CN100399721C - Transmission method of orthogonal beam shaping in advance based on sending assistant selection of user's feedbacks - Google Patents

Abstract

The present invention discloses a data transmitting method for wireless networks. A system is firstly initialized and then the system sends signals according to cycles; in current sending cycle, the system selects partial users in the system to carry out feedback according to the feedback threshold gamma of a former sending cycle, and feedback users select sub-channels to increase code words and also feed back the increase of the code words to a sending end through a feedback channel. Users selecting feedback select the vector quantity of quantified pre-beams and feed back the vector quantity of quantified pre-beams together with the increase of code words in the sub-channel to the sending end. The sending end selects N < T > users among all users carrying out feedback according to a standard orthogonal design method to form N < T > signals for the selected users and to form main sending signals for the system. In the current sending cycle, the system is provided with a feedback threshold gamma and carries out broadcasting for all users by utilizing a system broadcasting channel for a next periodic system to select users.

Description

Based on the pre-beam shaping transmission method of quadrature that sends the assisted Selection user feedback

Technical field

The present invention relates to a kind of data transmission method of wireless network, relate in particular to a kind of based on the pre-beam shaping transmission method of quadrature that sends the assisted Selection user feedback.

Background technology

Multiple-input, multiple-output (MIMO) technology can improve the capacity and the availability of frequency spectrum of wireless communication system exponentially under the situation that does not increase bandwidth.Generally believe that Multiple Input Multiple Output all will have actual application in Cellular Networks of future generation, self-organization network, WLAN (wireless local area network) and broadband access network.The nineties is by AT﹠amp; TBell laboratory scholar at first finishes the work of laying a foundation of Multiple Input Multiple Output; After this, scholars such as Teladar, Foshinia, Tarokh have studied capacity, signal processing algorithm and the empty time-code of mimo systems in succession; People such as Wolniansky in 1998 adopt vertically-and (V-BLAST) algorithm has been set up a multiple-input, multiple-output experimental system during dblast; Recently, IS-856, HSDPA, IEEE 802.11n, standards such as 802.16 and 802.20 have formally been included Multiple Input Multiple Output in its standards system or as recommended technology.

Present single user's mimo systems is studied comparative maturity, and a large amount of work launches research over against multi-user MIMO system, and emphasis concentrates on multiple user signals and handles and the transmission method aspect.

The transmission method of mimo systems can be divided into two classes, the first kind is the constant transmissions method, specifically can be divided into space diversity and space division multiplexing again, the tradition space diversity method utilizes the Space Time Coding technology can effectively improve link performance, wherein orthogonal space time packet is the Space Time Coding that full diversity exponent number in the rayleigh fading channel was deciphered simply and can be obtained to a class, but its application is subjected to the restriction of antenna number, and can not provide antenna array gain as other utilizes the diversity technique of channel condition information; The tradition space division multiplexing method utilizes different antennae transmission independent sub-streams can obtain very high spectrum efficiency, than Space Time Coding lower complexity is arranged simultaneously.But unfortunately, traditional space division multiplexing makes its performance be subject to the deterioration of rank defect channel status owing to lack the redundancy in space.This two classes technology has remarkable advantages, but can not utilize channel condition information, and time-varying characteristics that can not adaptive channel have limited the further raising of systematic function.

At multiple-input, multiple-output channel time-varying characteristics, people have introduced adaptive transmission method, and transmitting terminal carries out self adaptation according to channel condition information to multidimensional Radio Resources such as power, bit and spaces and distributes, and can effectively improve the performance of system in fading channel.Adaptive approach has two class concurrent techniques, i.e. adaptive modulation technology and precoding technique.At present, the practicality of the precoding technique of mimo systems and achievement in research all will surpass adaptive modulation technology.Owing to the complexity reason, it is also few to take all factors into consideration this two classes Study on Technology.In contrast, obtained deep research by sending the precoding optimal design that receives Combined Treatment.The precoding optimal design allows system to adjust code word according to user's channel status self adaptation, obtains more performance by the maximization signal interference ratio.In the method for precoding research based on different criterions and linear reception structure, the simplest class technology is exactly that pre-beam shaping of multiple-input, multiple-output channel and received beam are shaped.The pre-beamforming system of multiple-input, multiple-output sends a data flow and uses a plurality of reception antennas by a plurality of transmitting antennas and merges received signal, to obtain good transmission performance.Compare with traditional Space Time Coding, under complete channel state information assumed condition, transmitting terminal can select the optimal beam vector to obtain complete diversity gain and significant antenna array gain.

Because multi-user MIMO system is subject to cochannel and disturbs, the Adaptive Transmission strategy must effectively be managed the interference between the user.Dirty paper code (DPC) disturbs and avoids by the method that subtracts in advance successively of subscriber signal, is the capacity-efficient acquisition strategy of multiple-input, multiple-output broadcast channel.But as many literature research, DPC is the coding and transmission method of a non-causal, and very complicated, impracticable.Multi-user diversity method based on the DPC technology has obtained certain research at present.Multi-user diversity can effectively resist channel fading in multi-user's packet-switched wireless networks, select effectively to approach channel capacity by optimal user, but its user choosing method and signal processing method still have higher complexity.

The capacity potentiality that mimo systems is huge depend on the utilization to the complete channel state, and same, the performance of adaptive transmission method also depends on the situation of knowing to channel status.In other words, in order to guarantee the performance of transmission method, transmitting terminal needs receiving terminal feedback complete channel state information.But the increase of antenna number causes the surge of channel condition information amount, makes the feedback of channel status become the heavy expense of system.Therefore, become a present research focus at the limited feedback method that reduces feedback information volume, its result of study will effectively promote the practicalization of mimo systems.

In multi-user system, reducing feedback information volume has two kinds of thinkings: a kind of is the feedback information volume that reduces each user; Another kind is to select user feedback.Second kind of thinking can well be used in the non-real-time service system in conjunction with the multi-user diversity method.Reduce the feedback of channel information amount and can keep systematic function in acceptable degree simultaneously, become the performance metric of limited feedback method.

Summary of the invention

The invention provides a kind of feedback information volume that can reduce based on the pre-beam shaping transmission method of quadrature that sends the assisted Selection user feedback.

The present invention adopts following technical scheme:

The first step: adopt following steps that system is carried out initialization:

A. determine system user channel correlation matrix R according to channel physical parameter and system parameters, and further determine maximum characteristic sub-channel gain profiles function of subscriber channel state matrix H distribution function and H and average statistical E[h by R ²],

B. count community user sum K, calculate normalization number of users kl=K/N _T, N _TBe number of transmit antennas,

C. initialization system ownership goal feedback quantity Kn is taken as transmitting antenna N _T6 to 10 times,

D. the initial feedback thresholding Γ that the system that provides starts working required makes feedback door be limited to E[h ²],

E., structure is set sends the required pre-beam vectors code book of signal, it is as follows that pre-beam vectors code book method is set:

A) give fixed system because the channel power degeneration factor desired value ζ that the quantification of beam vectors brings,

B) determine code word number N in the pre-beam vectors code book, N is the minimum positive integer that satisfies following two conditions:

1. the guaranteed output degeneration factor makes greater than desired value ζ $\widetilde{\mathrm{\&zeta;}}=\mathrm{\&zeta;}/E\left[h^2\right]$ Be normalized power degeneration factor desired value, promptly $N\&GreaterEqual;\frac{1-{\left(\widetilde{\mathrm{\&zeta;}}\right)}^{1/\mathrm{kl}}}{{((N_T-1)/{4N}_T)}^{N_T-1}(1-(N_T-1)/{4N}_T)},$

2. N is number of transmit antennas N _TIntegral multiple,

C) design Nm group vector codeword, Nm is N and the number of transmit antennas N that determines in the step b) _TThe merchant, every group of N _lIndividual vector codeword is mutually orthogonal, and detailed process is as follows:

1. generate a sample H of channel state matrix, carry out singular value decomposition H=A Λ B ^H, make first group of N _TIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ¹,

2. generate another sample H of channel state matrix, carry out singular value decomposition H=A Λ B ^HIf, B ^HB ¹, B ^HB ²..., B ^HB ^N-1In all element absolute value between 1/Nm and 1-1/Nm,

Then make n group N _lIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ⁿOtherwise, then regenerate sample H,

3. 2. repeating step organizes N up to generating Nm _TIndividual mutually orthogonal pre-this set of code words of beam vectors code word is designated as B ^Nm,

F. put structure and send the required subchannel gains increment code book of signal, the subchannel gains increment is meant that subchannel gains deducts the difference of feedback thresholding here.It is as follows that subchannel gains incremental code this method is set:

A) according to the maximum characteristic sub-channel gain profiles of H function, determine the subchannel gains incremental range, establishing the gain delta scope is 0～Mh ², to all greater than Mh ²Gain delta all to make it be Mh ²,

B) decide gain delta code book number M, M is the minimum positive integer that satisfies following two conditions:

1. satisfy the given system requirements of system,

2. M is that the logarithm at the end is a positive integer with 2,

C) to 0～Mh ²The gain delta scope is cut apart, and detailed process is as follows:

1. to 0～Mh ²The normalization of gain delta scope makes it in 0～1 distribution,

2. 0～1 gain delta value is pressed A and leads compression,

3. the gain delta value after will compressing on average is divided into M interval, is made as Δ={ Δ ¹, Δ ²... Δ ^M,

4. encoded in this M interval every log ₂M binary number constitutes a code word, and corresponding interval obtains gain delta code book C={C ¹, C ²... C ^M,

Second step: system sends signal by the cycle, and the signal configuration method in each transmission cycle is as follows:

A. in the current transmission cycle, system is according to feedback thresholding Γ in the last transmission cycle, and certain customers feed back in the selective system, and detailed process is as follows:

1. user k (k=1 ..., K, K are total number of users in the system) and after receiving terminal is finished the channel status estimation, channel matrix is carried out singular value decomposition $H_k=A_k{\mathrm{\&Sigma;}}_k{B}_k^H,$ Obtain maximum subchannel gains value h _k ²,

2. all maximum subchannel gains values are selected as feedback user more than or equal to the user of feedback thresholding, and its feedback information is to transmitting terminal, and all the other users wait for, and establish feedback user and add up to K ',

B. it is as follows that Fan Kui user's chooser channel gain increment code word, and feed back to transmitting terminal by feedback channel, the word select of subchannel gains incremental code are selected process:

1. user k ' (k '=1 ..., K ') and the maximum subchannel gains value h of calculating _{K '} ²With the difference of feedback thresholding,

2. determine the increment interval of this difference correspondence

3. the code word of selecting the interval corresponding code word of this increment will feed back for the user is made as

C. select the user of feedback to select to quantize pre-beam vectors, and feed back to transmitting terminal by feedback channel together with subchannel gains increment code word, it is as follows to quantize pre-beam vectors selection course:

1. user k ' (k '=1 ..., K ') and according to the channel singular value decomposition $H_{k^{\&prime;}}=A_{k^{\&prime;}}{\mathrm{\&Sigma;}}_{k^{\&prime;}}{B}_{k^{\&prime;}}^H$ Obtain optimum pre-wave beam

Vector b _{K '}, b _{K '}Be B _kThe first row column vector,

2. from pre-beam vectors code book, select to quantize pre-beam vectors code word, promptly

$f^{n_k}=\underset{f^i,i=1,N}{\arg \min }{\left|\right|b_{k^{\&prime;}}{b}_{k^{\&prime;}}^H-f^i{\left(f^i\right)}^H\left|\right|}_{\mathrm{<figure-callout\; id="2"\; label="F"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">F</figure-callout>}}^2,$

D. sending end is selected N according to accurate orthogonal design method in the user that all feed back _lIndividual user, detailed process is as follows:

1. calculate user k ' (k '=1 ..., K ') maximum subchannel gains ${\tilde{h}}_{k^{\&prime;}}^2=\mathrm{\&Gamma;}+{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^{m_k},$

2. calculate user k ' receiving terminal signal to noise ratio ${\mathrm{SNR}}_{k^{\&prime;}}=(P/N_l){\tilde{h}}_{k^{\&prime;}}^2/{\mathrm{\&delta;}}^2,$

3. select first user $\mathrm{use}$ $1=\arg \underset{k^{\&prime;}}{\min }1/{\mathrm{SNR}}_{k^{\&prime;}},$

4. select k " individual user ${\mathrm{usek}}^{\&prime;\&prime;}=\underset{i\&NotEqual;1,,k^{\&prime;\&prime;}-1}{\arg \min }(1/{\mathrm{SNR}}_i+{\mathrm{\&Sigma;}}_{j=1}^{k^{\&prime;\&prime;}-1}{\mathrm{\&eta;}}_{j,i}),$ Here ${\mathrm{\&eta;}}_{j,i}={\left|{\left(f^{n_i}\right)}^H{f}^{n_j}\right|}^2,$

5. repeating step d is up to k "=N _T,

E. construct N _TIndividual selecteed user's signal at first makes user k " (k "=1 ..., N _T) pre-beam vectors be $f_{k^{\&prime;\&prime;}}=\sqrt{P/N_l}{f}^{n_{k^{\&prime;\&prime;}}},$ " the transmission signal is f to obtain user k _{K "}x _{K "}, x here _{K "}Be user's information symbol to be passed,

F. tectonic system always sends signal, always sends signal by the N that directly superposes _TIndividual selecteed user's signal obtains, promptly $\mathrm{Tx}=f_1{x}_1+f_2{x}_2+\&CenterDot;\&CenterDot;\&CenterDot;+f_{N_T}{x}_{N_T},$

G. in the current transmission cycle, system is provided with feedback door limit value Γ, utilizes system broadcast channel to all users broadcastings, selects the user for next periodic system, and feedback door limit value Γ setting up procedure is as follows:

1. calculate the subchannel gains average h of all feedback users in the current period ²,

2. calculating K ' with maximum and the difference of Kn, and calculated difference and peaked merchant be multiply by the adjustment step-length that the subchannel gains average obtains feeding back thresholding with this merchant,

3. next periodic feedback thresholding is that current feedback door limit value deducts the adjustment step-length.

Compared with prior art, the present invention has following advantage:

The present invention control effectively to reduce feedback quantity to feedback user quantity by self adaptation feedback thresholding is set, beam vectors code book by the design quadrature quantizes feedback information, further reduce feedback information volume, utilize vector codebooks to carry out the end user simultaneously and select and direct structuring user's signal.Obey independent same distribution at the subscriber channel state, and under the fast-changing situation, the present invention supports user's short-term fairness and long-term fairness simultaneously.

1. performance can be approached best practice, for example at suburbs honeycomb communication environments:

A) three antenna systems are under the condition of different user number and the average signal-to-noise ratio SNR=20dB of system, along with the increase of number of users, the present invention that when code book N=6 and N=15, obtain with the capacity that obtains the very fast mean of access 1 of capacity.This is because of the increase along with number of users, and vector codeword has reduced quantization error with the optimum pre-beam vectors of increasing probability near certain customers in the vector codebooks.Same reason, the capacity that code book N=6 and N=15 obtain is along with the number of users increase is tending towards approaching.

B) three antenna systems are under the condition of different signal to noise ratios and number of users K=100, because the present invention is provided with self adaptation feedback thresholding, the subscriber channel state had robustness, therefore along with the increase of signal to noise ratio, the capacity that obtains of the present invention and method 1 keeps a very little relatively-stationary difference.

C) four antenna systems are under the condition of different user number and the average signal-to-noise ratio SNR=20dB of system, along with increasing of number of users, and the very fast approach method 1 of the capacity that the present invention obtains.It can also be seen that from figure when code book was set to N=8 and N=16 respectively, its capacity that obtains was more approaching, simultaneously along with the number of users increase is further approaching.Therefore when the number of users of system is very big, can adopt the code book of N=8.

D) four antenna systems are under different signal to noise ratio conditions, and the performance of method 3 and method 1 is very approaching.

2. feedback information volume is little, by user feedback quantity is control effectively, utilizes the wave beam quantification code book of orthogonal design can reduce feedback information volume widely simultaneously.System optimization method (method 1) feedack amount is K * 2 * N _T ²* f (M), wherein f (M) is the channel gain feedback information volume, and the present invention's (method 3) feedback information volume is Kn * log ₂N * log ₂M, Kn is more much smaller than K, the feedback quantity of method 3 ratio method 1 few about 90%.For example four antenna systems are under code book N=8 condition, and the feedback user number was set to 30 o'clock and capacity only is set to minimizing in 100 o'clock less than 1bits/s/Hz than feedback user number, and feedback quantity has reduced by 70%.

3. it is simple to send signal configuration, only is N _TThe direct stack of individual subscriber signal need not the Combined Treatment of user-user signal, can suppress inter-user interference effectively again simultaneously.

4. the present invention comprises two selection user procedures: first process is to utilize the feedback thresholding to select the user of those subchannel gains greater than the feedback thresholding, this process purpose is chooser channel quality user preferably, guarantees user's space diversity gain and array gain performance; Second process utilizes accurate orthogonal design method to select N _TIndividual user, this process purpose is to select to have each other the user of best orthogonal property, has guaranteed the orthogonal property between the user.Because the quadrature performance is a contradiction between space diversity gain and array gain performance and user, thus among the present invention two select user procedures that user diversity is gained and array gain performance and user's quadrature performance are traded off, effectively guaranteed the entire system performance.

5. the feedback thresholding of system broadcasts is according to the channel condition information adaptive change of ownership goal feedback quantity and user feedback among the present invention, compare with the method for determining fixing feedback thresholding according to system channel state long-time statistical characteristic, this method has robustness to channel state variations, has guaranteed the performance of method.

6. the code book method for designing that provides of the present invention by the minimum value of inner product absolute value minimum between the restriction code word, maximizes the chordal distance of interblock minimum as possible, thereby optimizes the performance of code book, and its design process is a kind of process of cyclic search.Simultaneously, the code book design has been considered correlation between channels to be a kind of general method for designing, has nothing to do with the reception antenna number, can be according to concrete channel condition flexible design code book.

