CN101594323A - A kind of transmission method for precoding of multi-input multi-output system - Google Patents

Abstract

A kind of transmission method for precoding of multi-input multi-output system, comprise: base station (NodeB) does auto-correlation to the local channel response of each multidiameter delay of up link of obtaining, obtain the auto-correlation result of this multidiameter delay, to auto-correlation result's summation of all multidiameter delays, with the correlation matrix of summed result as channel; Described correlation matrix is carried out characteristic value decomposition, obtain N characteristic vector of N maximum characteristic value correspondence, described N is descending transmitted stream way; The N that utilization an is obtained characteristic vector is carried out precoding as the antenna weighting of NodeB to descending transmission data.Use method of the present invention, can improve the efficiency of transmission and the data throughout of signal.

Description

A kind of transmission method for precoding of multi-input multi-output system

Technical field

The present invention relates to send precoding technique, the transmission method for precoding of particularly a kind of multiple-input and multiple-output (MIMO) system.

Background technology

In present multiple-input and multiple-output (MIMO) system, user side and base station (NodeB) all has many antennas to carry out transmitting and receiving of signal, to obtain data multiplex, improves the throughput of system.

Usually, in the macrocell communication environment of TD-SCDMA system, when NodeB carries out the downlink data transmission, NodeB carries out appropriately weighted according to the down channel characteristic to many antennas, utilize weight vectors to carry out precoding to sending data,, then the data transmission behind the wave beam forming is gone out to sending the wave beam forming of data with realization.The weighting of realizing ideal then needs to utilize the complete channel information of down channel.

But, on the one hand, owing to lack abundant training sequence resource, NodeB can not dispose a training sequence on every antenna, can only (equivalent channel after being beam shaping) be 1 training sequence of every circuit-switched data configuration on equivalent launch of data channels at most, the antenna number of the equivalent channel number of user side smaller or equal to user side arrived in concrete base station.Therefore, when NodeB sent data to user side, user side can't obtain the channel that dimension is Nr * Nt, and can only obtain (the channel of equivalent channel number * Nr).Wherein, Nt is the antenna number of NodeB, and Nr is the antenna number of user side.

On the other hand, at user side, its many antennas received signal simultaneously when receiving, but because the restriction of transmitter module, user side can only have an antenna to be used for emission when sending upward signal, and can not carry out the switching emission between different antennae.Because this restriction, NodeB also can not get access to complete channel information by up link, and can only get access to a certain transmit antennas of user side to the channel information between many antennas of NodeB, promptly 1 * Nt channel is called local channel with this information among the application and responds.As seen, though the up-downgoing channel of TD-SCDMA system has reciprocity, the base station all can't obtain complete channel information from up link or down link and carry out wave beam forming.

Processing mode when utilizing at present smart antenna to carry out the signal emission in the macrocell communication environment has three kinds, and adopting 8 antennas, 2 antennas of user side with NodeB is example:

One, NodeB obtains 1 * 8 channel by up link, utilizes 1 * 8 channel that is obtained to carry out wave beam forming, sends the single channel code stream to user side;

Two, NodeB obtains 1 * 8 channel by up link, is divided into two group of 1 * 4 channel, carries out wave beam forming in the group, and multiplexing wave beam sends the two-way code stream between group;

Three, NodeB obtains 1 * 8 channel by up link, is divided into two group of 1 * 4 channel, carries out wave beam forming in the group, and two equivalent wave beams that form is carried out singular value (SVD) decompose, and gets corresponding characteristic vector as antenna weighting.

In above-mentioned three kinds of modes, first kind of mode is owing to only form the single channel code stream, and therefore the data transfer rate of transmission is low, and efficiency of transmission is poor; In the second way and the third mode 1 * 8 channel is divided into two groups and carries out wave beam forming respectively, be about to two groups of antennas and regard incoherent as, but in fact two groups of antennas are correlated with, therefore after carrying out wave beam forming according to this dual mode, user side will receive the data decorrelation than difficulty, thereby make the performance of detection signal reduce, and then influenced data throughout.

Summary of the invention

In view of this, the invention provides the method for precoding in a kind of mimo system, can improve the efficiency of transmission and the data throughout of signal.

