CN102546123A - Uplink precoding method and base station - Google Patents

Abstract

The invention provides an uplink precoding method and a base station. The uplink precoding method is applied to a multi-mobile-station multi-input multi-output system and includes that the base station receives reference signals respectively sent by each of at least two mobile stations; aiming at each mobile station, the base station performs channel estimation according to corresponding reference signals to obtain channel response matrixes corresponding to each mobile station; the base station obtains indexes of precoders corresponding to each data stream to be sent of each mobile station, and the indexes of the precoders of the data streams to be sent of at least one mobile station are obtained according to the channel response matrixes of the mobile stations and channel response matrixes of the mobile stations corresponding to other data streams to be sent; the base station sends the indexes to corresponding mobile stations; and the base station performs operation of receiving matrixes, thereby working out useful signals. The uplink precoding method and the base station improve throughput of a system.

Description

A kind of up method for precoding and base station

Technical field

The present invention relates to the wireless mobile communications field, particularly a kind of up method for precoding and base station.

Background technology

Future broadband wireless communication systems requires the increasingly high rate of information throughput and communication quality can be provided.In order on limited frequency spectrum resources, to realize this goal, MIMO (Multi Input Multi Output, multiple-input and multiple-output) technology has become one of requisite means that adopted in the following radio communication.

In mimo system, transmitting terminal utilizes many antennas to carry out the transmission of signal, and receiving terminal utilizes many antennas to carry out the reception of signal.Research shows that than traditional single antenna transmissions method, the MIMO technology can improve channel capacity significantly, and improves the rate of information throughput.

The sending method of employing precoding Precoding can effectively improve the performance of mimo system in many travelling carriages MIMO up-link.The basic thought of precoding is according to current channel information, and data to be sent were carried out preliminary treatment before sending, and comprises linear process and Nonlinear Processing etc.

In the FDD system; All uplink channel state information CSI are supposed to know in base station (BS); Therefore the optimum precoder/decoder of each travelling carriage can be calculated in the base station; And select suitable travelling carriage, the index of the precoder behind selecteed travelling carriage feedback quantization then with top performance standard.

Up method for precoding in many travelling carriages multi-input multi-output system of prior art comprises:

In the steps A 1, at least two travelling carriages each is sent reference signal to the base station separately;

Steps A 2 to each travelling carriage, after the reference signal that send according to travelling carriage the base station is carried out channel estimating, is independently calculated the optimum precoder of this travelling carriage to each travelling carriage, and precoder is quantized;

Steps A 3, the index that the precoder after the base station will quantize is corresponding sends to corresponding travelling carriage, makes travelling carriage from code book, select corresponding pre-coding matrix that data flow to be sent is carried out sending to the base station after the precoding processing according to the index that receives separately.

Yet there is following shortcoming at least in above-mentioned up method for precoding:

Well-knownly be; In many travelling carriages multi-input multi-output system,, inevitably exist between the travelling carriage and disturb because a plurality of travelling carriages carry out the transmission of data flow simultaneously; And in the above-mentioned up method for precoding; Regard each travelling carriage as precoder that travelling carriage is selected in independently existence, it must choose optimum precoder, can cause throughput of system to reduce.

Summary of the invention

The purpose of this invention is to provide a kind of up method for precoding and base station, improve throughput of system.

To achieve these goals, the embodiment of the invention provides a kind of up method for precoding, is used for many travelling carriages multi-input multi-output system, and said up method for precoding comprises:

The base station receives the reference signal that each travelling carriage at least two travelling carriages sends separately;

To each travelling carriage, channel estimating is carried out according to the corresponding reference signal in the base station, obtains the corresponding channel response matrix of each travelling carriage;

The base station obtains the index of the corresponding precoder of each data flow to be sent of each travelling carriage, and wherein the index of the precoder of the data flow to be sent of at least one travelling carriage obtains according to the channel response matrix of the corresponding travelling carriage of the channel response matrix of travelling carriage self and other data flow to be sent;

The base station sends to corresponding travelling carriage with index, makes travelling carriage from code book, select corresponding pre-coding matrix that the data flow to be sent of correspondence is carried out sending to the base station after the precoding processing according to the index that receives separately;

The base station is carried out the computing of receiving matrix again, thereby useful signal is solved.

Above-mentioned up method for precoding, wherein, the index of the precoder that each data flow to be sent of said each travelling carriage of base station acquisition is corresponding specifically comprises:

The base station calculates the optimum t of the corresponding associating of each data flow to be sent ₁..., t _Mr ₁..., r _M, wherein, M is the number of data flow to be sent, t _m, m=1 ..., M is the precoding vectors of m data flow, r _m, m=1 ..., M is the decoded vector of m data flow;

Each vector quantization that the base station will be united in the optimum precoder is a vector in the code book, as follows:

${\hat{t}}_m=\underset{c_i^H,i=1,...2^B}{\arg \max }{\left|\right|t_m{c}_i\left|\right|}^2$

for being kept at the vector in the code book in base station and the travelling carriage in advance, B is the quantization bit of precoding vectors.

Above-mentioned up method for precoding, wherein, said base station utilizes the index of the corresponding precoder of each data flow to be sent of each travelling carriage of method of exhaustion acquisition, specifically comprises:

The base station is confirmed corresponding to the possible precoder combination of all data flow to be sent; Precoder in each precoder combination is corresponding one by one with data flow to be sent; Each vector in the precoder combination is a vector in the code book;

The base station is to each precoder combination calculation performance parameter;

The base station is according to performance parameter result of calculation, and the precoder combination that will have a top performance parameter is confirmed as this and transmitted employed precoder combination;

The index that said base station sends to travelling carriage is to confirm as this to transmit the index of vector in code book in the employed precoder combination;

Said performance parameter comprises power system capacity, throughput of system, the error rate, later power system capacity or the later throughput of weighting of weighting.

