CN101369871A - Multi-user MIMO system precoding method and apparatus based on limited feedback - Google Patents

Abstract

The invention discloses a channel quantization method of a multiuser MIMO system based on a limited feedback and a device thereof belonging to the wireless transmission technology field. The method comprises the following steps: firstly choosing a code word having a minimal included angle with a codebook from the codebook according to the channel state information, and calculating the quantized error; secondly choosing a code word vector having a minimal included angle with the quantized error from the codebook; projecting the channel vector to the space formed by the chosen two code word vectors and calculating to obtain a compound proportion factor; quantizing the compound proportion factor using a pre-generated quantization step set. A receiving terminal device comprises a channel condition information acquisition unit, a channel vector quantization unit and a quantization information feedback unit. The transmission terminal device comprises a feedback information reconstruction unit and a precoding unit. The channel quantization method of the invention can reduce the quantization error, improve the system capacity performance and have advantages of small used codebook and low complexity when the speed and capacity are the same.

Description

Multi-user MIMO system precoding method and device based on Limited Feedback

Technical field

The present invention relates to a kind of method and apparatus of radio transmission technical field, specifically is a kind of multi-user MIMO system precoding method and device based on Limited Feedback.

Background technology

Use many antennas on transmitter and receiver, promptly constituted multiple-input and multiple-output (MIMO) mode, this transmission means can strengthen the reliability of wireless connections and increase the capacity of wireless communication system.In the comparatively serious wireless environment of electromagnetic interference, have M for one _TTransmit antennas and M _RThe system of root reception antenna, the contribution that the MIMO transmission means improves channel capacity, with number of transmit antennas and reception antenna number, less relatively numerical value is linear.Therefore, the MIMO technology has become the outstanding alternative of following wide-band mobile communication system.Precoding technique can better overcome the decline of channel to signal at the fading characteristic of wireless channel with the feasible signal that sends of the mode of weighting, better follows the tracks of the variation of channel status, is one of effective ways that improve the mimo system performance.In the last few years, because the development of MIMO technology makes precoding technique become the important research direction of wireless communication field.Precoding technique was conceived to improve the message transmission rate of radio communication, became the part of IEEE 802.16e standard in 2005, comprised the method for precoding of open loop and closed loop.In the 3GPP standard, also used method for precoding based on the Limited Feedback of precoding.

Because the capacity of mimo system is relevant with the minimum antenna number of transmitting terminal and receiving terminal, and in real system, the mobile terminal is owing to the restriction of volume and power, and the quantity of antenna suffers restraints, and the base station is then not limited.Therefore, for implementation space diversity in this case, (Broadcast Channel, research BC) has caused academia's extensive interest to the multiuser MIMO broadcast channel in recent years.Existing result of study shows, (Dirty Paper Coding DPC) can reach the capacity territory of MIMO broadcast channel to dirty paper code.Yet, realize that at transmitting terminal and receiving terminal the required extra complexity of DPC is high.And the problem of how to construct DPC sign indicating number practical, that reach power system capacity is not resolved so far.

It is low that the linear predictive coding technology has a computation complexity, realizes simple advantage.Existing result of study shows, close-to zero beam is shaped, and (Zero-foring Beamforming ZFBF) can eliminate interference between each user antenna, and performance is near optimum.In actual Frequency Division Duplexing (FDD) (FDD) system, channel condition information (CSI) can be obtained by the channel estimating of receiving terminal, and is transferred to transmitting terminal by feedback channel.But,, generally can not can only represent CSI by the information of feedback limit bit with whole CSI message transmission to transmitting terminal because the bandwidth of feedback channel is limited.Because the information with limit bit quantizes CSI, exists the error of quantification.The performance that causes precoding is compared with the ideal situation of the complete known CSI of transmitting terminal, and bigger performance loss is arranged.

Find through literature search prior art, Nihar Jindal is in IEEE InternationalSymposium on Information Theory (IEEE information theory international conference, in July, 2006,2699-2703 page or leaf) " A Feedback Reduction Technique for MIMO BroadcastChannels " (" a kind of MIMO broadcast channel technology of low-feedback ") delivered on.In this article, under the situation to many antennas receiver user, proposed a kind of method for precoding that reduces number of bits of feedback, wherein the concrete grammar of precoding is: user's receiving terminal at first estimates the channel state vectors of each antenna; Then from the traversal code book of structure in advance, therefrom select the space that constituted with each antenna channel state vector, the code word of angle minimum is come the channel state vectors of representative of consumer; At last selected codeword number is fed back to transmitting terminal, generate pre-coding matrix by transmitting terminal.The advantage of this method is: under same codebook, reduced average quantization error, thereby can reduce the bit number of feedback.Its deficiency is: the situation of multi-aerial receiver is deteriorated to the single antenna receiver, thereby correspondingly, (BlockDiagonalization, BD) the precoding algorithm deteriorates to zero forcing algorithm to block diagonalization, makes the degree of freedom of precoding descend.Therefore, the performance of system has the leeway of further raising.

