CN101047478A - Low complexity partial feedback zero beam forming method - Google Patents

Abstract

A low-complicacy forming method of partial feedback ZF wave beam includes passing channel state information of user through a decision unit with feedback threshold formed by throughput and user total number as well as emission antenna number first before said information is sent to feedback link, feedback channel state information of user with channel gain greater than feedback threshold back to base station through uplink feedback link and carrying out activation user-set dispatch only in all users participating in feedback by base station.

Description

A kind of partial feedback zero beam forming method of low complex degree

Technical field:

The invention belongs to mobile communication system multi-antenna broadcast channel capacity technical field, particularly relate in the multi-antenna broadcast channel base station and dispatch and pretreated beam forming method according to user's feedback information.

Background technology:

Multi-user diversity is present mobile communication system multi-antenna broadcast channel capacity technical field broad research, the effective ways that improve the wireless system spectrum efficiency.

" international electronics and The Institution of Electrical Engineers's information theory journal " (IEEE Transactions on InformationTheory, vol.49 No.7, July 2003, pp.1691--1706) introduced and a kind ofly many antennas have been arranged at the base station end, user side has in the broadcast channel of single antenna and utilizes dpc techniques to reach the method for downlink broadcast channel and capacity, but because dirty paper code is non-linear, and need know other users' interfere information, make this method complexity too high, and the base station end needs channel condition information completely, and this is difficult to realize in practice.

" international electronics and The Institution of Electrical Engineers's signal processing journal " (IEEE Transactions on Signal Processing, Vol.52 No.2, February 2004, pp.461--471) introduced a kind of close-to zero beam forming technology and have been used for approaching single sub-district multi-antenna broadcast channel capacity.When user side had only an antenna, base station end wave beam preconditioning matrix was set to subscriber channel and reverses, thereby realized not disturbing mutually between the user.But best multi-subscriber dispatching is owing to need search for optimum user's combination in all possible user's combination, computation complexity can be exponential increase along with the increase of number of users.

" international electronics and The Institution of Electrical Engineers's international communication conference proceedings " (IEEE Internet.Conf.Communication ICC ' 05, Rio de Janeiro, Brazil, July 9-18,2005, pp.542--546) introduced a kind of low complex degree the close-to zero beam forming multi-user dispatching method.Utilize the selection of uniting of channel orthogonality between the user and subscriber channel gain, its scheduling complexity is linear growth with the increase of number of users.But this method still has higher complexity when number of users is many.

Summary of the invention:

The present invention proposes a kind of partial feedback zero beam forming method of low complex degree, utilize the selectivity that in the close-to zero beam forming multi-subscriber dispatching subscriber channel is gained, reject the poor user of some channel gains, can significantly reduce feedback overhead and reduce the complexity of multi-subscriber dispatching, and within the scope that the loss of throughput also can the system of being controlled at can be accepted.

The partial feedback zero beam forming method of low complex degree of the present invention, the user carries out channel estimating acquisition channel condition information at training stage pilot transmitted signal and this user in the received signal of training stage according to the base station; The user arrives the base station with channel condition information by uplink feedback; The method that the base station is selected by the semi-orthogonal vector is selected the excited users set; The transmitting power matrix is determined according to the maximum capacity principle and the algorithm of pouring water in the base station; Will launch after will launching data vector and close-to zero beam forming matrix multiple the base station, and described close-to zero beam forming matrix is the product of the evolution of the pseudoinverse of excited users aggregate channel matrix and transmitting power matrix;

It is characterized in that:

User's channel condition information feeds back the feedback decision device that thresholding is determined according to system throughput requirement, total number of users and number of transmit antennas at initial phase jointly by the base station through one earlier before being sent to based on feedback link; Channel gain feeds back to base station with channel condition information by up based on feedback link greater than the user of feedback thresholding; Excited users set scheduling is only carried out in the base station in all users that participate in feeding back.

The described channel estimating of carrying out comprises least mean-square error channel estimating, least square channel estimating or blind Channel Estimation.

Below pass through to analyze the principle of the inventive method institute foundation, and advantage of the present invention is described compared with prior art.