Above-mentioned up method for precoding, wherein, the iteration index of the precoder of each data flow correspondence to be sent of each travelling carriage of calculation mode acquisition is one by one passed through in said base station, specifically comprises:

Steps A; The data flow to be sent that is said at least two travelling carriages is selected initial precoder index, and the base station is independently calculated the initial precoder of this data flow to each data flow to be sent at least one data flow to be sent according to user's under this data flow channel estimation results;

Step B utilizes initial precoder and each user's of existing data flow to be sent channel information to come the index of the precoder of the one or more data flow in the data flow to be sent of said at least two travelling carriages of combined calculation.

Above-mentioned up method for precoding, wherein, said base station is carried out the computing of receiving matrix again; Thereby useful signal is solved in the step; The base station is utilized in the precoding vectors index of each data flow that calculates in the feedback procedure, and current channel information, calculates receiving matrix R.

To achieve these goals, the embodiment of the invention provide carry out above-mentioned up method for precoding the base station of pre-coding apparatus, be used for many travelling carriages multi-input multi-output system.

The embodiment of the invention has following beneficial effect:

In the embodiment of the invention; Because when the base station obtains the index of the corresponding precoder of each data flow to be sent of each travelling carriage, wherein the index of the precoder of the data flow to be sent of at least one travelling carriage obtains according to the channel response matrix of the corresponding travelling carriage of the channel response matrix of travelling carriage self and other data flow to be sent; That is to say that the base station does not have not regard each data flow as independently existence and confirms precoding vectors, it has considered interference mutual between the data flow, and therefore, the precoding vectors of selection tallies with the actual situation more, has improved throughput of system.

Description of drawings

Fig. 1 is the schematic flow sheet of the method for first embodiment of the invention;

Fig. 2-Fig. 9 is the simulation result sketch map of the embodiment of the invention.

Embodiment

In the up method for precoding and base station of the embodiment of the invention, be used for the FDD system, it selects each self-corresponding precoder of each travelling carriage through taking all factors into consideration influencing each other between a plurality of travelling carriages, to improve the throughput of system.

Suppose to have in the sub-district base station, K travelling carriage (user) is simultaneously to this base station upstream data.There is N the base station _RIndividual antenna, k travelling carriage has N _kIndividual antenna.K travelling carriage emission L _kIndividual data flow.L _k≤N _k, and

Receiving signal can be written as:

${\hat{s}}_m=r_m^T(\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{H}_i^T{\hat{t}}_i^T{s}_i+n_m)=r_m^T{H}_m^T{\hat{t}}_m^T{s}_m+r_m^T\underset{i\≠m}{\overset{M}{\underset{i=1}{\mathrm{\Σ}}}}{H}_i^T{\hat{t}}_i^T{s}_i+r_m^T{n}_m$

Wherein

M=1 ..., M is emission precoder (precoding vectors) (1 * N of m data flow _kDimension).H _m, m=1 ..., M is the response matrix (N of the channel of affiliated travelling carriage of m data flow and the channel between the base station _k* N _RDimension), wherein comprised transmitting power, large scale decline, and information such as small scale decline.

M=1 ..., M is decoder (the decoded vector) (N of m data flow _R* 1 dimension).s _m, m=1 ..., M is the emission data of m data flow. m=1; ..., M is the soft output of decoding of m data flow.n _m, m=1 ..., M be m data flow on noise.Subscript T is expressed as matrix transpose operation.

The up method for precoding of the embodiment of the invention is used for many travelling carriages multi-input multi-output system, and is as shown in Figure 1, is the schematic flow sheet of the up method for precoding of first embodiment of the invention, and this method comprises:

Step 11, the base station receives the reference signal that each travelling carriage at least two travelling carriages sends separately;

Step 12, to each travelling carriage, channel estimating is carried out according to the corresponding reference signal in the base station, obtains the corresponding channel response matrix of each travelling carriage;

Step 13; The base station obtains the index of the corresponding precoder of each data flow to be sent of each travelling carriage, and wherein the index of the precoder of the data flow to be sent of at least one travelling carriage obtains according to the channel response matrix of the corresponding travelling carriage of the channel response matrix of travelling carriage self and other data flow to be sent;

Step 14, the base station sends to corresponding travelling carriage with index, makes travelling carriage from code book, select corresponding pre-coding matrix that the data flow to be sent of correspondence is carried out sending to the base station after the precoding processing according to the index that receives separately;

Step 15, base station are carried out the computing of receiving matrix again, thereby useful signal is solved.

In specific embodiment of the present invention; Step 13 needs the index of the corresponding precoder (precoding vectors) of the data flow to be sent of each travelling carriage of acquisition; And wherein the acquisition of the index of the corresponding precoder of the data flow to be sent of at least one travelling carriage obtains according to the channel response matrix of the corresponding travelling carriage of the channel response matrix of travelling carriage self and other data flow to be sent; In specific embodiment of the present invention, can realize in several ways, specify as follows.

< implementation one >

In implementation one, step 13 specifically comprises:

Step 131, base station are calculated the optimum precoder t of each data flow associating to be sent ₁..., t _MWith decoder r ₁..., r _M

Wherein, M (more than or equal to 2) is the total quantity of data flow to be sent; t _m, m=1 ..., M is the precoding vectors of m data flow, is 1 * N _kDimension, r _m, m=1 ..., M is the decoder (decoded vector) of m data flow, is N _R* 1 dimension.