Summary of the invention

The objective of the invention is to overcome deficiency of the prior art, a kind of multi-user MIMO system precoding method and device based on Limited Feedback is provided, select several and the less alternative code word of reception antenna vector space, from alternative code word, select the code word of representing channel condition information again, when reducing quantization error, utilize block diagonalization precoding design freedom height, advantage that performance is better than zero forcing algorithm, improved the performance of multi-user pre-coding.

The present invention is achieved by the following technical solutions:

The method of multi-user MIMO system precoding based on Limited Feedback involved in the present invention comprises the steps:

1. at first obtain the channel state vectors of this user's every antenna of receiver at each user's receiving terminal, then according to from the code book that pre-sets, the code word of selecting several and channel state vectors space angle minimum is as alternative code word, and obtain the equivalent channel state vector and the weight vector of each alternative code word, from alternative code word, select N _iIndividual code word vector also obtains the equivalent channel state matrix (N of corresponding weights matrix and receiver user _iBe receiver user antenna number), and the numbering of the code word vector selected sent back the base station by feedback channel;

2. base station receiving feedback information, numbering according to the code word vector of code book of making an appointment and feedback, each user's of reconstruct channel state vectors, carry out the diagonalizable precoding of multi-user block according to the channel state vectors of reconstruct then, after user's receiving terminal receives the signal of precoding, at first carry out preprocessing process, precoding and weight matrix are multiplied each other, the channel matrix of reality is converted to corresponding equivalent channel matrix.

The described code word of selecting several and channel state vectors space angle minimum comprises the steps: as alternative code word

The first step is to the channel state vectors h of receiving terminal acquisition ₁, h ₂...,

Carry out the orthogonalization conversion, obtain one group of orthonormal basis q in space that channel state vectors constitutes ₁, q ₂...,

Wherein, N _iBe the antenna number of i receiver user, h ₁, h ₂...,

Channel state vectors for every antenna of receiver user;

Second step, from code book, pick out several code words as alternative code word, the code word number of being selected should be greater than the antenna number of user's receiver, and the method for selecting alternative code word is: travel through whole code book set, select and the channel vector space

The code word of angle minimum is used w ₁, w ₂..., The alternative code word that expression is selected, n _aBe the number of alternative code word,

For answering alternative code word and space

Angle, and i user's channel state matrix is expressed as: $H_i={\left[\begin{array}{cccc}{h}_1^H& h_2^H& \·\·\·& h_{N_i}^H\end{array}\right]}^H.$

Described equivalent channel state vector and the weight vector of obtaining each alternative code word is specially: to each alternative code word w _j, according to it in the channel vector space

On projection vector

Obtain its corresponding equivalent channel state vector and weight vector, the equivalent channel vector

Be the vector on the alternative code word projecting direction, weight vector γ _jConstitute by the scale factor between the actual channel vector of equivalent channel vector and reception antenna, be specially:

${\hat{h}}_i=\frac{w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}{H}_i}{\left|\right|w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}\left|\right|}$ (1)

${\mathrm{\γ}}_j=\frac{w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}}{\left|\right|w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}\left|\right|}$

Wherein: H _iIt is i user's channel state matrix.

Select N in the described alternative code word _iIndividual code word vector comes the channel condition information of representative of consumer, and gets access to corresponding equivalent channel state vector and weight vector, and is specific as follows:

The first step, initialization: establish matrix A=I _M, wherein, M is the number of transmit antennas of base station transmitter, selects code word with channel vector space angle minimum as first code word from alternative code word, being numbered of this code word: ${\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1={\arg \max }_{1\&le;l\&le;n_a}{\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">cos</figure-callout>}}^2{\mathrm{\&alpha;}}_l,$ Note n=1;

In second step, upgrade matrix $A=A-A{\hat{h}}_{r_n}^H{(1+{\hat{h}}_{r_n}A{\hat{h}}_{r_n}^H)}^{-1}{\hat{h}}_{r_n}A,$ Select n+1 code word to be, $r_{n+1}={\arg \min }_{l\&NotElement;\{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,r_n\}}{\hat{h}}_{l}A{\hat{h}}_l^H,$ Upgrade n=n+1;

In the 3rd step, repeated for second step up to selecting whole N _iIndividual code word vector, i.e. n=N _i, the code word vector of selecting makes the value maximum of formula (2), the selected N that goes out _iIndividual code word be numbered r ₁, r ₂..., Corresponding equivalent channel vector is Corresponding weight vector is ${\mathrm{\&gamma;}}_{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&gamma;}}_{r_N}.$

$C({\hat{h}}_{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1},{\hat{h}}_{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_{n+1}})={\mathrm{<figure-callout\; id="2"\; label="log"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">log</figure-callout>}}_2|I_M+{\hat{H}}_{n+1}^H{\hat{H}}_{n+1}|---\left(2\right)$

Wherein, Constitute by the code word vector of having selected:

${\hat{H}}_{n+1}={\left[\begin{array}{cccc}{\hat{h}}_{r_1}^H& {\hat{h}}_{r_2}^H& \&CenterDot;\&CenterDot;\&CenterDot;& {\hat{h}}_{r_{n+1}}^H\end{array}\right]}^H---\left(3\right)$

The code book that described basis is made an appointment and the codeword number of feedback, reconstruct user's channel state matrix

Specific as follows: ${\tilde{H}}_i={\left[\begin{array}{cccc}{w}_{r_1}^H& w_{r_2}^H& \&CenterDot;\&CenterDot;\&CenterDot;& w_{r_{N_i}}^H\end{array}\right]}^H---\left(4\right).$

The described channel state matrix that obtains according to reconstruct is carried out the precoding of block diagonalization, obtains each user's pre-coding matrix V after its precoding ₁, V ₂..., V _K, establishing and sending to i user's data is s _i, then the transmission signal after the precoding is $x=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{V}_i{s}_i,$ The signal of i user's reception is y _i=H _iX+n _i, wherein, H _iBe channel state matrix, n _iBe additive white Gaussian noise.

Described precoding and weight matrix are multiplied each other, thereby the channel matrix of reality is converted to corresponding equivalent channel matrix, specific as follows:

$y_i^{\mathrm{eff}}={\mathrm{\&Gamma;}}_i{y}_i={\mathrm{\&Gamma;}}_i{H}_{i}x+n_i---\left(5\right)$

Wherein: Γ _iBe weight matrix, y _iBe the signal of user's reception,

Received signal for equivalence.

In the present invention, according to (5) formula, the model of equivalent received signals, i user's speed is:

$R_i={\log }_2\frac{|I+\frac{P}{M}{\mathrm{\&Gamma;}}_i{H}_i\left({\mathrm{\&Sigma;}}_{j=1}^K{V}_j{V}_j^H\right){\left({\mathrm{\&Gamma;}}_i{H}_i\right)}^H|}{|I+\frac{P}{M}{\mathrm{\&Gamma;}}_i{H}_i\left({\mathrm{\&Sigma;}}_{j\&NotEqual;i}{V}_j{V}_j^H\right){\left({\mathrm{\&Gamma;}}_i{H}_i\right)}^H|}---\left(6\right)$

Wherein, P is the transmitting power of transmission end of base station.

By (6) formula as can be known, the performance of system depends on two aspects: the one, and shown in (6) formula molecule, the mould value size of equivalent channel matrix and pre-coding matrix product; Another is shown in (6) formula denominator, because quantization error, the phase mutual interference between each user fully causes the loss of performance.In order to improve the former, i.e. (6) formula bulk of molecule should allow between the equivalent channel vector quadrature as much as possible; And in order to reduce the interference between the user, i.e. the size of (6) formula denominator should reduce the size of quantization error, promptly selects for use and the space

The less code word of angle.These two parts of combined optimization make (6) formula best performance relatively more difficult.Therefore, in the present invention, at first select n _aIndividual code word is as alternative code word, to limit the size of quantization error.Then, in alternative code word, select as much as possible and make the maximum codeword set of mould value size of equivalent channel matrix and pre-coding matrix product, to improve the performance of system.

The device of the multi-user MIMO system precoding based on Limited Feedback involved in the present invention comprises: receiving terminal and transmitting terminal, wherein:

After receiving terminal receives the information of transmitting terminal emission, finish obtaining, select the code word vector, calculating equivalent channel matrix of channel condition information, feedback information is launched;

After transmitting terminal receives feedback information, finish feedback information reconstruct, pre-coding matrix calculating, and use the pre-coding matrix that calculates that the data of preparing to transmit are carried out precoding, finally form missile information.

Described receiving end device comprises: channel condition information acquiring unit, alternative code word selected cell, equivalent channel state vector computing unit, code word vector selected cell, quantitative information feedback unit, receiving terminal pretreatment unit, wherein:

The channel condition information acquiring unit obtains channel state vectors, and channel state vectors is transferred to alternative code word selected cell;

Alternative code word selected cell is according to the channel state vectors that obtains, from predefined code book, select alternative code word vector, at first one group of standard quadrature in calculating channel space that state vector constitutes is long-pending, calculate the angle of each code word and channel state vectors space in the code book then, from result of calculation, select minimum angle code word at last, constitute alternative codeword set;

Equivalent channel state vector computing unit calculates its equivalent channel state vector and corresponding weights vector to each code word in the alternative codeword set;

Code word vector selected cell from alternative codeword set, is selected the code word vector of representing this user's equivalent channel state vector according to the equivalent channel state vector;

The numbering of code word vector in code book that the quantitative information feedback unit is selected code word vector selected cell feeds back to transmitting terminal;

The receiving terminal pretreatment unit multiplies each other data-signal and the weight matrix that receives, and user's channel state matrix is converted into the equivalent channel state matrix, and wherein, weight matrix constitutes by selecting the pairing weight vector of code word.