For achieving the above object, the present invention adopts following technical scheme:

A kind of transmission method for precoding of multi-input multi-output system comprises:

Base station node B is done auto-correlation to the local channel response of each multidiameter delay of up link of obtaining, obtains the auto-correlation result of this multidiameter delay, to auto-correlation result's summation of all multidiameter delays, with the correlation matrix of summed result as channel;

Described correlation matrix is carried out characteristic value decomposition, and select N characteristic vector, described N is descending transmitted stream way;

Utilize the antenna weighting of N characteristic vector of selection, descending transmission data are carried out precoding as NodeB.

Preferably, the response of the local channel of described each multidiameter delay of up link is: a transmit antennas of user side is to the vectorial g of row of the channel response formation of all antennas of NodeB _Nr' (l)=[ρ _lh _{Nr, 1}(l) ... ρ _lh _{Nr, nt}(l) ... ρ _lh _{Nr, Nt}(l)], wherein, l is the numbering of multidiameter delay, and nr is the numbering of a described transmit antennas in all user side antennas of user side, and nt is the antenna number of NodeB, and 1≤nt≤Nt, and Nt is the antenna sum of NodeB.

Preferably, N characteristic vector of described selection is: select N maximum characteristic value characteristic of correspondence vector separately.

As seen from the above technical solution, among the present invention, NodeB does auto-correlation to the local channel response of each multidiameter delay of up link of obtaining, obtains the auto-correlation result of this multidiameter delay, to auto-correlation result's summation of all multidiameter delays, with the correlation matrix of summed result as channel; Described correlation matrix is carried out characteristic value decomposition, and select N characteristic vector, described N is descending transmitted stream way; The N that utilization an is obtained characteristic vector is carried out precoding as the antenna weighting of NodeB to descending transmission data.In method for precoding of the present invention, can obtain near the correlation matrix under the unity feedback channel, and utilize this correlation matrix to carry out precoding, thereby realize the transmission of multichannel code stream, send mode with respect to the single channel code stream has improved efficiency of transmission on the one hand, on the other hand, the mode with respect to wave beam forming that partial channel knowledge is divided into groups has also improved data throughout.

Description of drawings

Fig. 1 is the particular flow sheet that sends method for precoding in the mimo system of the present invention.

Embodiment

For making purpose of the present invention, technological means and advantage clearer,, the present invention done describing in further detail below in conjunction with accompanying drawing.

Basic thought of the present invention is: will adopt the correlation matrix of local channel to replace the correlation matrix of complete channel to carry out the selection of antenna weight vector, thereby carry out precoding to sending data.

Before introducing the specific embodiment of the present invention, at first analyze the method principle of utilizing complete channel information to carry out precoding at present.

On the whole, when utilizing the complete channel information of down link to carry out precoding, at first obtain instantaneous complete channel information, i.e. the channel matrix H of current down link is again to the correlation matrix H of channel ^HH does characteristic value decomposition, if emission single channel code stream then get eigenvalue of maximum characteristic of correspondence vector as antenna weighting; If emission two-way code stream would get eigenvalue of maximum and time big characteristic value characteristic of correspondence vector as antenna weighting.

As seen, key is instantaneous channel H is asked correlation matrix H in the above-mentioned method for precoding ^HH in case this correlation matrix is known, just can carry out suitable antenna weighting, thereby obtain desirable efficiency of transmission and data throughout.

Make a concrete analysis of the correlation matrix characteristic of determining by complete channel information below.

Suppose that the macrocell communication environment is the cellular system of Nr * Nt, antenna distance is closer, because the base station control tower is all very high, so angle spread is all very little, user's low-speed motion is little to the DOA influence.And because the TDD system, the up-downgoing channel has reciprocity, utilizes up link to estimate that the channel response that obtains can be used as the channel response structure channel matrix of down link.Suppose that Nr is the antenna number of user side, Nt is the antenna number of NodeB, and distinguishable multipath bar number is L, supposes that herein all antennas of user side all can send data, thereby can obtain complete channel information, and promptly the channel matrix representation is:

E{|h wherein _{I, j}| ²}=1, and h _{I, j}=[ρ ₁h _{I, j}(1), ρ ₂h _{I, j}(2) ..., ρ _Lh _{I, j}(L)] ^TMultipath channel response between the i root antenna of expression user side and the j root antenna of NodeB, E{ ρ _lh _{I, j}(l) | ²Satisfy a certain power delay profile.