Advantage 1 of the present invention and advantage 2 have been considered three antennas and four antenna systems in the honeycomb communication environments of suburbs, but its performance is not limited to above communication environments and three antennas and four antenna mimo systems.The present invention can be extended to the MIMO-OFDM wireless transmitting system easily.

Description of drawings

Fig. 1 is based on the mimo systems structured flowchart of the pre-beam shaping transmission method that sends the assisted Selection user feedback,

Fig. 2 the inventive method flow chart,

Under three antenna system different users of Fig. 3 said conditions and the capacity comparative graph,

Under the different signal to noise ratio conditions of three antenna systems of Fig. 4 and the capacity comparative graph,

Under four antenna system different users of Fig. 5 said conditions and the capacity comparative graph,

Under the different signal to noise ratio conditions of four antenna systems of Fig. 6 and the capacity comparative graph,

The different feedback user numbers of four antenna systems of Fig. 7 to the influence curve figure of capacity.

Embodiment

Embodiment 1

The first step: adopt following steps that system is carried out initialization:

A. determine system user channel correlation matrix R according to channel physical parameter and system parameters, and further determine maximum characteristic sub-channel gain profiles function of subscriber channel state matrix H distribution function and H and average statistical E[h by R ²],

B. count community user sum K, calculate normalization number of users kl=K/N _T, N _TBe number of transmit antennas,

C. initialization system ownership goal feedback quantity Kn is taken as transmitting antenna N _T6 to 10 times,

D. the initial feedback thresholding Γ that the system that provides starts working required makes feedback door be limited to E[h ²],

E., structure is set sends the required pre-beam vectors code book of signal, it is as follows that pre-beam vectors code book method is set:

A) give fixed system because the channel power degeneration factor desired value ζ that the quantification of beam vectors brings,

B) determine code word number N in the pre-beam vectors code book, N is the minimum positive integer that satisfies following two conditions:

1. the guaranteed output degeneration factor makes greater than desired value ζ $\widetilde{\mathrm{\&zeta;}}=\mathrm{\&zeta;}/E\left[h^2\right]$ Be normalized power degeneration factor desired value, promptly $N\&GreaterEqual;\frac{1-{\left(\widetilde{\mathrm{\&zeta;}}\right)}^{1/\mathrm{kl}}}{{((N_T-1)/{4N}_T)}^{N_T-1}(1-(N_T-1)/{4N}_T)},$

2. N is number of transmit antennas N _TIntegral multiple,

C) design Nm group vector codeword, Nm is N and the number of transmit antennas N that determines in the step b) _TThe merchant, every group of N _TIndividual vector codeword is mutually orthogonal, and detailed process is as follows:

1. generate a sample H of channel state matrix, carry out singular value decomposition H=A Λ B ^H, make first group of N _TIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ¹,

2. generate another sample H of channel state matrix, carry out singular value decomposition H=A Λ B ^HIf, B ^HB ¹, B ^HB ²..., B ^HB ^N-1In all element absolute value between 1/Nm and 1-1/Nm, then make n group N _lIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ⁿOtherwise, then regenerate sample H,

3. 2. repeating step organizes N up to generating Nm _TIndividual mutually orthogonal pre-this set of code words of beam vectors code word is designated as B ^Nm,

F. put structure and send the required subchannel gains increment code book of signal, the subchannel gains increment is meant that subchannel gains deducts the difference of feedback thresholding here.It is as follows that subchannel gains incremental code this method is set:

A) according to the maximum characteristic sub-channel gain profiles of H function, determine the subchannel gains incremental range, establishing the gain delta scope is 0～Mh ², to all greater than Mh ²Gain delta all to make it be Mh ²,

B) decide gain delta code book number M, M is the minimum positive integer that satisfies following two conditions:

1. satisfy the given system requirements of system,

2. M is that the logarithm at the end is a positive integer with 2,

C) to 0～Mh ²The gain delta scope is cut apart, and detailed process is as follows:

1. to 0～Mh ²The normalization of gain delta scope makes it in 0～1 distribution,

2. 0～1 gain delta value is pressed A and leads compression,

3. the gain delta value after will compressing on average is divided into M interval, is made as Δ={ Δ ¹, Δ ²... Δ ^M,

4. encoded in this M interval every log ₂M binary number constitutes a code word, and corresponding interval obtains gain delta code book C={C ¹, C ²... C ^M,

Second step: system sends signal by the cycle, and the signal configuration method in each transmission cycle is as follows:

A. in the current transmission cycle, system is according to feedback thresholding Γ in the last transmission cycle, and certain customers feed back in the selective system, and detailed process is as follows:

1. user k (k=1 ..., K, K are total number of users in the system) and after receiving terminal is finished the channel status estimation, channel matrix is carried out singular value decomposition $H_k=A_k{\mathrm{\&Sigma;}}_k{B}_k^H,$ Obtain maximum subchannel gains value h _k ²,

2. all maximum subchannel gains values are selected as feedback user more than or equal to the user of feedback thresholding, and its feedback information is to transmitting terminal, and all the other users wait for, and establish feedback user and add up to K ',

B. it is as follows that Fan Kui user's chooser channel gain increment code word, and feed back to transmitting terminal by feedback channel, the word select of subchannel gains incremental code are selected process:

1. user k ' (k '=1 ..., K ') and the maximum subchannel gains value h of calculating _{K '} ²With the difference of feedback thresholding,

2. determine the increment interval of this difference correspondence

3. the code word of selecting the interval corresponding code word of this increment will feed back for the user is made as

C. select the user of feedback to select to quantize pre-beam vectors, and feed back to transmitting terminal by feedback channel together with subchannel gains increment code word, it is as follows to quantize pre-beam vectors selection course:

1. user k ' (k '=1 ..., K ') and according to the channel singular value decomposition $H_{k^{\&prime;}}=A_{k^{\&prime;}}{\mathrm{\&Sigma;}}_{k^{\&prime;}}{B}_{k^{\&prime;}}^H$ Obtain optimum pre-beam vectors b _{K '}, b _{K '}Be B _kThe first row column vector,

2. from pre-beam vectors code book, select to quantize pre-beam vectors code word, promptly

$f^{n_k}=\underset{f^i,i=1,,N}{\arg \min }{\left|\right|b_{k^{\&prime;}}{b}_{k^{\&prime;}}^H-f^i{\left(f^i\right)}^H\left|\right|}_{\mathrm{<figure-callout\; id="2"\; label="F"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">F</figure-callout>}}^2,$

D. sending end is selected N according to accurate orthogonal design method in the user that all feed back _lIndividual user, detailed process is as follows:

1. calculate user k ' (k '=1 ..., K ') maximum subchannel gains ${\tilde{h}}_{k^{\&prime;}}^2=\mathrm{\&Gamma;}+{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^{m_k},$

2. calculate user k ' receiving terminal signal to noise ratio ${\mathrm{SNR}}_{k^{\&prime;}}=(P/N_l){\tilde{h}}_{k^{\&prime;}}^2/{\mathrm{\&delta;}}^2,$

3. select first user $\mathrm{use}$ $1=\arg \underset{k^{\&prime;}}{\min }1/{\mathrm{SNR}}_{k^{\&prime;}},$

4. select k " individual user ${\mathrm{usek}}^{\&prime;\&prime;}=\underset{i\&NotEqual;1,{,k}^{\&prime;\&prime;}-1}{\arg \min }(1/{\mathrm{SNR}}_i+{\mathrm{\&Sigma;}}_{j=1}^{k^{\&prime;\&prime;}-1}{\mathrm{\&eta;}}_{j,i}),$ Here ${\mathrm{\&eta;}}_{j,i}={\left|{\left(f^{n_i}\right)}^H{f}^{n_j}\right|}^2,$

5. repeating step d is up to k "=N _T,

E. construct N _TIndividual selecteed user's signal at first makes user k " (k "=1 ..., N _T) pre-beam vectors be $f_{k^{\&prime;\&prime;}}=\sqrt{P/N_T}{f}^{n_{k^{\&prime;\&prime;}}},$ " the transmission signal is f to obtain user k _{K "}x _{K "}, x here _{K "}Be user's information symbol to be passed,

F. tectonic system always sends signal, always sends signal by the N that directly superposes _TIndividual selecteed user's signal obtains, promptly $\mathrm{Tx}=f_1{x}_1+f_2{x}_2+\&CenterDot;\&CenterDot;\&CenterDot;+f_{N_T}{x}_{N_T},$

G. in the current transmission cycle, system is provided with feedback door limit value Γ, utilizes system broadcast channel to all users broadcastings, selects the user for next periodic system, and feedback door limit value Γ setting up procedure is as follows:

1. calculate the subchannel gains average h of all feedback users in the current period ²,

2. calculating K ' with maximum and the difference of Kn, and calculated difference and peaked merchant be multiply by the adjustment step-length that the subchannel gains average obtains feeding back thresholding with this merchant,

3. next periodic feedback thresholding is that current feedback door limit value deducts the adjustment step-length.

