horizontal dimensions with vertical dimensions CSI carries out code book design, but also the level of coverage of consideration code book to 3D channel, and 3D pre-coding matrix is for the inhibition ability of CCI.To sum up, the 3D MU-MIMO method for precoding based on quadrature associating code book collection that the present invention proposes can obtain stronger precoding gain, improves the overall performance of system.

Quadrature associating code book collection proposed by the invention has not only covered larger space, and between code book, has orthogonality, therefore, not only more accurate comprehensively for the coupling of 3D mimo channel, can also effectively suppress multi-user simultaneously and disturb in road altogether; In addition, the present invention can not increase under the condition of feedback quantity, and the channel information in level of comprehensive utilization and vertical dimensions has improved the overall performance of system.

Accompanying drawing explanation

Fig. 1 is the system block diagram of the 3D MU-MIMO method for precoding that proposes of the present invention;

Fig. 2 is the pre-encode operation flow chart based on code book;

Fig. 3 is quadrature sub-clustering code book organigram in the present invention;

Fig. 4 is quadrature associating code book collection organigram in the present invention;

Fig. 5 is 3D precoding concrete operations flow chart of the present invention;

Fig. 6 be under WINNER channel " A1 " scene method for precoding of the present invention with DFT codebook precoding and fixed codebook precoding bit error rate performance comparison diagram under identical feedback quantity;

Fig. 7 be method for precoding of the present invention with DFT codebook precoding and fixed codebook precoding the CDF curve comparison diagram under identical signal to noise ratio;

Fig. 8 is the performance comparison diagram of " A1 " scene method for precoding under different codebook size under WINNER channel;

Embodiment

Below in conjunction with instantiation, present invention is described.

Fig. 1 is the 3D MU-MIMO method for precoding system block diagram based on quadrature associating code book collection in the present invention.Base station end adopts n * m dimension surface antenna array structure, the line number of n presentation surface aerial array wherein, the columns of m presentation surface aerial array, n=m.Require the simultaneously information in processing horizontal dimension and vertical dimensions of each transmit antennas.User side reception antenna number represents with r.The signal that j user receives can be expressed as:

$y_i=H_j{W}_j{X}_j+H_j\underset{i=1i\&NotEqual;j}{\overset{J}{\mathrm{\&Sigma;}}}{W}_i{X}_i+n_j,j=\mathrm{1,2},...,J$

Receive signal and formed by three parts, wherein, H _jw _jx _jrepresent the useful signal part that receiving terminal receives,

represent j the interference from other subscriber signals that user receives, nj represents noise signal, for obeying CN (0, N ₀) probability distribution white Gauss noise.User side obtains 3D channel matrix H by channel estimating _{n * m * r}according to precoding Criterion of Selecting, respectively in horizontal dimensions and vertical dimensions, from quadrature associating code book, concentrate and select optimum precoding vectors, and the sequence number in corresponding code book of the optimum precoding vectors in two dimensions is fed back to base station end as call number, base station end recovers optimum precoding vectors according to the call number receiving, and then generating pre-coding matrix carries out precoding processing to input data.

Fig. 2 is the precoding basic operation flow chart based on code book, in LTE-Advanced system, base station end and user side store code book simultaneously, user side is according to CSI, in code book, travel through all code words, select the code word of mating with current C SI as optimum precoding vectors, and its institute's manipulative indexing number in code book is fed back to base station end, base station end recovers optimum precoding vectors according to the call number receiving, and then formation pre-coding matrix carries out precoding processing to input data.According to Fig. 5, and with reference to figure 3 and Fig. 4, in the present invention, the realization of 3D MU-MIMO method for precoding specifically comprises:

Transmitting terminal adopts surface antenna array structure, guarantees each transmit antennas information in processing horizontal dimension and vertical dimensions simultaneously, user side employing r root reception antenna.The structure of quadrature associating code book collection, shown in Fig. 3 and Fig. 4, is specifically expressed as follows:

A, basic code book F _basestructure

Consider discrete Fourier transform (Discrete Fourier Transformation, DFT) code book has the advantages that complexity is low, and the precoding codebook of stipulating in 3GPP standard is DFT code book, therefore selecting DFT code book basis of formation code book, this example to select following DFT matrix is that example illustrates enforcement of the present invention:

according to above formula, basic code book F _basebe:

Wherein $T_i=\left(\begin{array}{cccc}1e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}1\frac{i}{N}}& e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}2\frac{i}{N}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;}(N_t-1)\frac{i}{N}}\end{array}\right);i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,L;,$ Represent a certain row in DFT matrix, N is DFT code book number, and L is basic code book number, N=L.N _tfor antenna number, N _t=n.For other conventional code books, also can be according to the method described above in conjunction with antenna number basis of formation code book.