Comprise in the close-to zero beam forming system that the present invention is suitable for that a base station and K user (are numbered U_1, ..., U_K), the base station is equipped with M transmitting antenna, each is user assembled a reception antenna, and K＞M, the input/output relation of this system can be described as: Y=HSX+N (F1)

Each base station selected M user communicates, and this M user has formed the excited users set.Wherein, Y is user's received signal vector of M * 1, $H={\left[\begin{array}{ccc}{h}_{{\mathrm{\π}}_{\left(1\right)}}^T& ...& h_{{\mathrm{\π}}_{\left(M\right)}}^T\end{array}\right]}^T$ Be the excited users collection channel matrix of M * M,

() ^TThe transposition of representing matrix.h _{π (i)}It is 1 * M channel vector of i the user that is activated; S is the close-to zero beam forming matrix of M * M, and X is the user data of M * 1, and N is the additive white Gaussian noise vector that the user of M * 1 receives.

And definition:

I user's channel h _iGain be ‖ h _i‖ ²(F2)

User's average channel gain is g=E (‖ h _i‖ ²) (F3)

I user's normalization channel gain is α _i=‖ h _i‖ ²/ g (F4)

The normalization throughput of system is expressed as R/R ₀(F5)

R ₀Throughput during for the whole user feedback of system, R are the aims of systems throughput.

In traditional multi-user's close-to zero beam forming system, the base station end need be known all users' channel condition information, and promptly each user need be to base station feedback channel status h at each time slot _iThe base station after obtaining all subscriber channel states with optimum (method of exhaustion) or suboptimum (as semi-orthogonal user back-and-forth method) find out with the speed meaning on the excited users group of optimum or suboptimum, and pseudoinverse and the product of transmitting power matrix evolution, i.e. S=H that S is H are set ^T(HH ^T) ^-1P ^1/2, eliminate the interference between the user, will send behind the data X premultiplication S.And the inventive method does not need all feeding back channel state informations of all users, in system initialization, the base station requires, can dispatch total number of users and number of transmit antennas each user's configuration in the sub-district according to the throughput of system and is used to the channel gain thresholding that determines whether the user feeds back, at each time slot, each user carries out channel estimating, have only channel gain greater than the user of the feedback thresholding of configuration just to the base station feedback channel condition information.Along with normalization feedback thresholding increases, the number of users that can access the feedback chance descends rapidly, thereby effectively picks out the poor user of channel gain, reduces feedback overhead to a great extent; On the other hand, because reducing of number of users can be dispatched in the base station, the complexity of scheduling also can reduce.When obtaining above two benefits, the loss of the throughput of system is an acceptable.

In ren beginningization stage of system, the base station is according to system's normalization throughput requirement, number of users and number of transmit antennas, and each user disposes the normalized channel gain thresholding α that determines whether the user feeds back of being used in the sub-district ₀

The user is in order to judge whether oneself need carry out the state that channel estimating obtains needing to obtain channel by pilot signal and this user who utilizes the base station end to send at user side in the received signal of training stage to the base station feedback channel information.The operation of base station is divided into two stages: training stage and transfer phase.In the training stage, the base station does not send data at each antenna on pilot signal transmitted; At transfer phase, the base station only sends the output user data on each antenna, do not send pilot tone.Pilot tone and output user data carry out time division multiplexing by multiplexer.

In the training stage of base station, the base station sends mutually orthogonal pilot tone on each antenna, and each user estimates channel state vectors from M transmitting antenna of base station to this user's reception antenna by channel estimation module:

h _i＝[h _i1…h _iM] ^T (F6)

Wherein i is user's a sequence number.

Then with h _iBe input to the feedback decision device.The feedback decision device is the normalization channel gain of calculating channel at first:

α _i＝‖h _i‖ ²/g (F7)

Compare α _iAnd α ₀If, α _i＞α ₀, this user judges oneself to possess the feedback qualification, with channel h _i=[h _I1H _IM] ^TSend into uplink feedback channel.

If α _i≤ α ₀, this user does not just participate in feedback.

Among the present invention, to participating in the user of feedback, the every user feedback expense that participates in feedback is identical during still with whole user feedback, but because the reducing of feedback user number, total feedback overhead obviously reduces.

The base station end carries out user's selection according to following steps after obtaining all subscriber channel information with feedback qualification:

The first step: ren beginningization.