Embodiment 2

The present invention is on the basis of multi-user diversity and pre-beam shaping, a kind of user choosing method of simplification is proposed, avoided the federated user signal processing, select user feedback by sending auxiliary strategy simultaneously, effectively reduce the system feedback amount of information, the present invention has provided the beam vectors design and the quantization strategy of multi-user system, supports effective transmission of transmitting terminal signal by Limited Feedback

The present invention is directed to following mimo systems broadcast channel model: there is N the base station _TIndividual transmitting antenna, there is the user of K independent distribution in system, and each user has N _kIndividual reception antenna, the system signal model is:

$y=\left[\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">y</figure-callout>}}_1\\ .\\ .\\ .\\ y_K\end{array}\right]=\left[\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="H"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">H</figure-callout>}}_1\\ .\\ .\\ .\\ H_K\end{array}\right][{\mathrm{<figure-callout\; id="1"\; label="F"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">F</figure-callout>}}_1,\&CenterDot;\&CenterDot;\&CenterDot;{,F}_K]\left[\begin{array}{c}{x}_1\\ .\\ .\\ .\\ x_K\end{array}\right]+\left[\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">n</figure-callout>}}_1\\ .\\ .\\ .\\ n_K\end{array}\right]---\left(1\right)$

Wherein, Be user k received signal.

Be the channel matrix of base station to user k,

Be information sequence, satisfy x _kx _k ^H～N (0, I),

Be pre-coding matrix, make Q _k=F _kF _k ^HFor sending covariance matrix, satisfy total power constraint ${\mathrm{\&Sigma;}}_{k=1}^K\mathrm{Tr}\left(Q_k\right)\&le;P,$ P is the transmitting terminal gross power.

For receiving terminal independent same distribution, the multiple Gaussian noise vector of circulation symmetry, satisfy n _kn _k ^H～N (0, δ ²I).

The present invention at first finds the solution capacity territory under multiple access channel and the power constraint condition and user's transmission covariance matrix by the iteration water-filling method, because this can provide user's selection strategy of broadcast channel optimum.

Make the optimization transmission correlation matrix set of user in the multiple access channel be { ∑ _k} _K=1 ^K, its solution procedure can be expressed as following protruding optimization problem:

$\mathrm{<figure-callout\; id="2"\; label="max\; imize\; log"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">max</figure-callout>}\mathrm{imize}{\log }_{}{}_2|I+{\mathrm{\&delta;}}^{-2}\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}_k^H{\mathrm{\&Sigma;}}_k{H}_k|---\left(2\right)$

s.t.Tr(∑ _k)≤P _k， $\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{P}_k\&le;P,$ ∑ _k≥0，k＝1，…，K

Here P _kPower for user k distribution.Order $N_k={\mathrm{\&delta;}}^{2}I+\underset{j=1,j\&NotEqual;k}{\overset{K}{\mathrm{\&Sigma;}}}{H}_j^H{\mathrm{\&Sigma;}}_j{H}_j$ Be the noise plus interference correlation matrix of user i, $W_k=H_k^H{N}_k^{-1}{H}_k$ For the equivalent channel correlation matrix of user k, to W _kCarrying out the SVD decomposition obtains $W_k=U_k{\mathrm{\&Delta;}}_k{U}_k^H,$ Note ${\mathrm{\&Delta;}}_k=\mathrm{diag}({\tilde{h}}_{k,1}^2,\&CenterDot;\&CenterDot;\&CenterDot;,{\tilde{h}}_{k,N_k}^2),$ Then optimum ∑ _kExpression formula is:

${\mathrm{\&Sigma;}}_k=U_k{\widetilde{\mathrm{\&Delta;}}}_k{U}_k^H---\left(3\right)$

Here ${\widetilde{\mathrm{\&Delta;}}}_k=\mathrm{diag}({(\mathrm{\&lambda;}-{\tilde{h}}_{k,1}^{-2})}^{+},\&CenterDot;\&CenterDot;\&CenterDot;{(\mathrm{\&lambda;}-{\tilde{h}}_{k,N_k}^{-2})}^{+}),$ (x) ⁺=max (x, 0), λ is a fill level, satisfies total power constraint.Further, the power that distributes of user k is P _k=Tr (∑ _k).Said process is carried out to user K successively by user 1, carries out final optimization pass by iteration repeatedly then, obtains optimal solution { P _k} _K=1 ^K, at last according to ∑ _K=1 ^KP _kAdjust λ with the magnitude relationship of P.Provide multiple access channel and capacity (equaling broadcast channel and capacity) computational methods thus.

Method 1

1. initialization.A given suitable λ initial value and ρ initial value.

From user 1 to user K, suppose that other User Status is constant, obtain the active user by (3) formula and send correlation matrix.

3. repeating step 2, up to { ∑ _k} _K=1 ^KConverge to fixed value.

4. by { ∑ _k} _K=1 ^KCalculate { P _k} _K=1 ^KIf, ${\mathrm{\&Sigma;}}_{k=1}^K{P}_k>P,$ λ=λ-ρ then.

5. repeating step 2 arrives step 4, up to ${\mathrm{\&Sigma;}}_{k=1}^K{P}_k\&le;P,$ Finally separated { ∑ _k} _K=1 ^K

By duality and simple closed expression, can be by { ∑ _k} _K=1 ^KDirectly obtain { Q _k} _K=1 ^KHere the present invention does not provide { Q _k} _K=1 ^KExpression formula because the conclusion that the present invention needs can be by to { ∑ _k} _K=1 ^KAnalysis directly draws, and 2 conclusions are arranged: the one, and most ∑ _k(corresponding Q _k) be null matrix; The 2nd, remain non-null matrix diagonal element except first element, all be zero basically.

Method 1 is based on space division multiplexing method, promptly

The user can transmit N at every turn _kIndividual symbol.Above-mentioned conclusion explanation, in actual multi-user MIMO system, subscriber channel generally presents stronger correlation, therefore can not support N _kThe transmission of individual symbol.Claim the degree of freedom that the each transmissible symbolic number of user k is the user, be designated as L _kN _TIndividual transmitting antenna can provide N _TAccording to the specific method distribution that is at war with, user k maximum can obtain N between the user for individual spatial degrees of freedom, these degrees of freedom _kThe individual degree of freedom, i.e. L _k=N _kIf each user only takies one degree of freedom, i.e. L _k=1, then system can insert N by interference-free at most _TIndividual user.For method 1, when number of users K with respect to N _TWhen very big, its essence is to participate in the competition with optimum subchannel separately between the user, and its result normally each user only takies one degree of freedom, and the result of this moment method 1 is consistent with the beam-forming method result.

In addition, method 1 hypothesis transmitting terminal and receiving terminal obtain the complete channel state information, angle from power control (multi-user's power division), the close-loop power controlling method that method 1 provides adopts water-filling method can obtain the optimal power allocation result, but consequently only in the optimum subchannel of certain customers, distribute power, and near the mean allocation result.So method 1 can be modified to only at the best N of all users _TMean allocation power in the individual subchannel is then to this N _TIndividual subchannel carries out associating precoding optimal design, the performance of the inevitable approach method 1 of its performance.

Even but only to N _TIndividual sub-channel allocation power, the method for correction still need the complete channel state information to carry out iteration precoding design, and complexity is still very high.Further analyze as can be known, iteration precoding design essence is to suppress inter-user interference, and when transmitting terminal does not have the ideal communication channel state information, then inter-user interference can not get effective inhibition.

The present invention proposes a kind of accurate orthogonal transmission method based on pre-beam shaping thus, this method utilizes (standard) orthogonal property natural between the user to carry out the subscriber signal design, and independently carry out based on each user, do not need co-design, both effectively suppress inter-user interference, reduced the method complexity again.

The precoding vector that claims user k

Be beamforming vectors, provide f below _kDesign, this is an important content of the present invention and feature.Since each user only distributes one degree of freedom, then can only be distributed on the best subchannel, at first investigate each user's subchannel.To H _kCarrying out the SVD branch solves $H_k=A_k{\mathrm{\&Lambda;}}_k{B}_k^H,$ Make h _k=maxdiag (∑ _k), h _kBe B _kThe first row column vector, the equivalent channel matrix of definition user k is h _kh _k

The accurate orthogonal design method that the present invention proposes is characterized in that, selects N by ad hoc approach _TGroup has the h of (standard) quadrature mutually _k, directly make f _k=h _k, k=1 ... N _TConsider the retrievable data rate of user k

$R_k={\mathrm{<figure-callout\; id="2"\; label="log"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">log</figure-callout>}}_2|I+Z_k^{-1}{H}_k{f}_k{f}_k^H{H}_k^H|---\left(4\right)$

Here $Z_k={\mathrm{\&Sigma;}}_{i\&NotEqual;k}{H}_k{f}_i{f}_i^H{H}_k^H+{\mathrm{\&delta;}}^{2}I.$ Consider the linear structure (this point has also guaranteed the complexity that the present invention is lower) that receives, establishing the balanced matrix of user k receiving terminal is g _k, make g _kBe A _k ^HThe first every trade vector, then (4) formula can be rewritten as

$R_k={\log }_2(1+\frac{h_k^2{P}_{\mathrm{avg}}}{{\mathrm{\&delta;}}^2+{\mathrm{\&Sigma;}}_{i\&NotEqual;k}{\mathrm{\&eta;}}_{k,i}{h}_k^2{P}_{\mathrm{avg}}}),$ k＝1，…N _T (5)

P wherein _Avg=P/N _T ${\mathrm{\&eta;}}_{k,i}={\left|h_k^H{h}_i\right|}^2,$ Be the inner product squared absolute value of two beam vectors, can characterize its degree of correlation.The user selects to be expressed as following optimization problem

$\max {\mathrm{\&Sigma;}}_{k=1}^{N_i}{R}_k---\left(6\right)$

s.t.k∈S

Wherein S for user collection 1,2 ... arbitrary N of K} _TThe dimension subclass.Definition user k received signal to noise ratio is ${\mathrm{SNR}}_k=h_k^2{P}_{\mathrm{avg}}/{\mathrm{\&delta;}}^2,$ R then _kBe g _k=1/SNR _k+ ∑ _{I ≠ k}, η _{K, i}Function, can be by g _kCharacterize.(6) formula is not protruding optimization problem, is difficult to obtain closed solutions.The present invention proposes a kind of user choosing method, can obtain the approximate solution of (6) formula.

