CN101567716B - Orthogonal random beam forming transmission method based on partial channel information feedback - Google Patents

Abstract

The invention proposes an orthogonal random beam forming transmission method based on partial channel information feedback, which comprises the steps: a transmitting terminal transmits pilot signal vector to each receiving terminal which obtains a channel matrix by means of channel estimation; and then singular value decomposition is conducted on the channel matrix and partial channel information obtained from decomposition is fed back to the transmitting terminal which then determines alternative users and new orthogonal transmitting beams. The transmitting terminal selects a beam from the new orthogonal transmitting beams to send data to a user. The selected user receives a data signal and water filtering is conducted on the data signal. Based on the orthogonal random beam forming proposal and the beam reforming proposal of partial channel information feedback, the method eliminates mutual interference among multi-user beams while having relatively low feedback consumption, thereby greatly improving the capacity of a multi-user system.

Description

A kind of orthogonal random beam forming transmission method based on the partial channel knowledge feedback

Technical field

The present invention relates to multiple-input and multiple-output (MIMO) wireless communication system transmission technology; Particularly relate to a kind of MIMO transmission method that combines random beam forming and partial channel knowledge feedback wave beam forming; Be applicable to and have a transmitting terminal and a plurality of receiving terminal; Transmitting terminal comprises many transmit antennas, and receiving terminal comprises the communication system of an antenna.

Background technology

Along with the rapid increase of mobile communication subscriber quantity and the rapid rise of wireless broadband business (like multimedia service), people expect that future mobile communication system can provide higher data transmission rate (more than the 100Mbps), higher spectrum efficiency (more than the 10bps/Hz).

Multiple-input and multiple-output (MultipleInput Multiple Output; MIMO) be a kind of technology of using many antennas at transmitting terminal and receiving terminal simultaneously; This technology also can be reduced to the single output of many inputs (the Multiple Input SingleOutput that only uses many antennas at transmitting terminal; MISO) technology and the many output of single input (SingleInput Multiple Output, the SIMO) technology of only using many antennas at receiving terminal.MIMO is a kind of effective means that improves spectrum efficiency, and it can increase exponentially the spectrum efficiency of system under the prerequisite that does not increase transmitting power.

Beam forming is the MIMO technology of a quasi-representative, and its basic thought is on every transmit antennas, to be multiplied by a plural number to make up a beam direction.Therefore then need be multiplied by the complex vector of a N dimension to the M transmit antennas, call beam forming vector or a wave beam to the complex vector of this N dimension usually.Can make up a plurality of wave beams for multi-transmitting antenna system and launch data concurrently.

Random beam forming and portion C SI feedback wave beam forming are two types of beam forming technology.

Random beam forming is to make up M wave beam (M is generally the transmitting terminal antenna number) randomly at transmitting terminal to carry out data transmission simultaneously.Then, receiving terminal calculates the Signal to Interference plus Noise Ratio on each wave beam and feeds back maximum Signal to Interference plus Noise Ratio and corresponding beam index value is given transmitting terminal.At last, transmitting terminal selects the maximum user of Signal to Interference plus Noise Ratio to launch on each wave beam.

The advantage of random beam forming is that feedback overhead is little, it is low to implement complexity; And under the enough big situation of number of users, can obtain most multi-user diversity gain; Random beam forming can only obtain the part multi-user diversity gain under the situation but count the actual user, is far from reaching the capacity of multi-user MIMO system.Simultaneously, there is the interference between bigger multi-user beam in this transmission plan, even under antenna number expends condition with higher, systematic function also is difficult to get a desired effect.

Each receiving terminal only need feed back its partial channel-state information in the beam forming of portion C SI feedback; To obtain the main singular value of channel matrix; Obtain the right singular vector of its corresponding master, the base station makes up beam vector with this, and realization beam vector and user are accurately mated.Receiver uses main left singular vector butt joint collection of letters breath to carry out filtering, obtains the higher system capacity.

Portion C SI feedback wave beam forming is that cost obtains the higher system capacity with less feedback overhead, but still there is bigger inter-beam interference in beam vector and nonopiate.Particularly, also can make a big impact to power system capacity under the strong situation of channel relevancy.