B, quadrature code book F' _basestructure

According to basic code book F _base builds quadrature code book F' _base.To basic code book F _basemiddle column vector { T ₁, T ₂..., T _ldo orthogonal transform and obtain quadrature code book F' _base:

$F_{\mathrm{base}}^{\&prime;}=\{{\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>}}_1^{\&prime;},{\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>}}_2^{\&prime;},...,T_L^{\&prime;}\}=\left\{\begin{array}{ccccc}{\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">\; <figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{1,1}}^{\&prime;}& {\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">\; <figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{1,2}}^{\&prime;}& {\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>}}_{1,i}^{\&prime;}& ...& {\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>}}_{1,L}^{\&prime;}\\ {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">\; <figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{2,1}^{\&prime;}& {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">\; <figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{2,2}}^{\&prime;}& {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>}}_{2,i}^{\&prime;}& ...& {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>}}_{2,L}^{\&prime;}\\ T_{\mathrm{3,1}}^{\&prime;}& {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">\; <figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{3,2}}^{\&prime;}& {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">T</figure-callout>}}_{3,i}^{\&prime;}& ...& {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">T</figure-callout>}}_{3,L}^{\&prime;}\\ .& .& .& & .\\ .& .& .& & .\\ .& .& .& & .\\ T_{N_t,1}^{\&prime;}& T_{N_t,2}^{\&prime;}& T_{N_t,i}^{\&prime;}& ...& T_{N_t,L}^{\&prime;}\end{array}\right\}$ (as Fig. 3); Due to quadrature code book F' _baseeach element

T' ₁, T' ₂..., T' _lbetween there is orthogonality, therefore by F' _basethe pre-coding matrix forming will have orthogonality equally, thereby effectively guarantee the inhibition to CCI.

The formation of C, quadrature associating code book collection

In order better to utilize quadrature code book in 3d space, itself and basic code book can be done to transform expansion in the lump, concrete operations are as shown in Figure 3.In this invention, basis code book is corresponding with CSI overall space, by the dot product to any code word is done appointment in basic code book, convert, thereby obtain the sub-clustering of the quadrature associating code book corresponding with each basic code word, the corresponding CSI subspace of basic code word is corresponding therewith in this sub-clustering, the present invention carries out quadrature joint code the design, and then forms quadrature associating code book collection by all sub-clustering collection.

Fig. 4 is that a kind of detailed implementation procedure that forms quadrature sub-clustering code book illustrates, and gets basic code book F _base={ T ₁, T ₂..., T _lin any one code word T _l(l=1,2 ... L), utilize relational expression

find L corresponding with it vector

by this L vector, form quadrature sub-clustering code book F _{base_l}=[T ₁ ^l, T ₂ ^l..., T _l ^l], for example, in Fig. 3 $F_{\mathrm{base}\_3}=\left\{\begin{array}{ccccc}{\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">\; <figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{1,1}}^3& {\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">\; <figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{1,2}}^3& {\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>}}_{1,\mathrm{<figure-callout\; id="3"\; label="i"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">i</figure-callout>}}^3& ...& {\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>}}_{1,L}^3\\ {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">\; <figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{2,1}}^2& {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">\; <figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{2,2}}^3& {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>}}_{2,\mathrm{<figure-callout\; id="3"\; label="i"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">i</figure-callout>}}^3& ...& {\mathrm{<figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>}}_{2,L}^3\\ {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">\; <figure-callout\; id="1"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000042.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{3,1}}^3& {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">\; <figure-callout\; id="2"\; label="T"\; filenames="HDA0000451148750000031.png,HDA0000451148750000041.png"\; state="\{\{state\}\}">T</figure-callout>\; </figure-callout>}}_{\mathrm{3,2}}^3& {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">T</figure-callout>}}_{3,\mathrm{<figure-callout\; id="3"\; label="i"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">i</figure-callout>}}^3& ...& {\mathrm{<figure-callout\; id="3"\; label="T"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">T</figure-callout>}}_{3,\mathrm{<figure-callout\; id="3"\; label="L"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">L</figure-callout>}}^3\\ .& .& .& & .\\ .& .& .& & .\\ .& .& .& & .\\ T_{N_t,1}^3& T_{N_t,2}^3& T_{N_t,\mathrm{<figure-callout\; id="3"\; label="i"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">i</figure-callout>}}^3& ...& T_{N_t,\mathrm{<figure-callout\; id="3"\; label="L"\; filenames="HDA0000451148750000041.png,HDA0000451148750000051.png"\; state="\{\{state\}\}">L</figure-callout>}}^3\end{array}\right\}.$ All code words in basic code book Fbase are all done to above-mentioned conversion, obtain associated quadrature sub-clustering code book collection Γ a '={ F _{base_1}, F _{base_2}..., F _{base_L}, then by this quadrature sub-clustering code book collection Γ ' and basic code book F _baseaccording to formula Γ={ { F _base, F _{base_1}, { F _base, F _{base_2}..., { F _base, F _{base_L}carry out combination, thus final quadrature associating code book collection Γ, wherein Γ obtained _i={ F _base, F _{base_i}represent the quadrature associating code book that this code book is concentrated.