Described transmitting terminal device comprises: feedback information reconfiguration unit, precoding unit, wherein:

The feedback information reconfiguration unit receives feedack, reconstruct user's equivalent channel state matrix;

Precoding unit adopts the method for block diagonalization according to each user's equivalent channel state matrix, calculates corresponding pre-coding matrix.

The corresponding relation of the functional unit of said apparatus and the inventive method performing step is: receiving terminal channel condition information acquiring unit completing steps one; Alternative code word selected cell completing steps two; Equivalent channel state vector computing unit completing steps three; Code word vector selected cell completing steps four; Quantitative information feedback unit completing steps five; Receiving terminal pretreatment unit completing steps eight.Transmitting terminal feedback information reconfiguration unit completing steps six; Precoding unit completing steps seven.

Compared with prior art, the present invention has following beneficial effect:

The present invention proposes a kind of new multi-user MIMO system precoding method, at first select several and the less alternative code word of reception antenna vector space, from alternative code word, select the code word of representing channel condition information again, when reducing quantization error, utilize block diagonalization precoding design freedom height, advantage that performance is better than zero forcing algorithm, improved the performance of multi-user pre-coding.With respect to existing method, the speed of system of the present invention and Geng Gao, be well suited for using in the reality, can be the third generation (3G), super three generations (B3G), the 4th generation (4G) cellular mobile communication and the technical scheme of systems such as Digital Television, WLAN (wireless local area network), wireless wide area network important theory foundation and concrete implementation method are provided.

Description of drawings

Fig. 1 is based on the fundamental diagram of Limited Feedback multi-user MIMO pre-encoding system;

Fig. 2 is the system architecture diagram of apparatus of the present invention;

Fig. 3 is the workflow diagram of the inventive method;

Fig. 4 be system's speed and capacity simulation curve figure under the MIMO broadcast channel (M=4, N=2);

Fig. 5 be system's speed and capacity simulation curve figure under the MIMO broadcast channel (M=6, N=3);

Fig. 6 is parameter n _aTo the speed and capacity simulation curve figure of system (M=4, N=2, SNR=15dB);

Fig. 7 is parameter n _aTo the speed and capacity simulation curve figure of system (M=6, N=3, SNR=20dB).

Embodiment

Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed execution mode and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

Present embodiment is applied to the multi-user MIMO system based on Limited Feedback, as shown in Figure 1, multi-user MIMO system comprise 1 base station and K user (user 1, and the user 2 ..., user K), there is the M transmit antennas base station, and each receiver user has N root reception antenna.Among Fig. 1, Z _i(i=1,2 ..., K) for sending to i user's data; T _iBe expressed as the linear predictive coding process that i user sends data; R _iBe expressed as i user's receiver.

The channel model that present embodiment is used is MIMO broadcast channel (MIMO Broadcast Channel), specifically please refer to " Multiple Antenna Broadcast Channels With Shape Feedbackand Limited Feedback " (" waveform feedback and Limited Feedback under the multi-antenna broadcast channel "), IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL.55, NO.7, July 2007 (IEEE signal processing transactions, in July, 2007, the 55th volume, the 7th phase).

As shown in Figure 3, present embodiment relates to a kind of multi-user MIMO system precoding method based on Limited Feedback, comprises the steps:

Step 1 obtains the channel state vectors of this user's every antenna of receiver at each user's receiving terminal;

Step 2, precoding codebook

Present embodiment does not relate to the design problem of code book, and the random vector code book of 16 bit sizes is all used in all emulation, analysis and the explanation of being done.Promptly from M dimension space unit ball, select 65,536 vectors to constitute the code word of code book at random, equably.

Step 3, alternative code word is selected

The selection course of alternative code word is, each code word in the traversal code book is calculated the angle in itself and channel vector space, therefrom selects the n of angle minimum _aIndividual code word constitutes alternative codeword set.Wherein, n _aBe preset parameters, and n _a〉=N _i

For compute codeword and channel vector space

Angle, at first use Schmidt (Schimidt) orthogonalization procedure to calculate one group of orthonormal basis

q ₁＝h ₁/‖h ₁‖

${\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">q</figure-callout>}}_2=(h_2-h_2{q}_1^H)/\left|\right|(h_2-h_2{q}_1^H)\left|\right|$ (7)

...

$q_{N_i}=(h_{N_i}-h_{N_i-1}{q}_{N_i-1}^H-\&CenterDot;\&CenterDot;\&CenterDot;-h_{N_i-1}{q}_1^H)/\left|\right|(h_{N_i}-h_{N_i-1}{q}_{N_i-1}^H-\&CenterDot;\&CenterDot;\&CenterDot;-h_{N_i-1}{q}_1^H)\left|\right|$

Then the cosine value of the angle in code word vector and channel vector space square, equal the quadratic sum of the inner product of code word vector and each orthogonal basis, promptly ${\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">cos</figure-callout>}}^2{\mathrm{\&alpha;}}_j=\underset{k=1}{\overset{N_i}{\mathrm{\&Sigma;}}}{\left|w_j^H{q}_k\right|}^2.$ Select the n of angle minimum _aIndividual code word, equivalence is for selecting square cos of its cosine value ²a _jMinimum n _aIndividual code word.