Therefore guaranteeing low feedback overhead, low complex degree is sought a kind of low interference even glitch-free beam forming technology under the prerequisite of high power capacity, have very big realistic meaning.

Summary of the invention

The object of the present invention is to provide a kind of orthogonal random beam forming transmission method based on the partial channel knowledge feedback, feedback overhead is little, and power system capacity is big, effectively eliminates and disturbs.

A kind of orthogonal random beam forming transmission method based on the partial channel knowledge feedback; Relate to a transmitting terminal and more than one receiving terminal; Transmitting terminal includes the transmitting antenna more than, and receiving terminal comprises a reception antenna, and this method is carried out according to following steps:

Step 1 transmitting terminal is to each receiving terminal emission pilot signal vector;

K receiving terminal of step 2 obtains channel matrix through channel estimating, and it is made singular value decomposition, obtains main left singular vector U _{K, 1}With binary information CSI (V _{K, 1}, λ _{K, 1}), λ _{K, 1}Be main singular value, V _{K, 1}Be λ _{K, 1}Corresponding main right singular vector, k=1,2 ..., N, N are the number of receiving terminal;

K receiving terminal of step 3 is with binary information CSI (V _{K, 1}, λ _{K, 1}) feed back to transmitting terminal;

Step 4 transmitting terminal is selected the maximum M of main singular value from N the binary information that receives, the receiving terminal that it is corresponding is as alternative user;

Step 5 transmitting terminal makes up the beam vector matrix $\widetilde{\mathrm{\&Phi;}}=[{\tilde{V}}_{\mathrm{1,1}},{\tilde{V}}_{\mathrm{2,1}},\&CenterDot;\&CenterDot;\&CenterDot;,{\tilde{V}}_{M,1}]:$ Transmitting terminal is formed preconditioning matrix Q with other M-1 alternative users' beyond p the alternative user main singular value _p=[V _1,1..., V _P-1,1, V _P+1, V _{M, 1}], ask for preconditioning matrix Q _pOrthonormal basis $q_p=[{\mathrm{\&epsiv;}}_p^1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&epsiv;}}_p^{p-1},{\mathrm{\&epsiv;}}_p^{p+1},\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&epsiv;}}_p^M],$ Again with p alternative user's the right singular value V of master _{P, 1}At orthogonal basis q _pLast projection obtains ${\stackrel{\&OverBar;}{V}}_{p,1}=V_{p,1}-\underset{i=1,i\&NotEqual;p}{\overset{M}{\mathrm{\&Sigma;}}}<{\mathrm{\&epsiv;}}_p^i,V_{p,1}>\&CenterDot;{\mathrm{\&epsiv;}}_p^i,$ Do normalization at last and handle the wave beam that obtains p alternative user ${\tilde{V}}_{p,1}={\stackrel{\&OverBar;}{V}}_{p,1}/\left|\right|{\stackrel{\&OverBar;}{V}}_{p,1}\left|\right|,$ P=1,2 ..., M;

User and corresponding wave beam that step 6 transmitting terminal selects this secondary data to send from all alternative users and beam vector matrix;

Step 7 transmitting terminal is launched data with selected wave beam to selected user;

Step 8 is chosen the user and receives data-signal, and filtering.

Technique effect of the present invention is embodied in:

Compared with prior art, the present invention has obtained the little and big advantage of portion C SI feedback wave beam forming power system capacity of random beam forming feedback overhead simultaneously.System is through rejecting the little a large number of users of power system capacity influence, and feedback fraction channel information only, the feedback overhead that has reduced system with implement complexity; Simultaneously through the orthogonalization of portion C SI beam forming is eliminated interference, and obtain the lifting of power system capacity.

Description of drawings

Fig. 1 is system framework figure of the present invention;

Fig. 2 is a flow chart of the present invention;

Fig. 3 is the graph of a relation (N=10) of power system capacity and signal to noise ratio;

Fig. 4 is the graph of a relation (SNR=15dB) of power system capacity and number of users;

Embodiment

Below in conjunction with accompanying drawing the present invention is done further description.

As shown in Figure 1, MIMO communication system of the present invention comprises a transmitting terminal and N receiving terminal, and transmitting terminal has M root antenna, and receiving terminal has L root antenna, L=1 in the present invention.