Below to correlation matrix H ^HThe expression formula of H is carried out abbreviation:

Definition g _Nr(l)=[h _{Nr, 1}(l) ... h _{Nr, Nt}(l)], then:

$H^{H}H=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{\left[\begin{array}{c}{g}_1\left(l\right)\\ .\\ .\\ .\\ g_{\mathrm{Nr}}\left(l\right)\end{array}\right]}^H\×\left[\begin{array}{c}{g}_1\left(l\right)\\ .\\ .\\ .\\ g_{\mathrm{Nr}}\left(l\right)\end{array}\right]$

$=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2\{{g_1}^H\left(l\right)\×g_1\left(l\right)+{g_2}^H\left(l\right)\×g_2\left(l\right)+\·\·\·+{g_{\mathrm{Nr}}}^H\left(l\right)\×g_{\mathrm{Nr}}\left(l\right)\}---\left(1\right)$

$=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_1}^H\left(l\right)\×g_1\left(l\right)+\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_2}^H\left(l\right)\×g_2\left(l\right)+\·\·\·+\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{Nr}}}^H\left(l\right)\×g_{\mathrm{Nr}}\left(l\right)$

Can see H by formula (1) ^HH can be divided into channel relevancy that the different antennae of user side adds up respectively and.

Below formula (1) is asked expectation:

$E\left\{H^{H}H\right\}=E\{\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_1}^H\left(l\right)\×g_1\left(l\right)+\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_2}^H\left(l\right)\×g_2\left(l\right)+\·\·\·+\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{Nr}}}^H\left(l\right)\×g_{\mathrm{Nr}}\left(l\right)\}$

$=E\{\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_1}^H\left(l\right)\×g_1\left(l\right)\}+E\{\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_2}^H\left(l\right)\×g_2\left(l\right)\}+\·\·\·+E\{\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{Nr}}}^H\left(l\right)\×g_{\mathrm{Nr}}\left(l\right)\}---\left(2\right)$

$=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^{2}E\{{g_1}^H\×g_1\}+\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^{2}E\{{g_2}^H\×g_2\}+\·\·\·+\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^{2}E\{{g_{\mathrm{Nr}}}^H\×g_{\mathrm{Nr}}\}$

$=E\{{g_1}^H\×g_1\}+E\{{g_2}^H\×g_2\}+\·\·\·+E\{{g_{\mathrm{Nt}}}^H\×g_{\mathrm{Nt}}\}$

Next prove each E{g in the formula (2) _Nr ^H(l) * g _Nr(l) } all equate.Suppose the mimo system of a Nr * Nt.Owing in the macrocell communication environment, have correlation between the many antennas of NodeB and between many antennas of user side, therefore utilize Kronecker product (Kronecker Product) model modeling channel, promptly $H={R_r}^{\frac{1}{2}}\×Q\×{R_T}^{\frac{H}{2}},$ Wherein ${R_r}^{\frac{1}{2}}\×{R_r}^{\frac{H}{2}}=R_r$ Be the relevant battle array of reception antenna, ${R_T}^{\frac{1}{2}}\×{R_T}^{\frac{H}{2}}=R_T$ Be the relevant battle array of transmitting antenna, it is 1 multiple Gaussian Profile that matrix Q satisfies zero-mean, variance, and E{q _Ij ^*q _Ij}=1, E{q _Ij ^*q _Mn}=0, (i ≠ m, j ≠ n), q _IjBe the capable j column element of i among the matrix Q.

So

For a root user end antenna k ∈ Nr wherein, then:

Wherein $E\left\{h_{k1}^{*}{h}_{k1}\right\}=E\left\{h_{k2}^{*}{h}_{k2}\right\}=\·\·\·=E\left\{h_{\mathrm{kNt}}^{*}{h}_{\mathrm{kNt}}\right\}=1,$

For E{g _k ^H* g _kIn element on any one off-diagonal (m, n) m ≠ n has:

$E\left\{h_{\mathrm{km}}^{*}{h}_{\mathrm{kn}}\right\}=E\left\{{\left(\underset{j=1}{\overset{\mathrm{Nt}}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{Nr}}{\mathrm{\Σ}}}{r}_{\mathrm{rki}}{q}_{\mathrm{ij}}{r}_{\mathrm{tmj}}^{*}\right)}^{*}\underset{j=1}{\overset{\mathrm{Nt}}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{Nr}}{\mathrm{\Σ}}}{r}_{\mathrm{rki}}{q}_{\mathrm{ij}}{r}_{\mathrm{tnj}}^{*}\right\}$

Because matrix Q is for satisfying zero-mean, variance is multiple Gauss's matrix of 1, for each element E{q is arranged all _Ij ^*q _Ij}=1, E{q _Ij ^*q _Pq}=0, (i ≠ p, j ≠ q).Then:

$E\left\{h_{\mathrm{km}}^{*}{h}_{\mathrm{kn}}\right\}=\underset{j=1}{\overset{\mathrm{Nt}}{\mathrm{\Σ}}}\underset{i=1}{\overset{\mathrm{Nr}}{\mathrm{\Σ}}}{r}_{\mathrm{rki}}^{*}{r}_{\mathrm{rki}}{r}_{\mathrm{tmj}}{r}_{\mathrm{tnj}}^{*}=\underset{i=1}{\overset{\mathrm{Nr}}{\mathrm{\Σ}}}{r}_{\mathrm{rki}}^{*}{r}_{\mathrm{rki}}\underset{j=1}{\overset{\mathrm{Nt}}{\mathrm{\Σ}}}{r}_{\mathrm{tmj}}{r}_{\mathrm{tnj}}^{*}=\underset{j=1}{\overset{\mathrm{Nt}}{\mathrm{\Σ}}}{r}_{\mathrm{tmj}}{r}_{\mathrm{tnj}}^{*}---\left(4\right)$

So E{h _Km ^*h _KnBe the amount that has nothing to do with k, i.e. an E{g _k ^H* g _kAlso be and the irrelevant amount of k.

By above-mentioned derivation as seen, from long-time statistical, the resulting channel correlation matrix of each root user antenna all is identical.Next, in the implementation procedure of deriving arbitrarily once at random, the situation of the resulting channel correlation matrix of a root user antenna.

By formula (1) as seen, the resulting channel correlation matrix of a root user antenna nr is:

Wherein power delay profile is subtraction function, i.e. a ρ ₁＞ρ ₂＞...＞ρ _L, formula (5) is got the upper bound and lower bound, and suppose distinguishable multipath L →+∞, then have

$\underset{L\→+\∞}{\mathrm{Lim}}{\left[\begin{array}{ccc}{h}_{\mathrm{nr},1}& \·\·\·& h_{\mathrm{nr},\mathrm{Nt}}\end{array}\right]}^H\×{{\mathrm{\ρ}}_L}^{2}I\×\left[\begin{array}{ccc}{h}_{\mathrm{nr},1}& \·\·\·& h_{\mathrm{nr},\mathrm{Nt}}\end{array}\right]$

$\≤\underset{L\→+\∞}{\mathrm{Lim}}{\left[\begin{array}{ccc}{h}_{\mathrm{nr},1}& \·\·\·& h_{\mathrm{nr},\mathrm{Nt}}\end{array}\right]}^H\×{{\mathrm{\ρ}}_1}^{2}I\×\left[\begin{array}{ccc}{h}_{\mathrm{nr},1}& \·\·\·& h_{\mathrm{nr},\mathrm{Nt}}\end{array}\right]$

Then ${{\mathrm{\ρ}}_L}^2\underset{L\→+\∞}{\mathrm{Lim}}\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{g_{\mathrm{nr}}}^H\left(l\right)\×g_{\mathrm{nr}}\left(l\right)\≤\underset{L\→+\∞}{\mathrm{Lim}}\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{nr}}}^H\left(l\right)\×g_{\mathrm{nr}}\left(l\right)\≤{{\mathrm{\ρ}}_1}^2\underset{L\→+\∞}{\mathrm{Lim}}\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{g_{\mathrm{nr}}}^H\left(l\right)\×g_{\mathrm{nr}}\left(l\right)$

Promptly ${{\mathrm{\ρ}}_L}^2{R}_T\≤\underset{L\→+\∞}{\mathrm{Lim}}\frac{1}{L}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{nr}}}^H\left(l\right)\×g_{\mathrm{nr}}\left(l\right)\≤{{\mathrm{\ρ}}_1}^2{R}_T---\left(6\right)$

By formula (6) as seen, under transient state, the any transmit antennas of user side can count channel correlation matrix at NodeB, and matrix form is all identical in two limiting values of the correlation matrix that any transmit antennas of user side obtains, just coefficient is different, and utilizes the correlation matrix H of channel on NodeB ^HWhen H sends precoding, only need to utilize matrix form, then set according to power requirement for concrete coefficient.Therefore, can think that the channel correlation matrix that every antenna obtains is all identical.