Step 132, the base station quantizes the precoder of optimum, obtains the corresponding index of each vector in the optimum precoder, just will unite each the vector t in the optimum precoder _m, m=1 ..., M is quantified as a vector in the code book

M=1 ..., M, as follows:

${\hat{t}}_m=\underset{c_i^H,i=1,...2^B}{\arg \max }{\left|\right|t_m{c}_i\left|\right|}^2$

For being kept at the vector in the code book in base station and the travelling carriage in advance, be N _k* 1 dimension, B is the quantization bit of precoding vectors, subscript H is the operation of asking conjugate transpose.

Can realizing of step 131 through algorithms of different.Is that example is described in detail as follows at this with three kinds of algorithms.

Adopt algorithm A, step 131 comprises:

Steps A 1 is carried out initialization, obtains the corresponding precoder of all data flow to be sent of each travelling carriage

(L _k* N _kThe dimension), initial value, as follows:

$T_k={[I_{L_k},0,...,0]}_{L_k\&times;N_k},(k=1,...,K)$

Wherein

Be the L in the travelling carriage _kThe precoder that individual data flow is corresponding.

Steps A 2 is according to T _kThe computes decoded device

(N _R* L _kDimension), wherein

Be the L in the travelling carriage _kIndividual data flow corresponding decoder.R _kCalculate as follows:

$R_k={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{T}_i^H{T}_i{H}_i+v{I}_{N_R})}^{-1}{H}_k^H{T}_k^H,k=1,...,K$

Wherein, H _kChannel response matrix (N for the channel between travelling carriage k and the base station _k* N _RDimension).ν is a Lagrange multiplier.Subscript H is expressed as the conjugate transpose operation.

Steps A 3 is according to decoder R _kCalculate new T _k, as follows:

$T_k=R_k^H{H}_k^H{(H_k\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\left(R_i{R}_i^H\right)H_k^H+N_0{I}_{N_k})}^{-1},k=1,...,K$

Wherein, N ₀Be noise variance;

Steps A 4, repeating step A2 and A3, the norm of the difference of the precoder that obtains until twice iteration is less than a predefined numerical value.

Yet in the above-mentioned mode, in each iterative step, all need calculate Lagrange multiplier, operand is bigger, and in order to address the above problem, the embodiment of the invention further provides second kind to calculate optimum precoder T of associating and the scheme of decoder R.

Adopt algorithm B, step 131 comprises:

Step B1 carries out initialization to each travelling carriage, obtains the corresponding precoder of all data flow to be sent of each travelling carriage, as follows:

$T_k={[{I_L}_k,0,...,0]}_{L_k\&times;N_k},(k=1,...,K)$

Step B2 calculates the H of each travelling carriage ^-((L-L _k) * N _RDimension), as follows:

$H_k^{-}\stackrel{\mathrm{\&Delta;}}{=}{[H_1^T{T}_1^T,\&CenterDot;\&CenterDot;\&CenterDot;,H_{k-1}^T{T}_{k-1}^T,H_{k+1}^T{T}_{k+1}^T,\&CenterDot;\&CenterDot;\&CenterDot;,H_K^T{T}_K^T]}^T$

Wherein subscript T is the operation of asking transposition;

Step B3 calculates kernel Q (N separately to each travelling carriage _R* L _kDimension), as follows:

$Q_k=\mathrm{null}\left(H_k^{-}\right)$

Wherein null is the operation of asking kernel,

0 is L _k* (L-L _k) the dimension null matrix.

Step B4 carries out singular value decomposition to the HO of each travelling carriage, as follows:

$\mathrm{SVD}\left(H_k{Q}_k\right)=U_k{\mathrm{\&Lambda;}}_k{V}_k^H$

Step B5 utilizes the result of singular value decomposition to calculate corresponding decoder separately to each travelling carriage, as follows:

R _k＝Q _kV _k

Step B6 utilizes the result of singular value decomposition to calculate each self-corresponding T to each travelling carriage _k, as follows:

$T_k=U_k^H$

Step B7, repeating step B2 to B6, set up until following formula:

${\left|\right|H_k^{-}{R}_k\left|\right|}_{\mathrm{<figure-callout\; id="2"\; label="F"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">F</figure-callout>}}^2<{\mathrm{\&epsiv;}}_{\mathrm{th}}$

Precoder that above-mentioned calculating associating is optimum and decoder are for first kind of mode; Its amount of calculation reduces to some extent; But it is still long on iteration time; In order to reduce iteration time, further provide the third to calculate the optimum precoder of associating and the scheme of decoder in the specific embodiment of the invention.

Adopt algorithm C, step 131 comprises:

Step C1 carries out initialization to each travelling carriage, obtains each self-corresponding precoder of each travelling carriage, as follows:

$T_k=I_{N_k}$

Step C2, according to initialized precoder computes decoded device, as follows:

$R_k={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{T}_i^H{T}_i{H}_i+N_0{I}_{N_R})}^{-1}{H}_k^H{T}_k^H$

$={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{H}_i+N_0{I}_{N_R})}^{-1}{H}_k^H$

Wherein, N ₀Be noise variance.

Step C3 carries out singular value decomposition to the HR of each travelling carriage, as follows:

$\mathrm{SVD}\left(H_k{R}_k\right)=U_k{\mathrm{\&Lambda;}}_k{V}_k^H$

Step C4 utilizes the result of singular value decomposition to calculate each self-corresponding T to each travelling carriage _k,

Be T _kBe taken as

In corresponding to maximum L _kThe L of individual characteristic value _kIndividual characteristic vector.