Step 4 gets access to corresponding equivalent channel state vector and weight vector

Calculate the equivalent channel state vector of each alternative code word correspondence respectively, the direction of equivalent channel state vector and corresponding code word w _jAt space span (h ₁, h ₂...,

) on projection

Direction is identical.

The projection vector of code word

Calculating, can equivalence be code word w _jAt each orthogonal basis q ₁, q ₂...,

On the direction projection vector and, promptly

$w_j^{\mathrm{proj}}=\frac{{\mathrm{\&Sigma;}}_{k=1}^{N_i}{q}_k\left(w_j^H{q}_k\right)}{\left|\right|{\mathrm{\&Sigma;}}_{k=1}^{N_i}{q}_k\left(w_j^H{q}_k\right)\left|\right|}---\left(8\right)$

The equivalent channel vector is the channel vector h of receiver antenna ₁, h ₂..., Linear combination, and and projection vector

Direction identical.Note equivalent channel vector ${\hat{h}}_j={\mathrm{\&gamma;}}_j{H}_i,$ γ wherein _jWeight vector, $H_i={\left[\begin{array}{cccc}{h}_1^H& h_2^H& \&CenterDot;\&CenterDot;\&CenterDot;& h_{N_i}^H\end{array}\right]}^H$ It is i user's channel state matrix.Weight vector γ _jSize to the not influence of the size of receiver input signal-to-noise ratio, also can not influence the performance of whole system.Therefore in order to find the solution γ _jAnd equivalent channel vector

Limit weight vector γ _jBe unit vector.Can try to achieve,

${\mathrm{\&gamma;}}_j=\frac{w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}}{\left|\right|w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}\left|\right|}---\left(9\right)$

The equivalent channel vector For,

${\hat{h}}_i={\mathrm{\&gamma;}}_j{H}_i=\frac{w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}{H}_i}{\left|\right|w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}\left|\right|}---\left(10\right)$

Step 5 is selected N from alternative code word _iIndividual code word vector also obtains the equivalent channel state matrix (N of corresponding weights matrix and receiver user _iBe receiver user antenna number), and the numbering of the code word vector selected sent back the base station by feedback channel;

From alternative codeword set, select N _iIndividual code word is represented receiver N as feedback code _iThe channel vector of root antenna.In order to improve the performance of system, wish the selected feedback code that goes out, can make $C_{\mathrm{opt}}({\hat{h}}_{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1},{\hat{h}}_{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_{N_i}})={\mathrm{<figure-callout\; id="2"\; label="log"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">log</figure-callout>}}_2|I_M+{\hat{H}}_{N_i}^H{\hat{H}}_{N_i}|$ Maximization, wherein

Equivalent channel vector by the feedback code correspondence of picking out constitutes, ${\hat{H}}_{N_i}={[{\hat{h}}_{r_1}^H,{\hat{h}}_{r_2}^H,\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_{N_i}}^H]}^H.$

N therefrom _aSelect N in the individual alternative code word _iIndividual code word, total $\left(\begin{array}{c}{n}_a\\ N_i\end{array}\right)$ Plant combination.To every kind of code word combination, calculate C respectively _OptValue, the expense of calculating is very big.Therefore, in real system, adopt a kind of system of selection of suboptimum.Promptly at every turn from unselected alternative code word, select a code word vector, make $C({\hat{h}}_{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1},{\hat{h}}_{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_{n+1}})={\mathrm{<figure-callout\; id="2"\; label="log"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">log</figure-callout>}}_2|I_M+{\hat{H}}_{n+1}^H{\hat{H}}_{n+1}|$ Maximum is up to selecting whole N _iIndividual code word vector.Wherein,

Constitute by the code word vector of having selected,

${\hat{H}}_{n+1}={\left[\begin{array}{cccc}{\hat{h}}_{r_1}^H& {\hat{h}}_{r_2}^H& \&CenterDot;\&CenterDot;\&CenterDot;& {\hat{h}}_{r_{n+1}}^H\end{array}\right]}^H$

In each selection code word process, recomputate Overhead.Utilize following iterative relation,

$C({\hat{h}}_{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1},{\hat{h}}_{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_{n+1}})=C({\hat{h}}_{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1},{\hat{h}}_{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_n})+{\log }_2(1+{\hat{h}}_{r_{n+1}}{(I_M+{\hat{H}}_{r_n}^H{\hat{H}}_{r_n})}^{-1}{\hat{h}}_{r_{n+1}}^H)---\left(11\right)$

Therefore, select being numbered of code word

$r_{n+1}={\arg \max }_{l\&NotElement;\{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,r_n\}}{\hat{h}}_{r_{n+1}}{(I_M+{\hat{H}}_{r_n}^H{\hat{H}}_{r_n})}^{-1}{\hat{h}}_{r_{n+1}}^H---\left(12\right)$

According to the matrix inversion theorem as can be known, the process of (12) formula matrix inversion can be transformed to the inverse that calculates scalar numeric value,