Method 2

Step 1 is selected first user, $\mathrm{<figure-callout\; id="1"\; label="use"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">use</figure-callout>}1=\mathrm{<figure-callout\; id="1"\; label="arg\; min\; i"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">arg</figure-callout>}\underset{i}{\min }\underset{}{}1/{\mathrm{SNR}}_i.$

Step 2 is selected user k, $\mathrm{usek}=\underset{i\&NotEqual;1,}{\arg }\underset{k-1}{\min }(1/{\mathrm{SNR}}_i+{\mathrm{\&Sigma;}}_{j=1}^{k-1}{\mathrm{\&eta;}}_{j,i}).$

Step 3 repeating step 2 is up to finishing N _TThe user selects.

In user's selection course, there is the contradiction of orthogonality and diversity gain between subscriber signal, guarantee that promptly the subscriber signal orthogonality can reduce diversity gain, and can detract orthogonality between subscriber signal of the diversity gain that guarantees the user.The step 2 of the method 2 this contradiction of having traded off has effectively been obtained whole performance.

In above-mentioned analytic process, all be based on the hypothesis that transmitting terminal is known channel condition information.Method 1 is based on complete channel state hypothesis, and 2 of methods are based on partial channel-state hypothesis (beam vectors and maximum subchannel gains).Therefore in order to realize said method, system needs all users to the transmitting terminal feeding back channel state information.Consider the feedback Ji pin and the cost of real system, generally do not support to feed back the complete channel state, method 2 is simple than method 1, not only is method itself, is that also its required feedack is less, can reduce the feedback overhead of system.

When but the number of users K that works as system was very big, even each user feedback fraction channel condition information, system also can't bear.At this moment need further to reduce feedback information volume.Have two kinds to solve thinking: the one, feedback information is quantized; The 2nd, select user feedback, promptly only allow certain customers' feeding back channel state information.

Consider that the feedback information that method 2 needs concentrates on pre-beam vectors and the maximum subchannel gains, the present invention will quantize it.A kind of selection user feedback method of invention on the basis that quantizes can further reduce the system feedback expense simultaneously.

Provide quantization method below and based on the selection user feedback method that quantizes.

Because one of feature of the present invention is that each user uses same code book, so only consider the quantification feedback problem of single user's mimo systems beam-forming method.

It is exactly to determine a vector codebooks that beam vectors quantizes

And vector codeword wherein is numbered, numbering is known and consistent to transmitting terminal and receiving terminal.Receiving terminal only needs in vector codebooks, selects one near the code word of actual beam vector, and feeds back its numbering, because the required bit number of numbering is far smaller than beam vectors itself, so can reduce feedback information volume.

Supposing the system has an error-free feedback link, and then the system signal mode shown in (1) formula is rewritten as

y＝Hfx+n (7)

H is carried out the SVD branch solve H=A Λ B ^H, establish from vector codebooks

Middle selection beam vectors is f, the beam vectors of feedback $\sqrt{P}f=\sqrt{P}(b+\tilde{b}),$ B is the first row column vector of B,

Be quantization error vector, balanced matrix is that g is A ^HThe first every trade vector, h=maxdiag (Λ), then

$\hat{x}=\mathrm{gy}=\sqrt{P}\mathrm{hx}+\sqrt{P}h{b}^H\tilde{b}x+\mathrm{gn}---\left(8\right)$

Consider that n is independent Gauss's vector again, the unit complex vector does not change it to its rotation and distributes, and therefore the interference variance of being brought by quantization error is

$E{\left|\right|\sqrt{P}h{b}^H\tilde{b}x\left|\right|}^2={\mathrm{Ph}}^{2}E\left[\mathrm{Tr}\left(b^H\tilde{b}x{x}^H{\tilde{b}}^{H}b\right)\right]={\mathrm{Ph}}^{2}E{\left|\right|\tilde{b}\left|\right|}^2---\left(9\right)$

By (9) formula as seen, the error variance that wave beam quantizes to bring depends on the Euclid norm of quantization error vector, and defining the channel power degeneration factor equally is ‖ Hb ‖ _l ²-‖ Hf ‖ _H ², it satisfies

${\left|\right|\mathrm{Hb}\left|\right|}_F^2-{\left|\right|\mathrm{Hf}\left|\right|}_{\mathrm{<figure-callout\; id="2"\; label="F"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">F</figure-callout>}}^2\&le;{\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">h</figure-callout>}}^2{\left|\right|{\mathrm{bb}}^H-{\mathrm{ff}}^H\left|\right|}_F^2---\left(10\right)$

Here, ‖ ‖ _FExpression Frobenius norm.Because

With || bb ^H-ff ^H‖ _F ²Have same monotonicity, so interference variance also can be by ‖ bb ^H-ff ^H‖ _l ²Characterize.Obtain the present invention thus in vector codebooks

The middle criterion of selecting vector

$f=\underset{f^i,i=1,,N}{\arg \min }{\left|\right|{\mathrm{bb}}^H-f^i{\left(f^i\right)}^H\left|\right|}_l^2---\left(11\right)$

For providing the method for designing of vector codebooks, at first need to analyze the Distribution Statistics situation of channel state matrix H.If subscriber channel matrix H element is an independent same distribution gaussian variable again, B is for the right singular matrix (unitary matrice) in the multiple gaussian random matrix, to arbitrarily

B and HB have same distribution.If f is the unit column vector of B, for arbitrarily

F and Hf have same distribution.

F belongs to N _TThe dimension complex vector space Can open by f

An one-dimensional subspace S in space _f, belong to by all

The subspace S of the vector f in space ' open _{F '}Set is called multiple Grassmannian space.The chordal distance of definition Grassmannian space two sub spaces is $d(S_{f^i},S_{f^j})=\sqrt{1-{\left|{\left(f^i\right)}^H{f}^j\right|}^2}.$ Owing to be with limited beam vectors

Make up all subspaces in Grassmannian space, therefore need provide a new definition.Definition is by f ⁱThe bag of expansion is

Here $D=\underset{1\&le;i,j\&le;N}{\mathrm{<figure-callout\; id="1"\; label="min"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">min</figure-callout>}}d(S_{f^i},S_{f^j}),$ $D\&le;\sqrt{\frac{(N_l-1)N}{N_T(N-1)}}.$ The bag density in definition Grassmannian space is Wherein Ψ () is the normalization tolerance on the Grassmannian space, by

Ha Er tolerance is introduced on the space.Work as N _TWhen big

(10) formula upper bound can approximate expression be simultaneously

(12) formula is the monotone decreasing function of D, and therefore under the situation that N fixes, minimizing (12) formula border can realize by maximization D.Can very simply under the situation of given channel power degeneration factor desired value ζ, determine the N value equally by (12) formula.At first calculate maximum subchannel gains average statistical E[h ²], obtain normalized power degenerate system numerical value $\widetilde{\mathrm{\&zeta;}}=\mathrm{\&zeta;}/E\left[h^2\right],$ (12) formula of utilization obtains:

$N\&GreaterEqual;\frac{1-\widetilde{\mathrm{\&zeta;}}}{{((N_T-1)/{4N}_T)}^{N_T-1}(1-(N_T-1)/{4N}_T)}---\left(13\right)$

The vector quantization method of tradition maximization D is not considered the orthogonality between channel relevancy and vector, and being all users, core concept of the present invention uses same code book, final user selects to depend on the orthogonality between the user, and this orthogonality directly comes from the orthogonality between vector in the code book.Simultaneously, consider the multiple-input, multiple-output correlation between channels, establishing channel correlation matrix is R, and then the code book design should be considered R simultaneously.