Step C5 is according to initialized T _kCalculate new R _k, as follows:

$R_k={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{T}_i^H{T}_i{H}_i+N_0{I}_{N_k})}^{-1}{H}_k^H{T}_k^H$

Step C6 is to R _kIn each vector carry out normalization and handle, as follows:

$r_m^k=r_m^k/\left|\right|r_m^k\left|\right|,m=1,...L_k$

Step C7 is according to the R after the normalization _kCalculate T _k, as follows:

$T_k=R_k^H{H}_k^H{(H_k\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\left(R_i{R}_i^H\right)H_k^H+N_0{I}_{N_k})}^{-1}$

Step C8, the T that step C7 is obtained _kIn each vector

M=1 ... L _kCarry out renewal operation as follows:

$t_m^k=t_m^k/\left|\right|t_m^k\left|\right|,m=1,...L_k$

Step C9, repeating step C5 be to C8, and the norm of the difference of the precoder that obtains until twice iteration is less than a predefined numerical value.

< implementation two >

First kind of implementation can be introduced quantization error owing to need carry out quantization operation, makes that the actual precoder that uses not is best precoder.The then actual performance that can reach also can't reach the performance that best precoder can reach.To this problem, the present invention has proposed second kind of implementation again.

In the implementation two, the base station utilizes the index of the corresponding precoder of each data flow to be sent of each travelling carriage of method of exhaustion acquisition, specifically comprises:

Step 133, base station are confirmed corresponding to the possible precoder combination of all data flow to be sent; Precoder in each precoder combination is corresponding one by one with data flow to be sent; Each vector in the precoder combination is a vector in the code book;

Step 134, the base station is to each precoder combination calculation performance parameter;

Step 135, base station be according to performance parameter result of calculation, and the precoder combination that will have a top performance parameter is confirmed as this and transmitted employed precoder combination;

The index that said base station sends to travelling carriage is saidly to confirm as this and transmit the index of vector in code book in the employed precoder combination.

Said performance parameter comprises power system capacity, throughput of system, the error rate, or later power system capacity or the later throughput of weighting of weighting.Be that power system capacity is that example is explained as follows below with the performance parameter.

In M data flow, 2 ^BUnder the situation of individual code book, the precoder combination has (2 ^B) M

Kind.For example, work as M=4, during B=4, (2 ^B) ^M=65536.

At first, select a certain encoder combination.Suppose m data flow with the precoding vectors in the code book

emission, then the precoder that adopts of the combination of M data flow is combined as

Carry out following computing to each precoder combination:

Calculate the equivalent channels H that this precoder combination is constructed, as follows:

$H_{i_1,...,i_M}={[H_1^T{c}_{i_1},...,H_M^T{c}_{i_M}]}^T$

H wherein _m, m=1 ..., M is the channel response matrix of the travelling carriage under m the data flow.Certainly, when a travelling carriage has a plurality of data flow to be sent, promptly as m ₁Individual data flow and m ₂(m ₁≠ m ₂) when individual data flow belongs to a travelling carriage, $H_{m_1}={\mathrm{<figure-callout\; id="2"\; label="H\; m"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">H</figure-callout>}}_{m_{}},(m_1=1,...,M,{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">m</figure-callout>}}_2=1,...,M).$

Then;

that utilization calculates is to each data-flow computation Rcv decoder to be sent, as follows:

$r_m={\left[H_{i_1,{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">i</figure-callout>}}_2,i_3,i_4}^H{H}_{i_1,{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">i</figure-callout>}}_2,i_3,i_4}\right]}^{-1}{H}_m^H{c}_{i_m},(m=1,...,M)$

Then, calculate the pairing reception Signal to Interference plus Noise Ratio of each data flow (SINR), as follows:

${\mathrm{SINR}}_m^{i_1,...,i_M}=\frac{1}{{\mathrm{<figure-callout\; id="0"\; label="N"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">N</figure-callout>}}_0}{\left|r_m^T{H}_m^T{c}_{i_m}\right|}^2/(1+\underset{k\&NotEqual;m}{\mathrm{\&Sigma;}}\frac{1}{{\mathrm{<figure-callout\; id="0"\; label="N"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">N</figure-callout>}}_0}{\left|r_m^T{H}_k^T{c}_{i_k}\right|}^2),(m=1,...,M)$

At last, calculate the power system capacity (sum rate) of this precoder combination, as follows:

${\mathrm{sumrate}}_{i_1,...,i_M}=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\log }_2(1+{\mathrm{SINR}}_m^{i_1,...,i_M})$

After the throughput that obtains each precoder combination, select the maximum precoder combination of sumrate, as follows:

$\{{{\hat{t}}_1}^T,...,{\hat{t}}_M^T\}=\underset{c_{i_1},...,c_{i_M},i_1,...,i_M=1,...2^B}{\arg \max }{\left|\right|{\mathrm{sumrate}}_{i_1,...,i_M}\left|\right|}^2$

In second kind of implementation of the present invention; Because the precoder that travelling carriage can be selected is limited; Can only from some precoders that pre-set, select; So directly make up calculate throughput in advance according to possible precoder, select the maximum precoder combination of throughput at last, then precalculated throughput is very approaching with actual throughput.Avoided first kind of quantization error that implementation is brought.

< implementation three >

Though implementation two can reach best performance, make up pairing performance index owing to need to calculate all precoders, operand is bigger.To this problem, the present invention has proposed implementation in the ground three again, and operand is greatly reduced.