${\hat{h}}_{r_{n+1}}{(I_M+{\hat{H}}_{r_n}^H{\hat{H}}_{r_n})}^{-1}{\hat{h}}_{r_{n+1}}^H={\hat{h}}_{r_{n+1}}(I_M+{\hat{H}}_{r_n}^H{(1+{\hat{H}}_{r_n}{\hat{H}}_{r_n}^H)}^{-1}{\hat{H}}_{r_n}){\hat{h}}_{r_{n+1}}^H---\left(13\right)$

Before selecting n+1 code word, calculated for $A={(I_M+{\hat{H}}_{r_{n-1}}^H{\hat{H}}_{r_{n-1}})}^{-1},$ By the matrix inversion theorem,

${(I_M+{\hat{H}}_{r_n}^H{\hat{H}}_{r_n})}^{-1}={(A^{-1}+{\hat{h}}_{r_{n+1}}^H{\hat{h}}_{r_{n+1}})}^{-1}$ (14)

$=A-A{\hat{h}}_{r_n}^H{(1+{\hat{h}}_{r_n}A{\hat{h}}_{r_n}^H)}^{-1}{\hat{h}}_{r_n}A$

Equally, use (14) formula and upgrade matrix $A=A-A{\hat{h}}_{r_n}^H{(1+{\hat{h}}_{r_n}A{\hat{h}}_{r_n}^H)}^{-1}{\hat{h}}_{r_n}A,$ Avoided the expense of matrix inversion.

Then, according to (12) formula and (13) formula, select n+1 code word, promptly

$r_{n+1}={\arg \min }_{l\&NotElement;\{{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="A200810201254E00181.png,A200810201254E00201.png"\; state="\{\{state\}\}">r</figure-callout>}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,r_n\}}{\hat{h}}_{l}A{\hat{h}}_l^H---\left(15\right)$

Step 6, the base station receiving feedback information, numbering according to the code word vector of code book of making an appointment and feedback, each user's of reconstruct channel state vectors, carry out the diagonalizable precoding of multi-user block according to the channel state vectors of reconstruct then, after user's receiving terminal receives the signal of precoding, at first carry out preprocessing process, precoding and weight matrix are multiplied each other, the channel matrix of reality is converted to corresponding equivalent channel matrix.

As shown in Figure 2, present embodiment relates to a kind of multi-user MIMO system precoding device based on Limited Feedback, comprising: receiving terminal and transmitting terminal, wherein:

After receiving terminal receives the information of transmitting terminal emission, finish obtaining, select the code word vector, calculating equivalent channel matrix of channel condition information, feedback information is launched;

After transmitting terminal receives feedback information, finish feedback information reconstruct, pre-coding matrix calculating, and use the pre-coding matrix that calculates that the data of preparing to transmit are carried out precoding, finally form missile information.

Described receiving end device comprises: channel condition information acquiring unit, alternative code word selected cell, equivalent channel state vector computing unit, code word vector selected cell, quantitative information feedback unit, receiving terminal pretreatment unit, wherein:

The channel condition information acquiring unit obtains channel state vectors, and channel state vectors is transferred to alternative code word selected cell;

Alternative code word selected cell is according to the channel state vectors that obtains, from predefined code book, select alternative code word vector, at first one group of standard quadrature in calculating channel space that state vector constitutes is long-pending, calculate the angle of each code word and channel state vectors space in the code book then, from result of calculation, select minimum angle code word at last, constitute alternative codeword set;

Equivalent channel state vector computing unit calculates its equivalent channel state vector and corresponding weights vector to each code word in the alternative codeword set;

Code word vector selected cell from alternative codeword set, is selected the code word vector of representing this user's equivalent channel state vector according to the equivalent channel state vector;

The numbering of code word vector in code book that the quantitative information feedback unit is selected code word vector selected cell feeds back to transmitting terminal;

The receiving terminal pretreatment unit multiplies each other data-signal and the weight matrix that receives, and user's channel state matrix is converted into the equivalent channel state matrix, and wherein, weight matrix constitutes by selecting the pairing weight vector of code word.

Described transmitting terminal device comprises: feedback information reconfiguration unit, precoding unit, wherein:

The feedback information reconfiguration unit receives feedack, reconstruct user's equivalent channel state matrix;

Precoding unit adopts the method for block diagonalization according to each user's equivalent channel state matrix, calculates corresponding pre-coding matrix.

The present embodiment effect:

When present embodiment was applied to the multi-user MIMO system of 4 transmitting antennas, 2 reception antennas, Fig. 4 was the speed and the performance simulation curve chart of this embodiment system.

Comprised the performance simulation result who adopts method for precoding of the present invention (BD) among Fig. 4, simultaneously as a comparison, provided Nihar Jindal in " A Feedback Reduction Technique for MIMOBroadcast Channels " (" a kind of MIMO broadcast channel technology of low-feedback "), InformationTheory, 2006 IEEE International Symposium on (IEEE information theory international conferences, the performance simulation result of the algorithm (ZF) that proposes in July, 2006,2699-2703 page or leaf).