(13) formula is single user's code word number constraint formula, and in multi-user system, when number of users K was big, code word number N can reduce in the code book, and promptly N can reduce along with the increase of average user quantity K in the system.Make kl=K/N _lBe system's normalization number of users, then

Obtain thus:

$N\&GreaterEqual;\frac{1-{\left(\widetilde{\mathrm{\&zeta;}}\right)}^{1/\mathrm{kl}}}{{((N_T-1)/{4N}_T)}^{N_T-1}(1-(N_T-1)/{4N}_T)}---\left(14\right)$

The present invention provides a kind of new code book method for designing thus, and step is as follows

1. determine current channel correlation matrix R, and determine distribution function and maximum subchannel gains average statistical E[h by the channel state matrix H of R sign ²].

2. determine code word number N according to given channel power degeneration factor desired value ζ and number of transmit antennas, N is the minimum positive integer that satisfies following two conditions:

A) vector codeword number N satisfies (14) formula

B) vector codeword number N is N _TIntegral multiple, be made as Nm=N/N _T

3. design Nm organizes vector codeword, every group of N _TIndividual vector codeword is mutually orthogonal, under this precondition, and maximization D.

A) a sample H of generation channel state matrix carries out singular value decomposition H=A Λ B ^H, make first group of N _TIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ¹

B) another sample H of generation channel state matrix carries out singular value decomposition H=A Λ B ^HIf, B ^HB ¹, B ^HB ²..., B ^HB ^N-1In all element absolute value between 1/Nm and 1-1/Nm, then make n group N _lIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ⁿOtherwise, then regenerate sample H, re-execute b).

C) repeating step b), up to generating Nm group N _TIndividual mutually orthogonal pre-beam vectors code word, this set of code words is designated as B ^Nm

In the above-mentioned code book method for designing, owing to guaranteed that vector codeword inner product absolute value minimum value is not less than 1/Nm, so maximize D indirectly.

In order further to reduce the feedback quantity of multi-user system, can select user feedback.A kind of method just is based on the long-time statistical characteristic of system channel, determines a fixing decision threshold Γ, when user's channel gain during greater than thresholding, and feedback information.The shortcoming that this method exists is the number of users (too much or very few) of effectively Control and Feedback, has limited the performance of method.

The present invention provides a kind of selection user feedback method based on the transmitting terminal aid decision thus, its basic thought is: by the feedback information calculating channel gain statistical conditions of transmitting terminal according to all feedback users last time, as average etc., based on this, in conjunction with given object feedback number of users Kn, determine decision threshold Γ, and carry out system broadcasts that the user adjudicates according to this thresholding and whether feeds back.This method is adjusted decision threshold by self adaptation, and therefore effective Control and Feedback number of users has guaranteed the performance of method.

Self adaptation is adjusted decision threshold Γ and is meant that the feedback thresholding constantly changes according to performance requirement and feedback user quantity desired value.If user feedback channel gain average statistical is h ², then feed back thresholding Γ and depend on h ²With feedback user number of targets Kn.The design procedure of feedback thresholding Γ is:

1. calculate maximum and the difference of current actual feedback number of users Kf and feedback user number of targets Kn, and calculated difference and peaked merchant, this merchant be multiply by subchannel gains average h ²Obtain feeding back the adjustment step delta Γ of thresholding, promptly

$\mathrm{\&Delta;\&Gamma;}={\stackrel{\&OverBar;}{h}}^2\frac{\mathrm{Kn}-\mathrm{Kf}}{\max (\mathrm{Kf},\mathrm{Kn})}.$

2. then feed back thresholding Γ and be adjusted into Γ=Γ-Δ Γ.

In addition, reach receivable performance requirement, the present invention is based on the transmission householder method and propose a kind of new subchannel gains quantization method in order to quantize progression with lower M state.Its thinking is: the user is limited to basic point with the given feedback door of system, adopts the method for similar delta modulation that the subchannel gains increment is quantized and to quantizing state encoding.It is as follows that subchannel gains incremental code this method is set:

1. according to the maximum characteristic sub-channel gain profiles of H function, determine the subchannel gains incremental range, establishing the gain delta scope is 0～Mh ², to all greater than Mh ²Gain delta all to make it be Mh ²

2. determine gain delta code book number M, M is the minimum positive integer that satisfies following two conditions

A) satisfy the given system requirements of system.

B) M is that the logarithm at the end is a positive integer with 2.

3. to 0～Mh ²The gain delta scope is cut apart, and detailed process is as follows

A) to 0～Mh ²The normalization of gain delta scope makes it in 0～1 distribution.

B) 0～1 gain delta value is pressed A and lead compression.

C) the gain delta value after will compressing on average be divided into M interval, be made as Δ={ Δ ¹, Δ ²... Δ ^M.

D) encoded in this M interval every log ₂M binary number constitutes a code word, and corresponding interval obtains gain delta code book C={C ¹, C ²... C ^M.

User k subchannel gains is quantified as following a kind of mapping relations (is basic point with thresholding Γ):

$h_k^2-\mathrm{\&Gamma;}:\&RightArrow;{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^m:\&RightArrow;C^m---\left(15\right)$

Expression h _k ²-Γ difference drops on Δ ^mQuantized interval, its quantization encoding then, promptly quantization code word is a Δ ^mCorresponding code word C ^mIncrement state progression M can determine according to performance requirement.

Provide the pre-beam shaping transmission method of selection user feedback below based on the transmitting terminal aid decision.

Method 3

Step 1 transmitting terminal is broadcasted current decision threshold information.

The user that step 2 system selects maximum subchannel gains to be not less than the feedback thresholding feeds back.

The selecteed user of step 3 carries out the beam vectors code word according to (11) formula and selects, and carries out the gain delta code word according to (15) formula and selects, and feed back to transmitting terminal.

Step 4 order ${\mathrm{\&eta;}}_{k,i}={\left|{\left(f^{n_k}\right)}^H{f}^{n_i}\right|}^2,$

For user k feeds back pre-beam vectors, transmitting terminal further carries out user's selection according to method 2.

Step 5 directly makes the user's who selects in the step 4 precoding be $f_k=\sqrt{P/N_T}{f}^{n_k},$ K=1 ..., N _T, system sends signal $\mathrm{Tx}=f_1{x}_1+f_2{x}_2+\&CenterDot;\&CenterDot;\&CenterDot;+f_{N_T}{x}_{N_T},$ X wherein _kInformation symbol for user k.

Step 6 is calculated the feedback thresholding in next transmission cycle, promptly calculates the subchannel gains average h of all feedback users in the current period earlier ², calculate maximum and the difference of actual feedback number of users and Kn then, and calculated difference and peaked merchant, this merchant be multiply by the adjustment step-length that the subchannel gains average obtains feeding back thresholding.Next periodic feedback thresholding is that current feedback door limit value deducts the adjustment step-length.

Embodiment 3

The multi-user MIMO system transmission method structured flowchart that the present invention proposes as shown in Figure 1.At first, system determines the feedback thresholding of system according to the information and the ownership goal feedback quantity of user feedback, and utilizes system broadcast channel to all users broadcastings; Whether user's receiving terminal according to the comparative result of its maximum subchannel gains value and feedback thresholding, determines feedback information after finishing channel status and estimating; Selecteed user then selects optimum pre-beam vectors from vector codebooks, together with maximum subchannel gains increment quantized value, feed back to transmitting terminal by dedicated feedback channel; Then, transmitting terminal according to accurate orthogonal design method, carries out user's selection again in the user that all feed back; The subscriber signal that last transmitting terminal will finally be selected is synthetic, gives each user by many antenna transmission.

Consider typical suburbs honeycomb communication environments.Suppose that there be Lp scattering bunch in channel, there is identical time delay in the path of sending with the cluster scattering object, and different bunches channel matrix is irrelevant, and establishing l bunch of scattering object has identical average angle of arrival θ _l, the actual angle of arrival is ${\mathrm{\&theta;}}_l={\stackrel{\&OverBar;}{\mathrm{\&theta;}}}_l+{\widehat{\mathrm{\&theta;}}}_l,{\widehat{\mathrm{\&theta;}}}_l~N(0,{\mathrm{\&delta;}}_l^2),$ δ _lBe called the low-angle expansion, path gain is g _l ²The θ of all bunches _lStatistical variance Θ be called wide-angle expansion.The antenna for base station relative spacing is Δ d=d/ λ _w, λ _wBe carrier wavelength, d is the antenna physical separation, and then correlation matrix is defined as between reception antenna: ${\left[R_l\right]}_{a,b}=g_l^2\&CenterDot;e^{-j2\mathrm{\&pi;}(b-a)\mathrm{\&Delta;}d\cos \left({\stackrel{\&OverBar;}{\mathrm{\&theta;}}}_l\right)}\&CenterDot;e^{{-(1/2)\left(2\mathrm{\&pi;}(b-a)\mathrm{\&Delta;}d\sin \left({\stackrel{\&OverBar;}{\mathrm{\&theta;}}}_l\right){\mathrm{\&delta;}}_l\right)}^2,}$ A, b=1 ..., N _T, channel correlation matrix is $R=\underset{l=0}{\overset{\mathrm{Lp}-1}{\mathrm{\&Sigma;}}}{R}_l,$ Channel matrix is $H=\underset{l=0}{\overset{\mathrm{Lp}-1}{\mathrm{\&Sigma;}}}{R_l}^{1/2}{H}_w,$ $H_w{H}_w^H=I.$ Define system average signal-to-noise ratio SNR=P/ δ ²

The inventive method flow process describes the execution in step in the flow chart as shown in Figure 2 below:

The first step: initialization.