Suppose that the base station has obtained K altogether ₀Individual user's feedback information is worked as K ₀During 〉=M, set alternative user's set ${\mathrm{\&Gamma;}}_1=({\mathrm{<figure-callout\; id="1"\; label="U"\; filenames="A20061003944000122.png,A20061003944000131.png"\; state="\{\{state\}\}">U</figure-callout>}}_1,\&CenterDot;\&CenterDot;\&CenterDot;U_{K_0}).$ Calculate all K ₀User's normalization channel gain, and therefrom select the user of 1 channel gain maximum, this user is the seed SEED=U of initial user's Active Set _SeedInitialization excited users S set=(U _Seed); The alternative user of initialization gathers Γ ₂=Γ _1-SAuxiliary variable $g_{\left(1\right)}=h_{U_{\mathrm{seed}};}$ Auxiliary variable i=2.

Second step: for each user U _k∈ Γ _i, calculate

$g_k=h_k-h_k\underset{1}{\overset{i-1}{\mathrm{\&Sigma;}}}\frac{{g_{\left(j\right)}}^{*}{g}_{\left(j\right)}}{{\left|\right|g_{\left(j\right)}\left|\right|}^2}...\left(F8\right)$

Symbol () ^*Representing matrix specially put conjugation.

The 3rd step: i the user who selects the excited users set according to following method

${\mathrm{\&pi;}}_{\left(i\right)}=\underset{k\&Element;{\mathrm{\&Gamma;}}_i}{\arg \max }{\left|\right|g_k\left|\right|}^2,...\left(F9\right)$

The excited users collection is updated to

S:＝S∪(π _(i))，

Auxiliary variable g _(i) be updated to

$g_{\left(i\right)}:=g_{{\mathrm{\&pi;}}_{\left(i\right)}}$

If i＜M, auxiliary variable i is updated to: i:=i+1 forwarded for the 4th step to;

If i=M forwarded for the 5th step to.

The 4th step: alternative user gathers according to following Policy Updates:

${\mathrm{\&Gamma;}}_{i+1}:=(k\&Element;{\mathrm{\&Gamma;}}_i,\frac{\left|g_{\left(i\right)}{h}_k\right|}{\left|\right|g_{\left(i\right)}\left|\right|\left|\right|h_k\left|\right|}<\mathrm{\&beta;})...\left(F10\right)$

Wherein β is the system's orthogonality scheduling parameter that obtains by Computer Simulation in advance according to average signal-to-noise ratio, antenna for base station number, total number of users and feedback thresholding.

Repeated for second step and the 3rd step.

The 5th step:

1. the user's of the M among the S channel vector is formed channel matrix

$H={\left[\begin{array}{ccc}{h}_{{\mathrm{\&pi;}}_{\left(1\right)}}^T& ...& h_{{\mathrm{\&pi;}}_{\left(M\right)}}^T\end{array}\right]}^T$

2. calculate the actual channel gain matrix of M user among the S

G＝(HH ^H) ^-1 (F11)

The inverse of its diagonal entry $g_{{\mathrm{\&pi;}}_{\left(j\right)}}=\frac{1}{G_{\mathrm{jj}}}(j=1\&CenterDot;\&CenterDot;\&CenterDot;M)$ It is each user's actual channel gain.

3. according to g _{π (1)}..., g _{π (M)}With total transmitting power P ₀Determine M user's the algorithm power division of pouring water with the principle of maximum throughput:

$P=\mathrm{diag}(P_{{\mathrm{\&pi;}}_{\left(1\right)}},\&CenterDot;\&CenterDot;\&CenterDot;,P_{{\mathrm{\&pi;}}_{\left(M\right)}})...\left(F12\right)$

Symbol diag () expression becomes diagonal matrix with the element set in the bracket.