In the present embodiment, compare with the method that Nihar Jindal proposes, it is better that the systematic function of present embodiment method is wanted.Work as n _a=2 o'clock, present embodiment method, system's speed and performance improved 4.0 (bit/s/Hz); Work as n _a=6 o'clock, system's speed and performance improved 5.6 (bit/s/Hz).

Give the complete known channel state information of transmitting terminal (CSIT) among Fig. 4, do not exist to quantize under the error condition, the performance simulation result of block diagonalization (BD) precoding is as the performance upper bound of BD algorithm.Compare with the performance upper bound of BD algorithm, under low signal-to-noise ratio, very near the upper bound of BD algorithm, still, along with signal to noise ratio increases, the loss of performance also increases gradually for present embodiment method, performance.

Fig. 6 is parameter n _aInfluence to system's speed and capacity.Under SNR=15dB, work as n _a=8 o'clock, system's speed and maximum.n _aAfter 8, and because quantization error increasing, system speed and with n _aIncrease and reduce.

When present embodiment was applied to the multi-user MIMO system of 6 transmitting antennas, 3 reception antennas, Fig. 5 was the speed and the performance simulation curve chart of this embodiment system.

Comprised the performance simulation result who adopts the present embodiment method among Fig. 5, simultaneously as a comparison, provided Nihar Jindal in " A Feedback Reduction Technique for MIMO BroadcastChannels " (" a kind of MIMO broadcast channel technology of low-feedback "), Information Theory, 2006 IEEE International Symposium on (IEEE information theory international conferences, the performance simulation result of the algorithm (ZF) that proposes in July, 2006,2699-2703 page or leaf).

In the present embodiment, compare with the method that Nihar Jindal proposes, it is better that the systematic function of present embodiment method is wanted.Work as n _a=3 o'clock, present embodiment method, system's speed and performance improved 8.5 (bit/s/Hz); Work as n _a=7 o'clock, system's speed and performance improved 10.1 (bit/s/Hz).

Give the complete known channel state information of transmitting terminal (CSIT) among Fig. 5, do not exist to quantize under the error condition, the performance simulation result of block diagonalization (BD) precoding is as the performance upper bound of BD method.Compare with the performance upper bound of BD method, under low signal-to-noise ratio, very near the upper bound of BD method, still, along with signal to noise ratio increases, the loss of performance also increases gradually for present embodiment method, performance.

Fig. 7 is parameter n _aInfluence to system's speed and capacity.Under SNR=15dB, work as n _a=7 o'clock, system's speed and maximum.n _aAfter 7, and because quantization error increasing, system speed and with n _aIncrease and reduce.

Claims (5)

1. the multi-user MIMO system precoding method based on Limited Feedback is characterized in that, may further comprise the steps:

1. at first obtain the channel state vectors of this user's every antenna of receiver at each user's receiving terminal, then according to from the code book that pre-sets, the code word of selecting several and channel state vectors space angle minimum is as alternative code word, and obtain the equivalent channel state vector and the weight vector of each alternative code word, from alternative code word, select N _iIndividual code word vector also obtains the equivalent channel state matrix of corresponding weights matrix and receiver user, described N _iBe receiver user antenna number, and the numbering of the code word vector selected is sent back the base station by feedback channel;

2. base station receiving feedback information, numbering according to the code word vector of code book of making an appointment and feedback, each user's of reconstruct channel state vectors, carry out the diagonalizable precoding of multi-user block according to the channel state vectors of reconstruct then, after user's receiving terminal receives the signal of precoding, carry out preprocessing process earlier, precoding and weight matrix are multiplied each other, the channel matrix of reality is converted to corresponding equivalent channel matrix.

2. the multi-user MIMO system precoding method based on Limited Feedback as claimed in claim 1 is characterized in that, the described code word of selecting several and channel state vectors space angle minimum comprises the steps: as alternative code word

The first step is to the channel state vectors h of receiving terminal acquisition ₁, h ₂...,

Row orthogonalization conversion obtains one group of orthonormal basis q in space that channel state vectors constitutes ₁, q ₂..., Wherein, N _iBe the antenna number of i receiver user, h ₁, h ₂...,

Channel state vectors for every antenna of receiver user;

Second step, from code book, pick out several code words as alternative code word, the code word number of being selected should be greater than the antenna number of user's receiver, and the method for selecting alternative code word is: travel through whole code book set, select and channel vector space span (h ₁, h ₂...,

) code word of angle minimum, use w ₁, w ₂...,

The alternative code word that expression is selected, n _aBe the number of alternative code word, α ₁, α ₁...,

Be corresponding alternative code word and space span (h ₁, h ₂...,

) angle, and i user's channel state matrix is expressed as: $H_i={\left[\begin{array}{cccc}{h}_1^H& h_2^H& \&CenterDot;\&CenterDot;\&CenterDot;& h_{N_i}^H\end{array}\right]}^H.$