1. determine pre-beam vectors code book.

A) estimate sub-district average user sum K, for example K=100.

B) channel parameter: Lp=3, δ _l=5 °, Θ=180, Δ d=2.Calculate maximum characteristic sub-channel gain average statistical E[h ²], E[h for example ²]=3.5.

C) setting power degeneration factor desired value ζ calculates normalized power degeneration factor desired value ζ, for example ζ=0.1.

For three antenna N _T=3, according to (14) formula, obtain N=6, vector codebooks following (2 groups of orthogonal vectors, maximum inner product absolute value=0.5):

In order to improve systematic function, can make N=15, vector codebooks following (5 groups of orthogonal vectors, maximum inner product absolute value=0.8142):

$\left\{\begin{array}{ccccccc}05772& -03048& -09067& 04632& 05195& -07307& 09007\\ {-02602-07644}_1& {-05614-02422}_1& {-01953+03493}_1& 06183-{05450}_1& 05907+{01167}_1& {-01298-05404}_1& {-01313+01728}_1\\ {01217+00075}_1& {-02040-07011}_1& {01283+00351}_1& {02524-02061}_1& {-06025+00685}_1& {-01845-03511}_1& 00765-03684_1\end{array}\right.$

$\left.\begin{array}{cccccccc}-03305& -0.5014& -07917& -03646& -03095& -02620& 07308& -03216\\ {01756-04668}_1& {04700-02583}_1& {00809-00233}_1& -{01517-01272}_1& {01709+07415}_1& {04545-06201}_1& -0069{4+01682}_1& {07549+02459}_1\\ {-06829-04191}_1& -{06605+01572}_1& {-06044+00301}_1& {05623+07153}_1& {04236-03817}_1& 04174+04075_1& {-06131+02385}_1& {05146+00364}_1\end{array}\right\}$

For four antenna N _T=4, according to (14) formula, obtain N=16, vector codebooks following (4 groups of orthogonal vectors, maximum inner product absolute value=0.7630):

$\left\{\begin{array}{cccccccc}-00086& -04842& 05093& -06467& 06318& 03595& 01910& 02896\\ {-01126+04923}_1& {-01395+05353}_1& {-02962+01331}_1& {06764+00828}_1& {-00195-06027}_1& {-05502-01734}_1& {00535+08512}_1& {02964+02878}_1\\ {05551-03246}_1& {03116+03732}_1& {02629-06912}_1& {01303+01217}_1& {-00109-02779}_1& {04852+02642}_1& {02979+02964}_1& {-2256+02483}_1\\ {-05757+00057}_1& {-04293+01969}_1& {-00640-02903}_1& {-02802+00842}_1& {-03761-01358}_1& {03089+03706}_1& {-00315+02418}_1& {03520+07153}_1\end{array}\right.$

$\left.\begin{array}{cccccccc}03135& -03516& -08314& -04877& -07088& 07160& 01140& 05114\\ {03302+01154}_1& {-02907-04790}_1& {-01466+02437}_1& -01648{-01385}_1& {01300-04921}_1& {00391+02139}_1& {01641-02435}_1& {05330-01889}_1\\ {00242+07097}_1& {03020-05726}_1& {02680-03431}_1& {-07955-00550}_1& {-00057+00701}_1& {01154-01614}_1& {02813+00887}_1& {-04673-00379}_1\\ {-03435+03963}_1& {-03785+00137}_1& {-01681-01003}_1& {02784-00496}_1& {-04802+00542}_1& {-06313-00462}_1& {-08878+01602}_1& {-02863-03420}_1\end{array}\right\}$

In order further to reduce feedback information volume, can make N=8, vector codebooks following (2 groups of orthogonal vectors, maximum inner product absolute value=0.5):

$\left\{\begin{array}{cccccccc}-00086& 05093& 06318& 01910& 03135& -08314& -07088& 01140\\ -01126+{04923}_1& {-02962+01331}_1& {-00195-06027}_1& {00535+08512}_1& {03302+01154}_1& {-01466+02437}_1& {01300-04921}_1& {01641-02435}_1\\ {05551-03246}_1& {02692-06912}_1& {-00109-02779}_1& {02979+02964}_1& {00242+07097}_1& {02680-03431}_1& {-00057+00701}_1& {02813+00887}_1\\ {-05757+00057}_1& {-00640-02903}_1& {-03761-01358}_1& {-00315+02418}_1& {-03435+0.3963}_1& {-01681-01003}_1& {-04802+00542}_1& {-08878+01602}_1\end{array}\right\}$

2. determine that maximum subchannel gains increment quantizes progression (M).With N _T=4, M=8 is an example, and channel parameter is the same.

$\stackrel{\&OverBar;}{\mathrm{\&Delta;}}=\left\{\begin{array}{cccccccc}{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^1& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^2& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^3& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^4& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^5& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^6& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^7& {\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^8\\ 000& 001& 010& 011& 100& 101& 110& 111\\ 0.1& 0.2& 0.3& 0.4& 0.6& 1& 2& 3\end{array}\right\}$

C＝{000?001?010?011?100?101?110?111}

3. it is the maximum subchannel gains average statistical of channel E[h that feedback thresholding Γ initial value is set ²].

4. set and select user feedback desired value Kn (as transmitting antenna N _T=3 o'clock, Kn=20, N _T=4 o'clock, Kn=30).

Second step: each user carries out singular value decomposition to channel matrix $H_k=A_k{\mathrm{\&Sigma;}}_k{B}_k^H,$ Obtain maximum subchannel gains h _k ²With the pre-beam vectors b of optimum _kFor example at number of transmit antennas N _T=4 mimo systems, user 1 subchannel gains $h_{}^{}$ ${}_1^2=2.6558,$ Optimum pre-beam vectors is b ₁=[05272-04817+02855 ₁00594-01263 ₁03803-04945 ₁] ^T, user 2 subchannel gains is ${\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">h</figure-callout>}}_1^2=4.1632,$ Optimum pre-beam vectors is b ₁=[03283-03818+03123 ₁-01457-03202 ₁-04055-06007 ₁] ^T

The 3rd step: chooser channel gain h _K ²User more than or equal to feedback thresholding Γ feeds back.If for example current feedback thresholding Γ=3.5, user's 2 selected feedbacks then, and user 1 waits for.

The 4th step: selecteed user carries out pre-beam vectors code word and selects $f^{n_k}=\underset{f^i,i=1,,N}{\arg \min }{\left|\right|b_k{b}_k^H-f^i{\left(f^i\right)}^H\left|\right|}_{\mathrm{<figure-callout\; id="2"\; label="F"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">F</figure-callout>}}^2$ With the increment codebook selecting $h_k^2-\mathrm{\&Gamma;}:\&RightArrow;{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^m:\&RightArrow;C^m,$ For example the pre-beam vectors code word of user's 2 selections is ${\mathrm{<figure-callout\; id="2"\; label="f\; n"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">f</figure-callout>}}^{n_{}}={\left[\begin{array}{cccc}0.5093& -0.2962+0.1331i& 0.2629-0.6912i& -0.0640-0.2903i\end{array}\right]}^T,$ During N=8, vector codeword is numbered 001, during N=16, is numbered 0010.Gain delta is $h_k^2-\mathrm{\&Gamma;}=0.66,$ So the increment code word of selecting is 100.Then, these code words together feed back to transmitting terminal,

The 5th step: transmitting terminal carries out user's selection again according to the feedback information that step 5 obtains, and supposes to have 32 user feedbacks, then selects N again in these 32 users _T=4 users, promptly

1. select first user $\mathrm{use}$ $1=\mathrm{<figure-callout\; id="1"\; label="arg\; min\; i"\; filenames="C20051003766600281.png,C20051003766600282.png"\; state="\{\{state\}\}">arg</figure-callout>}\underset{i}{\min }\underset{}{}1/{\mathrm{SNR}}_i,$ I=1 ..., 32, wherein, ${\mathrm{SNR}}_i=(P/4){\tilde{h}}_{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">i</figure-callout>}}^2/{\mathrm{\&delta;}}^2,$ Here ${\tilde{h}}_1^2=\mathrm{\&Gamma;}+{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^{m_i}=3.5+0.6=4.1.$ P/ δ ²Be system's overall average signal to noise ratio, the value between for example desirable 0dB～20dB.

2. select user k, $\mathrm{usek}=\underset{i\&NotEqual;1}{\arg }\underset{k-1}{\min }(1/{\mathrm{SNR}}_i+{\mathrm{\&Sigma;}}_{j=1}^{k-1}{\mathrm{\&eta;}}_{j,i}),$ Here ${\mathrm{\&eta;}}_{k,i}={\left|{\left(f^{n_k}\right)}^H{f}^{n_i}\right|}^2.$

3. repeating step 2, select up to finishing 4 users.