As shown in Figure 2, flow process of the present invention comprises A1, A2, and A3, A4, A5, A6, A7, A8 be totally eight steps, and be specific as follows:

In steps A 1, transmitting terminal makes up by M random wave bundle φ _m,=1 ..., beamforming matrix Φ=[φ that M forms ₁, φ ₂..., φ _M], then to each receiving terminal emission pilot signal vector X=[x ₁, x ₂..., x _M] ^T, T representes the vector transposition, then k receiving terminal (k=1,2 ..., the pilot signal T that N) receives _kBe expressed as:

$\begin{array}{c}{Y}_k=H_k\mathrm{\&Phi;X}+W_k\\ =\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{H}_k{\mathrm{\&phi;}}_i{x}_i+W_k\end{array}---\left(1\right)$

H wherein _kBe the L * M complex channel matrix of k receiving terminal, W _kAdditive noise for L * 1.Here H _k={ h _{Ji, k}} _{L * M}, h _{Ji, k}I transmitting antenna of expression transmitting terminal supposes that to the channel characteristics of j reception antenna of k receiving terminal it is independent identically distributed multiple Gaussian random variable, distributes and satisfies h _{Ji, k}～CN (0,1), H _kThrough channel is estimated to obtain.W _kBe expressed as W _k=[w _{1, k}, w _{2, k}..., w _{L, k}] ^T, be assumed to white complex gaussian noise, distribute and satisfy w _{J, k}～CN (0,1), j=1,2 ..., L.For the communication system of power limited, arranging total transmitting power is ρ.

In steps A 2, each receiving terminal processes its channel matrix respectively.K receiving terminal is to channel matrix H _k(sigular value decomposition SVD) obtains to do singular value decomposition $H_k=U_k{\mathrm{\&Lambda;}}_k{V}_k^H,$ H representing matrix transposition, U _kBe the unitary matrice of L * L, U _kFirst classify main left singular vector U as _{K, 1}, V _kBe the unitary matrice of M * M, Λ _kBe the singular value matrix of a L * M,

It is the characteristic value of channel matrix.Reception antenna is counted L=1, then λ _{K, i}=0, i=2 ..., M obtains main singular value λ then _{K, 1}Corresponding main right singular vector V _{K, 1}

In steps A 3, k receiving terminal is with the binary information CSI (V that obtains _{K, 1}, λ _{K, 1}) feed back to transmitting terminal.Can find out, only need feedback fraction CSI, i.e. V here _{K, 1}And λ _{K, 1}, and need not feed back accurate CSI, i.e. H _k

In steps A 4, transmitting terminal receives binary information CSI (V _{K, 1}, λ _{K, 1}), from N main singular value, select maximum M, the receiving terminal that the individual main singular value of this M is corresponding is formed alternative user and is collected S as alternative user.

In steps A 5, transmitting terminal is handled M the corresponding binary information of alternative user that alternative user collects among the S: transmitting terminal adopts the main right singular vector V of Ge Lamu-Schmidt (Gram-Schmidt) orthogonalization method to p alternative user _{P, 1}Do preliminary treatment, other M-1 the corresponding main right singular vector of alternative user formed preconditioning matrix Q _p, i.e. subspace V, and obtain its corresponding orthonormal basis q _p, can be expressed as

Q _p＝V _1，1，…，V _p-1，1，V _p+1，1，V _M，1] (2)

$q_p=[{\mathrm{\&epsiv;}}_p^1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&epsiv;}}_p^{p-1},{\mathrm{\&epsiv;}}_p^{p+1},\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&epsiv;}}_p^M]---\left(3\right)$

p＝1，2，…，M

Main right singular vector V with p alternative user _{P, 1}At orthogonal basis q _pLast projection can get

${\stackrel{\&OverBar;}{V}}_{p,1}=V_{p,1}-\underset{i=1,i\&NotEqual;p}{\overset{M}{\mathrm{\&Sigma;}}}<{\mathrm{\&epsiv;}}_p^i,V_{p,1}>\&CenterDot;{\mathrm{\&epsiv;}}_p^i---\left(4\right)$

Doing has after normalization is handled

${\tilde{V}}_{p,1}={\stackrel{\&OverBar;}{V}}_{p,1}/\left|\right|{\stackrel{\&OverBar;}{V}}_{p,1}\left|\right|---\left(5\right)$

Because

With q _pIn any vectorial quadrature, so With Q _pThe subspace quadrature.