In the implementation three, the iteration index of the precoder of each data flow correspondence to be sent of each travelling carriage of calculation mode acquisition is one by one passed through in the base station, specifically comprises:

Step 136; The data flow to be sent that is said at least two travelling carriages is selected initial precoder index, and the base station is independently calculated the initial precoder of this data flow to each data flow to be sent at least one data flow to be sent according to user's under this data flow channel estimation results;

Step 137 utilizes initial precoder and each user's of existing data flow to be sent channel information to come the index of the precoder of the one or more data flow in the data flow to be sent of said at least two travelling carriages of combined calculation.

In implementation three of the present invention, be divided into two kinds of situation, be described below respectively.

Situation one

In situation one, in step 136, the vector that the base station is selected from code book and this subscriber channel correlation is maximum is as initial precoder, as follows:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|H_m^T{c}_i\left|\right|}^2,m=1,...,M$

Then in step 137; Channel with each user projects on the kernel of other user's equivalent channels again; Each user's precoding vectors is elected as and that maximum code vector of the projection correlation of this user on other user's equivalent channels kernels, and is as follows:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left(H\right|h_m)H_m^T{c}_i\left|\right|}^2,m=1,...,M$

H wherein _mBe m user's equivalent channels, be expressed as:

$h_m={\hat{t}}_m{H}_m,m=1,...,M$

Wherein:

m=1; ..., M is precoder or the initial precoder that m data flow obtained;

$H={[h_1^T,...,h_M^T]}^T$

H|h _mFor removing h among the H _mOperation, that is:

$H|h_m={[h_1^T,...,h_{m-1}^T,h_{m+1}^T,..,h_M^T]}^T$

Null is the operation of asking kernel.

Certainly, in specific embodiment of the present invention, be example with data flow to be sent of a travelling carriage, can upgrade according to following mode:

Travelling carriage 1, travelling carriage 2 ..., travelling carriage M

......

Travelling carriage 1, travelling carriage 2 ..., travelling carriage M

Concrete iteration how many times can be decided as required.

Certainly; In specific embodiment of the present invention; Also can be only to travelling carriage 1, travelling carriage 2 ..., the part travelling carriage among the travelling carriage M upgrades, its throughput that can improve is certainly less than the situation of update all, but should be understood that; For the precoder of all travelling carriages of prior art all independently calculates, still can improve throughput.

In situation one; At first each data flow to be sent of each travelling carriage in said at least two travelling carriages is all independently calculated the initial precoder of this travelling carriage; In situation two; Can independently calculate initial precoder to part data flow to be sent, and then utilize existing precoder to come the precoder of the one or more data flow in the data flow to be sent of said at least two travelling carriages of combined calculation.

For situation two, before calculating following formula:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left(H\right|h_m)H_m^T{c}_i\left|\right|}^2,m=1,...,M$

Need in the remaining data flow to be sent of first combined calculation precoder once, as follows:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left({[H_1^T{t}_1^T,...,H_{m-1}^T{t}_{m-1}^T]}^T\right)H_m^T{c}_i\left|\right|}^2,m=1,...M$

The precoder of supposing data flow 1 to be sent independently calculates, and the device that prelists of data flow 2 then to be sent is following:

${\hat{t}}_2^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left({\left[H_1^T{t}_1^T\right]}^T\right)H_2^T{c}_i\left|\right|}^2$

And the device that prelists of data flow 3 to be sent is following:

${\hat{t}}_3^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left({[H_1^T{t}_1^T,H_2^T{t}_2^T]}^T\right)H_3^T{c}_i\left|\right|}^2$

And the device that prelists of data flow 4 to be sent is following:

${\hat{t}}_4^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left({[H_1^T{t}_1^T,H_2^T{t}_2^T,H_3^T{t}_3^T]}^T\right)H_4^T{c}_i\left|\right|}^2$

After the precoder of all data flow to be sent all has an initial value, utilize following formula to upgrade again:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left(H\right|h_m)H_m^T{c}_i\left|\right|}^2,m=1,...,M$

Concrete iterative computation how many times can be decided as required.

In the step 15, in base station end, new decoder R is recomputated receiving travelling carriage selects data flow that corresponding pre-coding matrix carries out sending after the precoding processing to data flow to be sent from code book according to the index that receives after in the base station, explains as follows.

The first step, the combination H. of the equivalent channels of M data flow of structure has two kinds of implementations to construct H, specifies as follows:

Implementation one: travelling carriage is to the reference signal of base station transmission non-precoded, and the base station obtains current channel information through receiving reference signal.The base station is utilized in the precoding vectors index of each data flow that calculates in the feedback procedure, and current channel information, structure H.Promptly

$H={[H_1^T{\hat{t}}_1^T,...,H_M^T{\hat{t}}_M^T]}^T$

m=1 wherein; ..., M is a m data flow corresponding precoding vector.

Implementation two: travelling carriage sends the reference signal after precoding to the base station, and the base station obtains current channel information H ' after precoding through receiving through the reference signal after the precoding _m, m=1 ..., M.The base station utilizes current channel information structure H after precoding.Promptly

$H={[{H^{\&prime;}}_1^T,...,{H^{\&prime;}}_M^T]}^T.$

In second step, the base station utilizes the H of structure, calculates the corresponding decoded vector of m data flow to be:

$r_m={\left[H^{H}H\right]}^{-1}{H}_m^H{t}_m^H,(m=1,...,M),$

Then total BS decoder can be written as R=[r ₁..., r _M].