Again as seen according to formula (1), the correlation matrix of unity feedback lower channel be the different user antennas channel correlation matrix of adding up respectively and, and prove by said process, the relevant battle array of the channel that different user antennas is added up respectively is all identical, therefore can get, the channel correlation matrix of any root user antenna statistics is identical with the matrix form of the relevant battle array of complete channel, it is the coefficient difference, the channel correlation matrix that any root user antenna can be added up sends precoding as the correlation matrix of complete channel for NodeB.

In addition, recall the precondition of carrying out aforementioned derivation, be respectively:, utilize the correlated channels model to carry out Channel Modeling the channel response of up link channel response as down link.Under the macrocell communication environment, user moving speed is not very fast, so channel meets the condition that becomes slowly, so first precondition satisfies; All have correlation between user side and NodeB antenna separately, therefore second precondition also satisfies.And in NodeB, utilize smart antenna to send precoding also normally under the macrocell environment usually, therefore when utilizing smart antenna to send precoding, can in NodeB, utilize the correlation matrix of the correlation matrix of any root user antenna statistics, thereby send precoding as the channel of unity feedback.

On the other hand according to formula (6) as can be known, for a root user antenna, the process of collecting multipath is approximate can regard the process of asking channel correlation matrix as, and multipath bar number is many more, the correlation matrix of the approaching more complete channel of the channel correlation matrix of gained.Wherein, the so-called multipath of collecting is exactly under each multidiameter delay, to ask the auto-correlation of a described root user antenna to the vector of the channel impulse response formation of all NodeB antennas, and then the auto-correlation result of all multidiameter delays is sued for peace.

Thus, utilize a root user antenna to ask correlation matrix among the present invention, and its correlation matrix as channel is sent precoding at NodeB.Particularly, Fig. 1 is for sending the particular flow sheet of method for precoding among the present invention.As shown in Figure 1, this method flow comprises:

Step 101, NodeB obtains the local channel response of each multidiameter delay according to the transmission signal of user side.

As previously mentioned, though user side has many antennas, but when signal sends, can only utilize wherein an antenna as transmitting antenna, therefore NodeB only can obtain the channel response of this transmitting antenna to self all reception antenna, and utilize the channel reciprocity of TDD system, with the channel of this channel as down link.The response of local channel that this step is obtained is the channel response of transmitting antenna all reception antennas in the NodeB of user side.

The mode of specifically obtaining channel response is: NodeB utilizes the transmission signal of user side to carry out channel estimating, thereby obtains the channel response g of each multidiameter delay of up link _Nr' (l)=[ρ _lh _{Nr, 1}(l) ... ρ _lh _{Nr, nt}(l) ... ρ _lh _{Nr, Nt}(l)], particularly, this channel response g _Nr' (l) be the row vector that channel impulse response constituted of the transmitting antenna of user side in this instantaneous moment to the every antenna of NodeB.Wherein, l is the numbering of current multidiameter delay, [ρ _lh _{Nr, nt}(l) be the transmitting antenna of user side to the channel impulse response between the nt root antenna of NodeB, nr is the numbering of transmitting antenna in all user side antennas of user side, nt is the antenna number of NodeB, and 1≤nt≤Nt, Nt is the antenna sum of NodeB.According to the definition g in the aforementioned derivation _Nr=[h _{Nr, 1}(l) ... h _{Nr, Nt}(l)] the channel response g of each multidiameter delay among the present invention as can be known, _Nr' (l) with derivation in the g that uses _Nr(l) following relation: g is arranged _Nr' (l)=ρ _lg _Nr(l).Channel estimation methods can utilize existing mode to realize, just repeats no more here.