The present invention is based on the 3D pre-coding scheme of quadrature associating code book collection, the base station quadrature that DFT matrix consist of same with the common storage of user side combined code book collection, client feeds back call number comprises the optimum precoding vectors call number of horizontal dimension and the optimum precoding vectors call number of vertical dimension, wherein, the corresponding a certain optimum precoding vectors that mates present level Vc SI most of the optimum precoding vectors call number of horizontal dimension is at basic code book F _basesequence number, the optimum precoding vectors call number of vertical dimension comes from the corresponding quadrature sub-clustering of the optimum precoding vectors call number of horizontal dimension code book F _{base_l}, therefore, there is same feedback quantity with the 3D precoding based on DFT code book.Base station end recovers the optimum precoding vectors of all user's horizontal dimension and the optimum precoding vectors of vertical dimension according to the call number receiving, and generate 3D pre-coding matrix according to point multiplication operation, guaranteed that in 3D pre-coding matrix, each element has orthogonality, so just can suppress well CCI.In sum, the 3D pre-coding matrix in the present invention not only more mates current channel information space, can effectively suppress CCI simultaneously.

Quadrature is combined to code book collection and be stored in base station end and user side simultaneously, user side is based on maximizing minimum singular value criterion, from basic code book F _basein determine the optimum precoding vectors of coupling present level dimension channel information, from quadrature associating code book, concentrate and select the quadrature sub-clustering code book F corresponding with this horizontal dimension optimum precoding vectors _{base_l}.As vertical dimensions code book, further according to maximizing minimum singular value criterion from this quadrature sub-clustering code book F _{base_l}the optimum precoding vectors of current vertical dimension channel information is mated in middle selection the most.

The optimum precoding vectors selection criterion of horizontal dimensions and vertical dimensions is and maximizes minimum singular value criterion:

$T_{\mathrm{opt}}=\arg \underset{1\&le;l\&le;L}{\max }{\mathrm{\&lambda;}}_{\min }\left\{H{W}_i\right\}$

W wherein _irepresent code word in current code book, HW _irepresent equivalent channel, λ _min{ H _xw _ybe HW _iminimum non-zero singular value, T _optrepresent the optimum precoding vectors of therefrom selecting after this code book of traversal.

Through aforesaid operations, by the optimum precoding vectors of the horizontal dimensions obtaining at basic code book F _basemiddle corresponding sequence number is as the optimum precoding vectors call number of horizontal dimension, meanwhile, by the optimum precoding vectors of the vertical dimension obtaining at quadrature sub-clustering code book F _{base_l}the sequence number of middle correspondence is as the optimum precoding vectors call number of vertical dimension, last, and user side feeds back to base station end by the optimum precoding vectors call number of this horizontal dimension together with the optimum precoding vectors call number of vertical dimension.

At base station end, according to the feedback call number of receiving, recover corresponding horizontal dimension and the optimum precoding vectors of vertical dimension, then form horizontal dimension pre-coding matrix W according to the optimum precoding vectors of horizontal dimension and the optimum precoding vectors of vertical dimension _hwith vertical dimension pre-coding matrix and W _v, by W _hwith W _vbe extended to respectively n and m, then do dot product extended arithmetic and generate the pre-coding matrix corresponding with number of antennas _w, W=[W _h; W _h; ... W _h]. * [W _v; W _v; ... W _v], like this, the W that user utilizes W that point multiplication operation obtains and other user to utilize same procedure to obtain _ibetween there is very strong orthogonality, thereby can suppress well CCI.

Sending and receiving end stores a same code book consisting of DFT matrix.