Wherein $\left\{\begin{array}{c}{P}_{{\mathrm{\&pi;}}_{\left(j\right)}}={({\mathrm{\&mu;g}}_{{\mathrm{\&pi;}}_{\left(j\right)}}-{\mathrm{\&sigma;}}_n^2)}^{+}\\ \underset{j=1\&CenterDot;\&CenterDot;\&CenterDot;M}{\mathrm{\&Sigma;}}{(\mathrm{\&mu;}-\frac{{\mathrm{\&sigma;}}_n^2}{g_{{\mathrm{\&pi;}}_{\left(j\right)}}})}^{+}=P_0\end{array}\right.,$ μ is the level that power is poured water, (z) ⁺=max (z, 0) (F13)

Calculating chases after the zero beam forming matrix $S=H^T{\left({\mathrm{HH}}^T\right)}^{-1}{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="A20061003944000122.png,A20061003944000131.png"\; state="\{\{state\}\}">p</figure-callout>}}^{1/2}...\left(F14\right)$

H concentrates each user's channel vector to obtain by rows by excited users in the formula.M user's emission vector $X={[x_{{\mathrm{\&pi;}}_{\left(1\right)}},\&CenterDot;\&CenterDot;\&CenterDot;,x_{{\mathrm{\&pi;}}_{\left(M\right)}}]}^T.$

X premultiplication close-to zero beam matrix S is obtained dateout.Go out by transmission antennas transmit then.

M the user who activates is numbered π ₍₁₎..., π _(M), user π _(j)Received signal be:

$y_{{\mathrm{\&pi;}}_{\left(j\right)}}=g_{{\mathrm{\&pi;}}_{\left(j\right)}}{x}_{{\mathrm{\&pi;}}_{\left(j\right)}}+n_{{\mathrm{\&pi;}}_{\left(j\right)}},j=1\&CenterDot;\&CenterDot;\&CenterDot;M...\left(F15\right)$

n _{π (j)}Be that variance equals σ _n ²White Gaussian noise.

Thereby K among the present invention ₀The throughput that finally can obtain during＞M is

$R=\underset{1}{\overset{M}{\mathrm{\&Sigma;}}}\log (1+\frac{{p_{\mathrm{\&pi;}}}_{\left(j\right)}}{{\mathrm{\&sigma;}}_n^2})$ Bps/hertz (F16)

K ₀During＜M, the base station can not be dispatched to maximum M user, and base station scheduling stops, and the throughput that scheduling this time obtains is: R=0 (F17)

System can obtain average throughput: R=E (R) (F18)

Symbol E () expression is to time average.

Need all user feedback channel condition informations to compare with existing dirty paper code and optimum and suboptimum multi-subscriber dispatching close-to zero beam forming system (exhaustive with the system of selection of semi-orthogonal Vector Groups), the present invention is owing to adopted the technology of User Part feedback, make and have only channel gain just can be fed, and do not need all users all to participate in feedback greater than the user of feedback thresholding.Along with the increase of feedback thresholding, the number of users of feedback reduces very fast, thereby has significantly saved feedback overhead.

Need in all users, the user of exhaustive throughput maximum compare with the optimum multi-subscriber dispatching of existing close-to zero beam forming, because the total number of users that the present invention has adopted the User Part feedback technique to make the base station to dispatch reduces, even exhaustive optimal scheduling method is used in the base station, complexity also can significantly reduce.

Since in multi-user's close-to zero beam forming system to the selectivity of subscriber channel gain, this throughput that does not need the reduction system that method that whole users feed back can be not clearly.Be that the present invention loses with very little throughput, exchanged the remarkable reduction of feedback overhead and scheduling complexity for.

Description of drawings:

Fig. 1 is the implementation system structured flowchart of the partial feedback zero beam forming method of low complex degree of the present invention.

Fig. 2 is the normalization throughput and the relation curve that feeds back thresholding of the partial feedback zero beam forming method system of low complex degree.

Fig. 3 is the feedback user ratio and the relation curve that feeds back thresholding of the partial feedback zero beam forming method of low complex degree.

Fig. 4 is the throughput and the relation curve that feeds back thresholding of the partial feedback zero beam forming method of low complex degree.

Embodiment:

Embodiment below in conjunction with description of drawings the inventive method.

Embodiment 1:

With 4 antennas of base station configuration, each user disposes 1 antenna in the present embodiment, but has 100 dispatched users altogether in the sub-district, and the normalization throughput requires 0.97, scheduled correlation factor-beta=0.5, total signal to noise ratio P ₀/ σ _n ²=10, each user's channel is that the uncorrelated channel of Rayleigh is an example, and the embodiment of User Part feedback zero beam forming multi-user broadcast system is described.