Step 102, NodeB does auto-correlation to the local channel response of each multidiameter delay of up link of obtaining, obtains the auto-correlation result of this multidiameter delay, to auto-correlation result's summation of all multidiameter delays, with the correlation matrix of summed result as channel.

Identical operations is carried out in local channel response to each multidiameter delay in this step.Here the local channel with arbitrary time delay responds g _Nr' (l) be example explanation specific operation process.

Local channel to arbitrary multidiameter delay responds g earlier _Nr' (l) carry out auto-correlation processing, i.e. ρ _l ²g _Nr ^H(l) * g _Nt(l), obtain the auto-correlation result of this multidiameter delay, the auto-correlation result with all multidiameter delays sues for peace then, promptly $\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{nr}}}^H\left(l\right)\×g_{\mathrm{nr}}\left(l\right),$ The process of just aforementioned collection multipath.The result of multipath will be collected $\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\mathrm{\ρ}}_l^2{g_{\mathrm{nr}}}^H\left(l\right)\×g_{\mathrm{nr}}\left(l\right)$ Correlation matrix as channel.

By aforesaid derivation as can be known, the result who collects multipath in the above described manner can be equivalent to the channel correlation matrix of user side one transmit antennas statistics, also just can replace the correlation matrix of complete channel to send precoding.

Step 103, the correlation matrix that step 102 is obtained carries out characteristic value decomposition, selects characteristic vector according to the transmitted stream way.

Wherein, the process of matrix being carried out characteristic value decomposition is identical with existing implementation, just repeats no more here.After carrying out characteristic value decomposition, select N characteristic vector according to the transmitted stream way, wherein, N is descending transmitted stream way.The concrete mode of characteristic vector of selecting is preferably: N characteristic vector selecting N maximum characteristic value correspondence.For example, if descending transmission two-way code stream is then selected two maximum characteristic values characteristic of correspondence vector respectively.

Wherein, descending transmitted stream way smaller or equal to the reception antenna number of user side, therefore selects characteristic vector also can think to select according to the reception antenna number of user side usually.

Step 104 with the antenna weighting of definite characteristic vector as NodeB, is carried out precoding to descending transmission data.

The process that sends precoding according to characteristic vector in this step can realize according to existing mode, just repeats no more here.

So far, transmission method for precoding flow process of the present invention finishes.

In the present invention, under the macrocell communication environment, feeding back a root user antenna can be in the hope of channel correlation matrix to the channel of base station, do precoding and adopt complete channel information to do the precoding performance to be more or less the same with it, with respect to existing precoding mode, can improve the efficiency of transmission and the data throughout of signal.Therefore, solved the situation that can't obtain complete channel information because the user side hardware condition is limited.Further, even user side can feed back the channel of all antennas to the base station, also must cause the increasing of feedback quantity, and utilize method of the present invention can only feed back the channel of a root user antenna, thereby realize low-feedback and near the transmission precoding of unity feedback performance to NodeB.

Being preferred embodiment of the present invention only below, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (3)

1, a kind of transmission method for precoding of multi-input multi-output system is characterized in that, this method comprises:

Base station node B is done auto-correlation to the local channel response of each multidiameter delay of up link of obtaining, obtains the auto-correlation result of this multidiameter delay, to auto-correlation result's summation of all multidiameter delays, with the correlation matrix of summed result as channel;

Described correlation matrix is carried out characteristic value decomposition, and select N characteristic vector, described N is descending transmitted stream way;

Utilize the antenna weighting of N characteristic vector of selection, descending transmission data are carried out precoding as NodeB.

2, method according to claim 1 is characterized in that, the local channel response of described each multidiameter delay of up link is: a transmit antennas of user side is to the vectorial g of row of the channel response formation of all antennas of NodeB _Nr' (l)=[ρ _lh _{Nr, l}(l) ... ρ _lh _{Nr, nt}(l) ... ρ _lh _{Nr, Nt}(l)], wherein, l is the numbering of multidiameter delay, and nr is the numbering of a described transmit antennas in all user side antennas of user side, and nt is the antenna number of NodeB, and 1≤nt≤Nt, and Nt is the antenna sum of NodeB.

3, method according to claim 1 and 2 is characterized in that, N characteristic vector of described selection is: select N maximum characteristic value characteristic of correspondence vector separately.

Images