3. the multi-user MIMO system precoding method based on Limited Feedback as claimed in claim 1 is characterized in that, described equivalent channel state vector and the weight vector of obtaining each alternative code word is specially: to each alternative code word w _j, according to it at channel vector space span (h ₁, h ₂..., ) on projection vector

Obtain its corresponding equivalent channel state vector and weight vector, the equivalent channel vector

Be the vector on the alternative code word projecting direction, weight vector γ _jConstitute by the scale factor between the actual channel vector of equivalent channel vector and reception antenna, be specially:

${\hat{h}}_i=\frac{w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}{H}_i}{\left|\right|w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}\left|\right|}$

${\mathrm{\&gamma;}}_j=\frac{w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}}{\left|\right|w_j^{\mathrm{proj}}{H}_i^H{\left(H_i{H}_i^H\right)}^{-1}\left|\right|}$

Wherein: H _iIt is i user's channel state matrix.

4. the multi-user MIMO system precoding method based on Limited Feedback as claimed in claim 1 is characterized in that, selects N in the described alternative code word _iIndividual code word vector comes the channel condition information of representative of consumer, and obtains the equivalent channel state matrix of corresponding weights matrix and receiver user, and is specific as follows:

The first step, initialization: establish matrix A=I _M, wherein, M is the number of transmit antennas of base station transmitter, selects angle α from alternative code word _iMinimum code word is as first code word, being numbered of this code word:

$r_1=\arg {\max }_{1\&le;l\&le;n_a}{\cos }^2{\mathrm{\&alpha;}}_l,$ Note n=1;

In second step, upgrade matrix $A=A-A{\hat{h}}_{r_n}^H{(1+{\hat{h}}_{r_n}A{\hat{h}}_{r_n}^H)}^{-1}{\hat{h}}_{r_n}A,$ Select n+1 code word to be, $r_{n+1}=\arg {\min }_{l\&NotElement;\{r_1,\&CenterDot;\&CenterDot;\&CenterDot;,r_n\}}{\hat{h}}_{l}A{\hat{h}}_l^H,$ Upgrade n=n+1;

In the 3rd step, repeated for second step up to selecting whole N _iIndividual code word vector, i.e. n=N _i, the value maximum of formula below the code word vector of selecting makes: $C({\hat{h}}_{r_1},{\hat{h}}_{r_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{h}}_{r_{n+1}})={\log }_2|I_M+{\hat{H}}_{n+1}^H{\hat{H}}_{n+1}|,$ Wherein, Constitute by the code word vector of having selected: ${\hat{H}}_{n+1}={\left[\begin{array}{cccc}{\hat{h}}_{r_1}^H& {\hat{h}}_{r_2}^H& \&CenterDot;\&CenterDot;\&CenterDot;& {\hat{h}}_{r_{n+1}}^H\end{array}\right]}^H,$ The selected N that goes out _iIndividual code word be numbered r ₁, r ₂..., Corresponding equivalent channel vector is

Corresponding weight vector is

5. device based on the multi-user MIMO system precoding of Limited Feedback, comprise: receiving terminal and transmitting terminal, after receiving terminal receives the information of transmitting terminal emission, finish obtaining, select the code word vector, calculating equivalent channel matrix of channel condition information, feedback information is launched; After transmitting terminal receives feedback information, finish feedback information reconstruct, pre-coding matrix calculating, and use the pre-coding matrix that calculates that the data of preparing to transmit are carried out precoding, finally form missile information;

Described receiving end device, comprise: channel condition information acquiring unit, alternative code word selected cell, equivalent channel state vector computing unit, code word vector selected cell, quantitative information feedback unit, receiving terminal pretreatment unit, wherein: the channel condition information acquiring unit obtains channel state vectors, and channel state vectors is transferred to alternative code word selected cell; Alternative code word selected cell is according to the channel state vectors that obtains, from predefined code book, select alternative code word vector, at first one group of standard quadrature in calculating channel space that state vector constitutes is long-pending, calculate the angle of each code word and channel state vectors space in the code book then, from result of calculation, select minimum angle code word at last, constitute alternative codeword set; Equivalent channel state vector computing unit calculates its equivalent channel state vector and corresponding weights vector to each code word in the alternative codeword set; Code word vector selected cell from alternative codeword set, is selected the code word vector of representing this user's equivalent channel state vector according to the equivalent channel state vector; The numbering of code word vector in code book that the quantitative information feedback unit is selected code word vector selected cell feeds back to transmitting terminal; The receiving terminal pretreatment unit multiplies each other data-signal and the weight matrix that receives, and user's channel state matrix is converted into the equivalent channel state matrix, and wherein, weight matrix constitutes by selecting the pairing weight vector of code word;

Described transmitting terminal device comprises: feedback information reconfiguration unit, precoding unit, and wherein: the feedback information reconfiguration unit receives feedack, reconstruct user's equivalent channel state matrix; Precoding unit adopts the method for block diagonalization according to each user's equivalent channel state matrix, calculates corresponding pre-coding matrix.

Images