In this configuration, average channel gain according to (F3) is: g=E (‖ h _i‖ ²)=4.

The base station is according to the relation of 100 users' normalization throughput and feedback thresholding, feedback decision device normalization throughput is set is normalization feedback thresholding α under 0.97 the situation ₀=1.5.

Fig. 1 has provided the implementation system structured flowchart of the partial feedback zero beam forming method of present embodiment low complex degree:

The base station end system is by forming with the lower part: Scheduler module 2 receives all users' data to be sent 1 and from the user state information 13 of up based on feedback link 12, carry out after user's scheduling the user data 3 that is scheduled being sent into beam forming module 4, scheduled user's channel condition information 14 in the up based on feedback link 12 also is admitted to beam forming module 4 simultaneously, beam forming module 4 is carried out transmit power assignment, close-to zero beam forming matrix computations, and the data behind the beam forming that transmitting antenna 5 is sent beam forming module 4 here send.

User terminal system is by forming with the lower part: transmission data or scramble information that reception antenna 6 receives from the base station, and data message is sent into data demodulation module 7 carry out recovering initial data 10 after the data demodulates; Scramble information is sent into channel estimation module 8 carry out after the channel estimating channel condition information being sent into feedback decision device 9, if judgement needs feedback then channel condition information 11 is sent into up based on feedback link 12.

In the training stage of base station, the base station sends mutually orthogonal pilot tone on each antenna, each user adopts the least mean-square error channel estimation methods to estimate channel gain vector h from M transmitting antenna of base station to this user's reception antenna by channel estimation module 8 _i=[h _I1, h _I2, h _I3, h _I4], wherein i is user's a sequence number.The user calculates normalization channel gain: α according to (F4) _i=‖ h _i‖ ²/ g=‖ h _i‖ ²/ 4

And with α _iSend into feedback decision device 9.

Feedback decision device 9 is α relatively _iAnd α ₀(α=1.5) are if α _i＞1.5, with h _i=[h _I1, h _I2, h _I3, h _I4] and user's sign i send into up link 12 and feed back to the base station.If α _i≤ 1.5,9 decisions of feedback decision device this time do not participate in feedback.

The base station obtains carrying out after all subscriber channels feed back following operation by up link:

1. the base station receives the channel status that all participate in the user of feedback in Scheduler module 2, and the sum of statistics feedback user.If the feedback user sum is less than 4, then base station dispatcher module 2 decisions this time do not send any signal.

The throughput that this base station scheduling obtains is R=0;

If the feedback user sum more than or equal to 4, continues to carry out following steps.

2. base station dispatcher module 2 is carried out the excited users selection according to following steps:

The first step: the set of base station dispatcher module 2 all feedback users is set to Γ ₁, calculate the K that all participate in feedback according to (F4) ₀Individual user's normalization channel gain, and therefrom select the user of 1 channel gain maximum, this user is the seed SEED=U of initial user's Active Set _SeedInitialization excited users S set=(U _Eed).The alternative user of initialization gathers Γ ₂, Γ ₂=Γ ₁-S.Auxiliary variable $g_{\left(1\right)}=h_{U_{\mathrm{seed}}}.$ Auxiliary variable i=2.

Second step: for each user U _k∈ Γ _i, calculate

$g_k=h_k-h_k\underset{1}{\overset{i-1}{\mathrm{\&Sigma;}}}\frac{{g_{\left(j\right)}}^{*}{g}_{\left(j\right)}}{{\left|\right|g_{\left(j\right)}\left|\right|}^2}$

The 3rd step: i the user who selects the excited users set according to following method

${\mathrm{\&pi;}}_{\left(i\right)}=\underset{k{\&Element;\mathrm{\&Gamma;}}_i}{\arg \max }{\left|\right|g_k\left|\right|}^2,$

The excited users set is updated to

S:＝S∪(π _(i))，

Auxiliary variable g _(i)Be updated to

$g_{\left(i\right)}:=g_{{\mathrm{\&pi;}}_{\left(i\right)}}$

If i＜4, auxiliary variable i is updated to i:=i+1, forwards for the 4th step to;

If i=4 forwarded for the 5th step to.