Utilize decoder, solve each and every one signal.M soft being output as of the decoded signal of data flow:

final total transmission rate (capacity) is

Wherein

${\mathrm{SINR}}_m=\frac{{\left|r_m^T{H}_m^T{t}_m\right|}^2}{({\mathrm{<figure-callout\; id="0"\; label="N"\; filenames="HSA00000400341500021.png,HSA00000400341500022.png"\; state="\{\{state\}\}">N</figure-callout>}}_0+\underset{k\&NotEqual;m}{\mathrm{\&Sigma;}}{\left|r_m^T{H}_m^T{t}_k\right|}^2)},(m=1,...M).$

In order to verify the technique effect of the embodiment of the invention, such scheme is carried out emulation, simulated conditions is following:

System bandwidth 10MHz

Carrier frequency 2GHz

FFT size 1024

Number of subcarriers 600

Subcarrier spacing 15KHz

Resource Block size (subcarrier) 12

Time-delay 1.5ms

Travelling carriage number of transmission antennas 4

Base station reception antenna quantity 4

Code book Rel.10UL SU-MIMO code book

The linear ZF of precoder (compeling zero)

The linear MMSE of decoder

Number of mobile stations 4

Channel typical case Urban6 road piece fading channel

Channel estimation errors 0,10%

Translational speed 3km/h

Wherein, among Fig. 24 * 4MMIO, the travelling carriage number is 4; Adopt the MMSE decoder, translational speed is 3Km/h, and step 13 is utilized the emulation sketch map under the situation of the 3rd kind of implementation in the mode one; From Fig. 2, can find; Along with the increase of SNR, the gap between the upper limit (curve that has circle among the figure) of method of the embodiment of the invention (curve that has star-like mark among the figure) and throughput maintains about 3bps/Hz basically, but for the method for prior art; Increase along with SNR; The difference of the throughput of the throughput of the method for the embodiment of the invention and the method for prior art (having leg-of-mutton curve among the figure) increases gradually, and when SNR was 20dB, the two differed and reaches about 6bps/Hz.

Wherein, among Fig. 34 * 4MMIO, the travelling carriage number is 4; Adopt the MMSE decoder, translational speed is 3Km/h, and step 13 is utilized the emulation sketch map under the situation that the exhaustive search in the mode two realizes; From Fig. 3, can find; Along with the increase of SNR, the gap between the upper limit (curve that has circle among the figure) of method of the embodiment of the invention (curve that has rectangle marked among the figure) and throughput maintains about 2bps/Hz basically, but with respect to the method for prior art; Increase along with SNR; The difference of the throughput of the throughput of the method for the embodiment of the invention and the method for prior art (having leg-of-mutton curve among the figure) is big more, and when SNR was 20dB, the two differed and reaches about 7bps/Hz.

Wherein, among Fig. 44 * 4MMIO, the travelling carriage number is 4; Adopt the MMSE decoder, translational speed is 3Km/h, and step 13 is utilized the emulation sketch map under the situation that first kind of mode in the mode three realize; From Fig. 4, can find; Along with the increase of SNR, the gap between the upper limit (curve that has circle among the figure) of method of the embodiment of the invention (curve that has fork among the figure) and throughput maintains about 2.5bps/Hz basically, but with respect to the method for prior art; Increase along with SNR; The difference of the throughput of the throughput of the method for the embodiment of the invention and the method for prior art (having leg-of-mutton curve among the figure) is big more, and when SNR was 20dB, the two differed and reaches about 6.5bps/Hz.

Wherein, among Fig. 54 * 4MMIO, the travelling carriage number is 4; Adopt the MMSE decoder, translational speed is 3Km/h, and step 13 is utilized the emulation sketch map under the situation that the second way in the mode three realizes; From Fig. 4, can find; Along with the increase of SNR, the gap between the upper limit (curve that has circle among the figure) of method of the embodiment of the invention (curve that has rhombus among the figure) and throughput maintains about 2bps/Hz basically, but with respect to the method for prior art; Increase along with SNR; The difference of the throughput of the throughput of the method for the embodiment of the invention and the method for prior art (having leg-of-mutton curve among the figure) is big more, and when SNR was 20dB, the two differed and reaches about 6.6bps/Hz.

Fig. 6 is corresponding to Fig. 5 with Fig. 2 respectively to Fig. 9; Difference between the two is that Fig. 6 is in the pairing policy condition of Fig. 9; Channel estimation errors is 10%; But the simulation result of Fig. 6-Fig. 9 also shows same result, and the method for the embodiment of the invention can be brought the increase of tangible throughput with respect to the method for prior art.

The base station of the embodiment of the invention is used for many travelling carriages multi-input multi-output system, and said base station comprises the pre-coding apparatus of the up method for precoding that is used to carry out above description, does not describe in detail at this.

The above only is a preferred implementation of the present invention; Should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; Can also make some improvement and retouching, these improvement and retouching also should be regarded as protection scope of the present invention.

Claims (12)

1. a up method for precoding is used for many travelling carriages multi-input multi-output system, it is characterized in that, said up method for precoding comprises:

The base station receives the reference signal that each travelling carriage at least two travelling carriages sends separately;

To each travelling carriage, channel estimating is carried out according to the corresponding reference signal in the base station, obtains the corresponding channel response matrix of each travelling carriage;

The base station obtains the index of the corresponding precoder of each data flow to be sent of each travelling carriage, and wherein the index of the precoder of the data flow to be sent of at least one travelling carriage obtains according to the channel response matrix of the corresponding travelling carriage of the channel response matrix of travelling carriage self and other data flow to be sent;

The base station sends to corresponding travelling carriage with index, makes travelling carriage from code book, select corresponding pre-coding matrix that the data flow to be sent of correspondence is carried out sending to the base station after the precoding processing according to the index that receives separately;

The base station is carried out the computing of receiving matrix again, thereby useful signal is solved.