According to the method described above the user is collected among the S all alternative users' main right singular vector and do same treatment, make up and obtain new beam vector matrix $\widetilde{\mathrm{\&Phi;}}=[{\tilde{V}}_{\mathrm{1,1}},{\tilde{V}}_{\mathrm{2,1}},\&CenterDot;\&CenterDot;\&CenterDot;,{\tilde{V}}_{M,1}].$

In steps A 6, transmitting terminal is selected corresponding user and beam combination based on dispatching algorithm.Dispatching algorithm can be chosen according to actual conditions; The present invention adopts following algorithm: all select alternative user to collect whole M users and corresponding wave beam

p=1 among the S at every turn; 2; ...., M.

In steps A 7, transmitting terminal is launched data with selected wave beam to selected user.

In steps A 8, the user who is chosen receives data, and carries out filtering according to existing main left singular vector to receiving data, is specially: adopt main left singular vector U _{P, 1} ^H(U _{P, 1}Obtain H representing matrix transposition in steps A 2) be chosen the data-signal x that the user receives to p _pCarry out filtering, the reception signal after then handling does

$\begin{array}{c}{\hat{Y}}_p=U_{p,1}^H{H}_p\widetilde{\mathrm{\&Phi;}}X+U_{p,1}^H{W}_p\\ ={\mathrm{\&lambda;}}_{p,1}{V}_{p,1}^H{\tilde{V}}_{p,1}{x}_p+{\mathrm{\&lambda;}}_{p,1}\underset{i=1,i\&NotEqual;p}{\overset{M}{\mathrm{\&Sigma;}}}\left(V_{p,1}^H{\tilde{V}}_{i,1}\right)x_i+U_{p,1}^H{W}_h\end{array}---\left(6\right)$

And can know by front orthogonalization process result, for $\&ForAll;p\&NotEqual;i,$ $V_{p,1}^H{\tilde{V}}_{i,1}=0,$ So have

${\hat{Y}}_p={\mathrm{\&lambda;}}_{p,1}{V}_{p,1}^H{\tilde{V}}_{p,1}{x}_p+U_{p,1}^H{W}_p---\left(7\right)$

λ _{P, 1}Represent p main singular value that is chosen the user, V _{P, 1}Represent p main right singular vector that is chosen the user,

Be expressed as p and be chosen the corresponding wave beam of user, U _{P, 1}Represent p main left singular vector that is chosen the user, W _pRepresent p noise that is chosen the user, H representing matrix transposition.

Therefore can eliminate the interference between the multi-beam fully, the signal interference ratio (SINR, Signal-to-Interference and NoiseRatio) after the individual receiver that is chosen the user of such p is handled does

${\mathrm{\&gamma;}}_p=\frac{{\left|{\mathrm{\&lambda;}}_{p,1}\right|}^2{\left|V_{p,1}^H{\tilde{V}}_{p,1}\right|}^2}{M/\mathrm{\&rho;}}---\left(8\right)$

P=1 wherein, 2 ..., M adopts average power allocation between each transmitting antenna, and p is promptly arranged _i=ρ/M.If adopt adaptive power division, then according to Lagrange (Lagrange) algorithm, under constraints ${\mathrm{\&Sigma;}}_{i=1}^M{p}_i=\mathrm{\&rho;},$ Can distribute the corresponding performance number of each wave beam down in the hope of adaptive power.

Further can obtain based on the power system capacity under the local channel feedback orthogonal beams moulding multi-user diversity do

$C=E\left\{\underset{k=1}{\overset{M}{\mathrm{\&Sigma;}}}{\log }_2(1+{\mathrm{\&gamma;}}_k)\right\}=E\left\{\underset{k=1}{\overset{M}{\mathrm{\&Sigma;}}}{\log }_2(1+p_k{\left|{\mathrm{\&lambda;}}_{k,1}\right|}^2{\left|V_{k,1}^H{\tilde{V}}_{k,1}\right|}^2)\right\}---\left(9\right)$

p _k=ρ/M,

Can be by similar the obtaining of formula (5).