The 4th step: according to the alternative user's set of following Policy Updates

${\mathrm{\&Gamma;}}_{i+1}=(k\&Element;{\mathrm{\&Gamma;}}_i,\frac{\left|g_{\left(i\right)}{h}_k\right|}{\left|\right|g_{\left(i\right)}\left|\right|\left|\right|h_k\left|\right|}<0.5),$

Repeated for second step and the 3rd step.

3. module 4 is with user U _{π (1)}..., U _{π (4)}Channel state vector form channel matrix $H={{{[h}_{{\mathrm{\&pi;}}_{\left(1\right)}}}^T\&CenterDot;\&CenterDot;\&CenterDot;{h_{{\mathrm{\&pi;}}_{\left(4\right)}}}^T]}^T.$

4. module 4 is calculated each user's actual channel gain $g_{{\mathrm{\&pi;}}_{\left(j\right)}}=\frac{1}{G_{\mathrm{jj}}},(i=1\&CenterDot;\&CenterDot;\&CenterDot;4).$ G _JjRepresenting matrix G=(HH ^H) ^-1J diagonal entry.Gross power P according to this transmission ₀And g _{π (1)}G _{π (4)}Utilize the transmitting power P of each user in (F13) computing activation user set according to the maximum capacity principle _{π (1)}P _{π (4)}: $\left\{\begin{array}{c}{P}_{{\mathrm{\&pi;}}_{\left(j\right)}}={({\mathrm{\&mu;g}}_{{\mathrm{\&pi;}}_{\left(j\right)}}-{\mathrm{\&sigma;}}_n^2)}^{+}\\ \underset{j=1\&CenterDot;\&CenterDot;\&CenterDot;M}{\mathrm{\&Sigma;}}{(\mathrm{\&mu;}-\frac{{\mathrm{\&sigma;}}_n^2}{g_{{\mathrm{\&pi;}}_{\left(j\right)}}})}^{+}=P_0\end{array}\right.,$ μ is the level that power is poured water, (z) ⁺=max (z, 0).Obtain the power emission matrix $P=\mathrm{diag}(P_{{\mathrm{\&pi;}}_{\left(1\right)}}\&CenterDot;\&CenterDot;\&CenterDot;P_{{\mathrm{\&pi;}}_{\left(4\right)}})$

5. module 4 is calculated close-to zero beam forming matrix S=H ^T(HH ^T) ^-1P ^1/2

6. module 4 determines that concentrated 4 users' of excited users emission data vector is $X={\left[x_{{\mathrm{\&pi;}}_{\left(1\right)}}{\&CenterDot;\&CenterDot;\&CenterDot;x}_{\mathrm{\&pi;}\left(4\right)}\right]}^T,$ And calculate and launch vectorial X '=SX.

7. 4 data among the X ' are launched from 4 antennas of transmitting antenna module 5 respectively.

User π in the excited users set _(j)Receive that by reception antenna module 6 data are:

$y_{{\mathrm{\&pi;}}_{\left(j\right)}}=g_{{\mathrm{\&pi;}}_{\left(j\right)}}{x}_{{\mathrm{\&pi;}}_{\left(j\right)}}+n_{{\mathrm{\&pi;}}_{\left(j\right)}},j=1\&CenterDot;\&CenterDot;\&CenterDot;M,n_{{\mathrm{\&pi;}}_{\left(j\right)}}$ For variance is σ _n ²White Gaussian noise.

User π (j) delivers to demodulation module 7 with these data, can obtain the base station and sends data x the signal accent that is concerned with according to the channel state vectors that obtains by channel estimation module 8 in the training stage _{π (j)}

The throughput that this dispatching patcher obtains is:

$R=\underset{1}{\overset{4}{\mathrm{\&Sigma;}}}\log (1+\frac{{g_{{\mathrm{\&pi;}}_{\left(j\right)}}{p}_{\mathrm{\&pi;}}}_{\left(j\right)}}{{\mathrm{\&sigma;}}_n^2})$ Bps/hertz

Total throughput that system obtains is:

R=E (R)=15.33 bps/hertz, compare whole user feedback channel information situation throughputs and only descended 0.45 bps/hertz.Symbol E () expression is to time average.