Emulation of the present invention adopts " A1 " (Indoor office/residential) scene under WINNER, and Fig. 6 has provided 3D method for precoding of the present invention and contrasted with 3D precoding and the bit error rate performance of the 3D method for precoding based on fixed codebook (fixed codebook only adopts two column vectors as code book) under identical feedback quantity based on DFT.It is 3(dB that Fig. 7 provides signal to noise ratio) time, method for precoding of the present invention and the 3D precoding based on DFT and the CDF curve comparison of the 3D method for precoding based on fixed codebook.As we know from the figure, the 3D pre-coding scheme based on quadrature associating code book collection proposed by the invention has obvious advantage with respect to aforementioned two kinds of pre-coding schemes on BER performance and CDF curve.This is because the 3D method for precoding in the present invention has been realized coupling more accurately to current channel information, has again stronger CCI simultaneously and suppresses ability.Figure 8 shows that the Performance Ratio of 3D method for precoding of the present invention under different codebook size, code book is larger, the channel information space covering is just more accurate, but code book is excessive, the feedback quantity of system will increase, consider above-mentioned two aspects, select balanced codebook size, the present invention is 16 according to factor, codebook size optimal selections such as channel space, performance, feedbacks.

The constructed quadrature associating code book collection of 3D MU-MIMO method for precoding that the present invention proposes has not only been realized and comprehensively having been covered 3D mimo channel is more accurate, but also can effectively suppress multi-user, disturbs in road altogether; , in the present invention, for the choosing of optimum pre-coding matrix, can not increase under the condition of feedback quantity meanwhile, take full advantage of level and vertical dimensions channel information, therefore, the overall performance of system has obtained obvious lifting.

Claims (6)

1. the 3D MU-MIMO method for precoding based on quadrature associating code book collection, it is characterized in that, comprise step: base station end adopts surface antenna array structure, each root antenna port receives the signal message in two dimensional surface and processes signal message in three dimensions, structure quadrature associating code book collection is also stored in base station end and user side simultaneously, user side channel estimates to obtain 3D channel matrix, according to 3D channel matrix, in horizontal dimensions, from quadrature, combine code book with vertical dimensions respectively and concentrate the optimum precoding vectors of selection, and the sequence number in corresponding code book feeds back to base station end as call number using the optimum precoding vectors in two dimensions, base station end recovers optimum precoding vectors according to the call number of client feeds back, generating pre-coding matrix carries out precoding processing to input data.

2. method according to claim 1, it is characterized in that, the structure of quadrature associating code book collection comprises: according to code book, the level of coverage of channel space, code book complexity and the pre-coding matrix Dui Gong road selected by code book are disturbed to (Co-Channel Interference, CCI) inhibition ability, chooses basic code book F _base; By basic code book F _basecarry out orthogonal transform, obtain quadrature code book F' mutually orthogonal between each code word _base; Extract base code book F _base={ T ₁, T ₂..., T _lin any one code word T _l(l=1,2 ... L), according to formula

find L corresponding with it vector

by L vector, form quadrature sub-clustering code book

obtain thus quadrature sub-clustering code book collection Γ '={ F _{base_1}, F _{base_2}..., F _{base_L}, by this quadrature sub-clustering code book collection Γ ' and basic code book F _baseaccording to formula: Γ={ { F _base, F _{base_1}, { F _base, F _{base_2}..., { F _base, F _{base_L}combination, final quadrature associating code book collection Γ obtained.

3. method according to claim 1, it is characterized in that, base station end generating pre-coding matrix is specially, and base station end recovers the optimum precoding vectors of all user's horizontal dimension and the optimum precoding vectors of vertical dimension according to the call number receiving, and generates 3D pre-coding matrix according to point multiplication operation.

4. the method described in one of them according to claim 1-3, is characterized in that, user side, based on maximizing minimum singular value criterion, according to 3D channel matrix, carries out respectively vertical dimensions and the calculating of horizontal dimensions feedback quantity.

5. the method described in one of them according to claim 1-3, it is characterized in that, the call number of client feeds back comprises the optimum precoding vectors call number of horizontal dimension and the optimum precoding vectors call number of vertical dimension, wherein, the optimum precoding vectors that the optimum precoding vectors call number of horizontal dimension correspondence is mated present level dimension channel condition information CSI is most at basic code book F _basesequence number, the optimum precoding vectors call number of vertical dimension is the corresponding quadrature sub-clustering of the optimum precoding vectors call number of horizontal dimension code book F _{base_l}.

6. method according to claim 2, is characterized in that, when according to discrete Fourier transform DFT code book:

basis of formation code book F _basetime: according to formula: $T_i=\left(\begin{array}{cccc}1e^{-j2\mathrm{\&pi;}1\frac{i}{N}}& e^{-j2\mathrm{\&pi;}2\frac{i}{N}}& \&CenterDot;\&CenterDot;\&CenterDot;& e^{-j2\mathrm{\&pi;}(N_t-1)\frac{i}{N}}\end{array}\right);i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,L;$ Determine basic code book F _base={ T ₁, T ₂..., T _lin element T _i.

Images