Fig. 3, Fig. 4 are simulation result figure, and among Fig. 3, base station transmit antennas is counted M=4, receive terminal number N=10, and user's reception antenna is counted L=1; Among Fig. 4, base station transmit antennas is counted M=4, and user's reception antenna is counted L=1, SNR=10.Can be found out that by figure the method that the present invention proposes obviously is superior to existing random beam forming and portion C SI feedback wave beam forming, the present invention has promoted the multi-user system capacity significantly when having less feedback overhead.

Claims (2)

1. orthogonal random beam forming transmission method based on partial channel knowledge feedback; Relate to a transmitting terminal and more than one receiving terminal; Transmitting terminal includes the transmitting antenna more than, and receiving terminal comprises a reception antenna, and this method is carried out according to following steps:

Steps A 1 transmitting terminal is to each receiving terminal emission pilot signal vector;

Steps A 2 a k receiving terminal obtain channel matrix through channel estimating, and it is made singular value decomposition, obtain main left singular vector U _{K, 1}And binary information (V _{K, 1}, λ _{K, 1}), λ _{K, 1}Be main singular value, V _{K, 1}Be λ _{K, 1}Corresponding main right singular vector, k=1,2 ..., N, N are the number of receiving terminal;

Steps A 3 a k receiving terminal are with binary information I (V _{K, 1}, λ _{K, 1}) feed back to transmitting terminal;

Steps A 4 transmitting terminals are selected the maximum M of main singular value from N the binary information that receives, the receiving terminal that it is corresponding is as alternative user;

Steps A 5 transmitting terminals make up the beam vector matrix

: transmitting terminal is formed preconditioning matrix Q with other M-1 alternative users' beyond p the alternative user main singular value _p=[V _1,1..., V _P-1,1, V _P+1,1, V _{M, 1}], ask for preconditioning matrix Q _pOrthonormal basis $q_p=[{\mathrm{\&epsiv;}}_p^1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&epsiv;}}_p^{p-1},{\mathrm{\&epsiv;}}_p^{p+1},\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&epsiv;}}_p^M],$ Again with p alternative user's the right singular value V of master _{P, 1}At orthogonal basis q _pLast projection obtains ${\stackrel{\&OverBar;}{V}}_{p,1}=V_{p,1}-\underset{i=1,i\&NotEqual;p}{\overset{M}{\mathrm{\&Sigma;}}}<{\mathrm{\&epsiv;}}_p^i,V_{p,1}>\&CenterDot;{\mathrm{\&epsiv;}}_p^i,$ Do normalization at last and handle the wave beam that obtains p alternative user ${\tilde{V}}_{p,1}={\stackrel{\&OverBar;}{V}}_{p,1}/\left|\right|{\stackrel{\&OverBar;}{V}}_{p,1}\left|\right|,p=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,M;$

User and corresponding wave beam that steps A 6 transmitting terminals select this secondary data to send from all alternative users and beam vector matrix;

Steps A 7 transmitting terminals are launched data with selected wave beam to selected user;

Steps A 8 is chosen the user and receives data-signal.

2. the orthogonal random beam forming transmission method based on the partial channel knowledge feedback according to claim 1 is characterized in that, e is chosen the user and receives data-signal x _eAfter, also it is obtained the final data signal as Filtering Processing ${\hat{Y}}_e={\mathrm{\&lambda;}}_{e,1}{V}_{e,1}^H{\tilde{V}}_{e,1}{x}_e+U_{e,1}^H{W}_e,$ E=1,2 ..., f, f represent the number of users that is chosen, λ _{E, 1}Represent that e is chosen the main singular value that the user obtains as singular value decomposition in steps A 2, V _{E, 1}Expression λ _{E, 1}Corresponding main right singular vector,

Represent that e is chosen the corresponding wave beam of user, U _{E, 1}Represent that e is chosen the main left singular vector that the user obtains as singular value decomposition in steps A 2, W _eRepresent e reception noise that is chosen the user, H representes the vector transposition.

Images