When Fig. 2 had provided base station configuration 4 transmit antennas, the relation curve of user's normalization throughput and feedback thresholding: user's channel was assumed that the uncorrelated channel of Rayleigh, and is independent mutually between the user.Abscissa is a normalization feedback thresholding.Ordinate is the normalization throughput of system.Curve A is 200 user's situations, and curve B is 100 user's situations, and curve C is 40 user's situations, and curve D is 20 user's situations.Wherein curve B has been represented the situation of embodiment 1.Curve B is 0.97 o'clock at ordinate, and abscissa is 1.5, i.e. normalization feedback thresholding is arranged at 1.5 o'clock, and the normalization throughput of system is 0.97.Among Fig. 2, be limited in 1 at the normalization feedback door, 20,40,100,200 users' normalization throughput is respectively 0.92,0.97,0.99,0.99.Loss is also not obvious.Along with the increase of feedback thresholding, the normalization throughput can reduce gradually.At corresponding different user number, different system's normalization throughput requirements can be chosen suitable feedback thresholding.

Curve E among Fig. 3 has provided 4 antennas of base station end, the relation curve of feedback user ratio and feedback thresholding, and promptly user's normalization channel gain is greater than the probability of normalization feedback thresholding.User's channel is assumed that the uncorrelated channel of Rayleigh, and is independent mutually between the user.Abscissa is a normalization feedback thresholding, and ordinate is a ratio.Along with the increase of normalization feedback thresholding, the probability that the user is fed reduces rapidly, thereby total feedback user number also can reduce rapidly.The feedback thresholding is set to 1.5 in example 1, and the total number of users of feedback only accounts for 15% of total number of users, and promptly feedback overhead has only 15% of whole user feedbacks, and the scheduling complexity of base station also has only 15% when feeding back all.

Associating Fig. 2 and Fig. 3, in embodiment 1, under 100 users' the situation, the partial feedback zero beam forming method system of low complex degree is to account for the feedback overhead of whole user feedbacks 15%, with the scheduling complexity of whole user feedbacks 15%, realized that whole users participate in feeding back 97% throughput.

When Fig. 4 has provided base station 4 transmit antennas, the relation curve of user's throughput and feedback thresholding.System throughput and normalization throughput are counted under the situation and the relation of feedback thresholding at different user.Abscissa is a normalization feedback thresholding.Ordinate is the throughput of system, and unit is every hertz of a bits per second.Curve F is 200 user's situations, and curve G is 100 user's situations, and curve H is 40 user's situations, and curve I is 20 user's situations.The situation that the relatively whole users of embodiment 1 curve G situation, throughput feed back descends and has only every hertz of 0.45 bits per second, but feedback overhead and scheduling complexity all have only 15% of whole feedback overheads.Associating Fig. 2 and Fig. 4, as long as the threshold setting of normalization feedback is not too big, the loss of system throughput is just also not obvious so; And as can be seen from Figure 3 along with the increase of feedback thresholding, the complexity of feedback overhead and scheduling all is significantly reduced.

Claims (2)

1, a kind of partial feedback zero beam forming method of low complex degree, the user carries out channel estimating acquisition channel condition information at training stage pilot transmitted signal and this user in the received signal of training stage according to the base station; The user arrives the base station with channel condition information by uplink feedback; The method that the base station is selected by the semi-orthogonal vector is selected the excited users set; The transmitting power matrix is determined according to the maximum capacity principle and the algorithm of pouring water in the base station; Will launch after will launching data vector and close-to zero beam forming matrix multiple the base station, and described close-to zero beam forming matrix is the product of the evolution of the pseudoinverse of excited users aggregate channel matrix and transmitting power matrix;

It is characterized in that:

User's channel condition information feeds back the feedback decision device that thresholding is determined according to system throughput requirement, total number of users and number of transmit antennas at initial phase jointly by the base station through one earlier before being sent to based on feedback link; Channel gain feeds back to base station with channel condition information by up based on feedback link greater than the user of feedback thresholding; Excited users set scheduling is only carried out in the base station in all users that participate in feeding back.

2,, be characterised in that the described channel estimating of carrying out comprises least mean-square error channel estimating, least square channel estimating or blind Channel Estimation as the partial feedback zero beam forming method of claim 1 low complex degree.

Images