2. up method for precoding according to claim 1 is characterized in that, the index of the precoder that each data flow to be sent of said each travelling carriage of base station acquisition is corresponding specifically comprises:

The base station calculates the optimum t of the corresponding associating of each data flow to be sent ₁..., t _Mr ₁..., r _M, wherein, M is the number of data flow to be sent, t _m, m=1 ..., M is the precoding vectors of m data flow, r _m, m=1 ..., M is the decoded vector of m data flow;

Each vector quantization that the base station will be united in the optimum precoder is a vector in the code book, as follows:

${\hat{t}}_m=\underset{c_i^H,i=,...2^B}{\arg \max }{\left|\right|t_m{c}_i\left|\right|}^2$

is for be kept at the vector in the code book in base station and the travelling carriage in advance; B is the quantization bit of precoding vectors, and subscript H is the operation of asking conjugate transpose.

3. up method for precoding according to claim 2 is characterized in that, the base station calculates the optimum precoder of the corresponding associating of each data flow to be sent and specifically comprises:

Steps A 1 is carried out initialization, obtains the corresponding precoder of all data flow to be sent of each travelling carriage, as follows:

$T_k={[{I_L}_k,0,...,0]}_{L_k\&times;N_k},k=1,...,K$

Wherein, K is a number of mobile stations, L _kBe the number of data streams of travelling carriage k, N _kNumber of antennas for travelling carriage k;

Be L _k* L _kThe dimension unit matrix

Steps A 2 is according to T _kComputes decoded device R _k, as follows:

$R_k={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{T}_i^H{T}_i{H}_i+v{I}_{N_R})}^{-1}{H}_k^H{T}_k^H$

Wherein, H _kBe the channel response matrix of the channel between travelling carriage k and the base station, N _RBe the number of antennas of base station, K is a number of mobile stations, ν is a Lagrange multiplier;

Steps A 3 is according to decoder R _kCalculate new T _k, as follows:

$T_k=R_k^H{H}_k^H{(H_k\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\left(R_i{R}_i^H\right)H_k^H+N_0{I}_{N_k})}^{-1}$

Wherein, N ₀Be noise variance, I is a unit matrix;

Steps A 4, repeating step A2 and A3, the norm of the difference of the precoder that obtains until twice iteration is less than a predefined numerical value.

4. up method for precoding according to claim 2 is characterized in that, the base station calculates the optimum precoder of the corresponding associating of each data flow to be sent and specifically comprises:

Step B1 carries out initialization to each travelling carriage, obtains the corresponding precoder of all data flow to be sent of each travelling carriage, as follows:

$T_k={[{I_L}_k,0,...,0]}_{L_k\&times;N_k},k=1,...,K$

L _kBe the number of data streams of travelling carriage k, N _kBe the number of antennas of travelling carriage k, K is a number of mobile stations;

Step B2 calculates the H of each travelling carriage ^-, as follows:

$H_k^{-}\stackrel{\mathrm{\&Delta;}}{=}{[H_1^T{T}_1^T,\&CenterDot;\&CenterDot;\&CenterDot;,H_{k-1}^T{T}_{k-1}^T,H_{k+1}^T{T}_{k+1}^T,\&CenterDot;\&CenterDot;\&CenterDot;,H_K^T{T}_K^T]}^T$

Subscript T is the operation of asking transposition;

Step B3 calculates H separately to each travelling carriage ^-Corresponding kernel Q, as follows:

$Q_k=\mathrm{null}\left(H_k^{-}\right)$

Step B4 carries out singular value decomposition to the HQ of each travelling carriage, as follows:

$\mathrm{SVD}\left(H_k{Q}_k\right)=U_k{\mathrm{\&Lambda;}}_k{V}_k^H$

Step B5 utilizes the result of singular value decomposition to calculate corresponding decoder separately to each travelling carriage, as follows:

R _k＝Q _kV _k

Step B6 utilizes the result of singular value decomposition to calculate each self-corresponding T to each travelling carriage _k, as follows:

$T_k=U_k^H$

Step B7, repeating step B2 to B6, set up until following formula:

${\left|\right|H_k^{-}{R}_k\left|\right|}_F^2<{\mathrm{\&epsiv;}}_{\mathrm{th}};$

ε wherein _ThBe a predefined numerical value.

5. up method for precoding according to claim 2 is characterized in that, the base station calculates the optimum precoder of the corresponding associating of each data flow to be sent and specifically comprises:

Step C1 carries out initialization to each travelling carriage, obtains each self-corresponding precoder of each travelling carriage, as follows:

$T_k=I_{N_k}$

Wherein, N _kNumber of antennas for travelling carriage k;

Step C2, according to initialized precoder computes decoded device, as follows:

$R_k={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{T}_i^H{T}_i{H}_i+N_0{I}_{N_R})}^{-1}{H}_k^H{T}_k^H$

$={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{H}_i+N_0{I_N}_R)}^{-1}{H}_k^H$

Wherein, N ₀Be noise variance, H _k, k=1 ..., K is the channel response matrix of the channel between travelling carriage k and the base station, N _RBe the number of antennas of base station, L _kNumber of data streams for travelling carriage k;

Step C3 carries out singular value decomposition to the HR of each travelling carriage, as follows:

$\mathrm{SVD}\left(H_k{R}_k\right)=U_k{\mathrm{\&Lambda;}}_k{V}_k^H$

Step C4 utilizes the result of singular value decomposition to calculate each self-corresponding T to each travelling carriage _k,

$T_k=U_k^H{|}_{1,...,L_k}$

Step C5 is according to initialized T _kCalculate new R _K, as follows:

$R_k={(\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_i^H{T}_i^H{T}_i{H}_i+N_0{I_N}_k)}^{-1}{H}_k^H{T}_k^H$

Step C6 is to R _KIn each vector carry out normalization and handle, as follows:

$r_m^k=r_m^k/\left|\right|r_m^k\left|\right|,m=1,...L_k$

Step C7 is according to the R after the normalization _kCalculate T _k, as follows:

$T_k=R_k^H{H}_k^H{(H_k\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}\left(R_i{R}_i^H\right)H_k^H+N_0{I_N}_k)}^{-1}$

Step C8, the T that step C7 is obtained _kIn each vector

M=1 ... L _kCarry out renewal operation as follows:

$t_m^k=t_m^k/\left|\right|t_m^k\left|\right|,m=1,...L_k$

Step C9, repeating step C5 to C8, the norm of the difference of the precoder that obtains until twice iteration is less than a predefined numerical value.

6. up method for precoding according to claim 1 is characterized in that, said base station utilizes the index of the corresponding precoder of each data flow to be sent of each travelling carriage of method of exhaustion acquisition, specifically comprises:

The base station is confirmed corresponding to the possible precoder combination of all data flow to be sent; Precoder in each precoder combination is corresponding one by one with data flow to be sent; Each vector in the precoder combination is a vector in the code book;

The base station is to each precoder combination calculation performance parameter;

The base station is according to performance parameter result of calculation, and the precoder combination that will have a top performance parameter is confirmed as this and transmitted employed precoder combination;

The index that said base station sends to travelling carriage is to confirm as this to transmit the index of vector in code book in the employed precoder combination;

Said performance parameter comprises power system capacity, throughput of system, the error rate, later power system capacity or the later throughput of weighting of weighting.

7. up method for precoding according to claim 1 is characterized in that, the iteration index of the precoder of each data flow correspondence to be sent of each travelling carriage of calculation mode acquisition is one by one passed through in said base station, specifically comprises:

Steps A; The data flow to be sent that is said at least two travelling carriages is selected initial precoder index, and the base station is independently calculated the initial precoder of this data flow to each data flow to be sent at least one data flow to be sent according to user's under this data flow channel estimation results;

Step B utilizes initial precoder and each user's of existing data flow to be sent channel information to come the index of the precoder of the one or more data flow in the data flow to be sent of said at least two travelling carriages of combined calculation.

8. up method for precoding according to claim 7 is characterized in that, in the steps A, the vector that the base station is selected from code book and this subscriber channel correlation is maximum is as initial precoder, as follows:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|H_m^T{c}_i\left|\right|}^2,m=1,...,M$

Wherein, M is the number of data flow to be sent;

Among the step B, the base station projects to each user's channel on the kernel of other user's equivalent channels, and each user's precoding vectors is elected as and that maximum code vector of the projection correlation of this user on other user's equivalent channels kernels then, as follows:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left(H\right|h_m)H_m^T{c}_i\left|\right|}^2,m=1,...,M$

Wherein:

$H={[h_1^T,...,h_M^T]}^T$

h _mBe m user's equivalent channels, be expressed as:

$h_m={\hat{t}}_m{H}_m,m=1,...,M,$

m=1 wherein; ..., M is precoder or the initial precoder that m data flow obtained;

H|h _mFor removing h among the H _mOperation, that is:

$H|h_m={[h_1^T,...,h_{m-1}^T,h_{m+1}^T,..,h_M^T]}^T;$

Null is the operation of asking kernel.

9. up method for precoding according to claim 8 is characterized in that, in the steps A, the data flow to be sent of each travelling carriage in said at least two travelling carriages is all independently calculated initial precoder.

10. up method for precoding according to claim 8; It is characterized in that; In the steps A; Part data flow to be sent is independently calculated initial precoder, and then utilizes following formula to come to get into behind the initial precoder of the remaining data flow to be sent of combined calculation step B according to existing precoder:

${\hat{t}}_m^T=\underset{c_i,i=1,...2^B}{\arg \max }{\left|\right|\mathrm{null}\left({[H_1^T{t}_1^T,...,H_{m-1}^T{t}_{m-1}^T]}^T\right)H_m^T{c}_i\left|\right|}^2,m=1,...M.$

11. up method for precoding according to claim 1 is characterized in that, said base station is carried out the computing of receiving matrix again, specifically comprises thereby useful signal is solved step:

The reference signal of the non-precoded that the base station receiving mobile sends obtains current channel information, and is utilized in the precoding vectors index of each data flow that calculates in the feedback procedure, and current channel information, structure H, as follows:

$H={[H_1^T{\hat{t}}_1^T,...,H_M^T{\hat{t}}_M^T]}^T$

m=1 wherein; ..., M is a m data flow corresponding precoding vector; The base station utilizes the H of structure to calculate the corresponding decoded vector of m data flow:

$r_m={\left[H^{H}H\right]}^{-1}{H}_m^H{t}_m^H,(m=1,...,M)$

Total base station decodes device is R=[r ₁..., r _M];

The base station utilizes total base station decodes device to solve each signal.

12. a base station is used for many travelling carriages multi-input multi-output system, it is characterized in that, said base station comprises the pre-coding apparatus that is used for any described up method for precoding of enforcement of rights requirement 1-11.

Images