The 6th step: directly make the user's who selects in the step 6 precoding be $f_k=\sqrt{P/N_T}{f}^{n_k},$ K=1 ..., N _T, for example, if user's 2 information symbols to be sent are x ₂=1+i, then user 2 transmission signal is ${\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="C20051003766600292.png"\; state="\{\{state\}\}">X</figure-callout>}}_2=\sqrt{P/4}{f}^{n_2}{x}_2=\sqrt{P/4}{\left[\begin{array}{cccc}0.5093+0.5093i& -0.4293-0.1631i& 0.9541-0.4283i& 02263-03543i\end{array}\right]}^T,$ System sends signal $\mathrm{Tx}=f_1{x}_1+f_2{x}_2+\&CenterDot;\&CenterDot;\&CenterDot;+f_{N_T}{x}_{N_T},$ X wherein _kInformation symbol for user k.

The 7th step: in the current transmission cycle, system is provided with feedback door limit value Γ, utilizes system broadcast channel to all users broadcastings, selects the user for next periodic system.Feedback door limit value Γ setting up procedure is as follows

1. the subchannel gains average h of all feedback users in the calculating current period ², for example ${\stackrel{\&OverBar;}{h}}^2=3.5+\frac{1}{32}\underset{i=1}{\overset{32}{\mathrm{\&Sigma;}}}{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^{m_i}.$

2. calculate maximum and the difference of actual feedback number of users and Kn, and calculated difference and peaked merchant, this merchant be multiply by the adjustment step delta Γ that the subchannel gains average obtains feeding back thresholding, for example $\mathrm{\&Delta;\&Gamma;}=\frac{32-30}{32}{\stackrel{\&OverBar;}{h}}^2.$

3. next periodic feedback thresholding is that current feedback door limit value deducts the adjustment step-length, for example Γ=3.5-Δ Γ.

Claims (1)

1. one kind based on the pre-beam shaping transmission method of quadrature that sends the assisted Selection user feedback, it is characterized in that: the first step: adopt following steps that system is carried out initialization:

A. determine system user channel correlation matrix R according to channel physical parameter and system parameters, and further determine maximum characteristic sub-channel gain profiles function of subscriber channel state matrix H distribution function and H and average statistical E[h by R ²],

B. meter systems total number of users K calculates normalization number of users kl=K/N _I, N _IBe number of transmit antennas,

C. initialization system ownership goal feedback quantity Kn is taken as transmitting antenna N _T6 to 10 times,

D. the initial feedback thresholding Γ that the system that provides starts working required makes that described initial feedback thresholding is E[h ²],

E., structure is set sends the required pre-beam vectors code book of signal, it is as follows that pre-beam vectors code book method is set:

A) give fixed system because the channel power degeneration factor desired value ζ that the quantification of beam vectors brings,

B) determine code word number N in the pre-beam vectors code book, N is the minimum positive integer that satisfies following two conditions:

1. the guaranteed output degeneration factor makes greater than desired value ζ $\widetilde{\mathrm{\&zeta;}}=\mathrm{\&zeta;}/E\left[h^2\right]$ Be normalized power degenerate system

The number scale value, promptly $N\&GreaterEqual;\frac{1-{\left(\widetilde{\mathrm{\&zeta;}}\right)}^{1/\mathrm{kl}}}{{((N_T-1)/4N_T)}^{N_T-1}(1-(N_T-1)/4N_T)},$

2. N is number of transmit antennas N _TIntegral multiple,

C) design Nm group vector codeword, Nm is N and the number of transmit antennas N that determines in the step b) _TThe merchant, every group of N _TIndividual vector codeword is mutually orthogonal, and detailed process is as follows:

1. generate a sample H of channel state matrix, carry out singular value decomposition H=A Λ B ^H, make first group of N _TIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _TIndividual column vector, this set of code words is designated as B ¹,

2. generate another sample H of channel state matrix, carry out singular value decomposition H=A Λ B ^HIf, B ^HB ¹, B ^HB ²..., B ^HB ^N-1In all element absolute value between 1/Nm and 1-1/Nm,

Then make n group N _TIndividual mutually orthogonal pre-beam vectors code word is respectively the N of B _IIndividual column vector, this set of code words is designated as B ⁿOtherwise,, then regenerate sample H,

3. 2. repeating step organizes N up to generating Nm _TIndividual mutually orthogonal pre-beam vectors code word, this set of code words is designated as B ^Nm,

F., structure is set sends the required subchannel gains increment code book of signal, the subchannel gains increment is meant that subchannel gains deducts the difference of feedback thresholding here, and it is as follows that subchannel gains incremental code this method is set:

A) according to the maximum characteristic sub-channel gain profiles of H function, determine the subchannel gains incremental range, establishing the gain delta scope is 0～Mh ², to all greater than Mh ²Gain delta all to make it be Mh ²,

B) decide gain delta code book number M, M is the minimum positive integer that satisfies following two conditions:

1. satisfy the given system requirements of system,

2. M is that the logarithm at the end is a positive integer with 2,

C) to 0～Mh ²The gain delta scope is cut apart, and detailed process is as follows:

1. to 0～Mh ²The normalization of gain delta scope makes it in 0～1 distribution,

2. 0～1 gain delta value is pressed A and leads compression,

3. the gain delta value after will compressing on average is divided into M interval, is made as Δ={ Δ ¹, Δ ²..., Δ ^M,

4. encoded in this M interval every log ₂M binary number constitutes a code word, and corresponding interval obtains gain delta code book C={C ¹, C ²..., C ^M,

Second step: system sends signal by the cycle, and the signal configuration method in each transmission cycle is as follows:

A. in the current transmission cycle, system is according to feedback thresholding Γ in the last transmission cycle, and certain customers feed back in the selective system, and detailed process is as follows:

1. user k, k=1 wherein ..., K, K are total number of users in the system, after receiving terminal is finished the channel status estimation, channel matrix are carried out singular value decomposition $H_k=A_k{\mathrm{\&Sigma;}}_k{B}_k^H,$ Obtain maximum subchannel gains value h _k ²,

2. all maximum subchannel gains values are selected as feedback user more than or equal to the user of feedback thresholding, and its feedback information is to transmitting terminal, and all the other users wait for, and establish feedback user and add up to K ',

B. it is as follows that Fan Kui user's chooser channel gain increment code word, and feed back to transmitting terminal by feedback channel, the word select of subchannel gains incremental code are selected process:

1. user k ', k '=1 wherein ..., K ' calculates maximum subchannel gains value h _{K '} ²With the difference of feedback thresholding,

2. determine the interval Δ of increment of this difference correspondence ^m _k,

3. the code word of selecting the interval corresponding code word of this increment will feed back for the user is made as C ^m _k,

C. the user of selected feedback selects to quantize pre-beam vectors, and feeds back to transmitting terminal by feedback channel together with subchannel gains increment code word, and it is as follows to quantize pre-beam vectors selection course:

1. user k ', k '=1 wherein ..., K ' is according to the channel singular value decomposition $H_{k^{\&prime;}}=A_{k^{\&prime;}}{\mathrm{\&Sigma;}}_{k^{\&prime;}}{B}_{k^{\&prime;}}^H$ Obtain optimum pre-beam vectors b _{K '}, b _{K '}Be B _kThe first row column vector,

2. from pre-beam vectors code book, select to quantize pre-beam vectors code word, promptly

$f^{n_k}=\underset{f^i,i=1,,N}{\arg \min }{\left|\right|b_{k^{\&prime;}}{b}_{k^{\&prime;}}^H-f^i{\left(f^i\right)}^H\left|\right|}_F^2,$ f ¹Be i vector codeword in the pre-beam vectors code book, f ⁿ _{K '}For user k ' feeds back pre-beam vectors;

D. transmitting terminal is selected N according to accurate orthogonal design method in the user that all feed back _TIndividual user, detailed process is as follows:

1. calculate the maximum subchannel gains of user k ' ${\tilde{h}}_{k^{\&prime;}}^2=\mathrm{\&Gamma;}+{\stackrel{\&OverBar;}{\mathrm{\&Delta;}}}^{m_k},$ Wherein, k '=1 ..., K ',

2. calculate user k ' receiving terminal signal to noise ratio ${\mathrm{SNR}}_{k^{\&prime;}}=(P/N_T){\tilde{h}}_{k^{\&prime;}}^2/{\mathrm{\&delta;}}^2,$ P is the transmitting terminal gross power, and δ is the low-angle expansion;

3. select first user $\mathrm{use}1=\arg \underset{k^{\&prime;}}{\min }1/{\mathrm{SNR}}_{k^{\&prime;}},$

4. select k " individual user $\mathrm{use}{k}^{\&prime;\&prime;}=\underset{i\&NotEqual;1,,k^{\&prime;\&prime;}-1}{\arg \min }(1/{\mathrm{SNR}}_i+{\mathrm{\&Sigma;}}_{j=1}^{k^{\&prime;\&prime;}-1}{\mathrm{\&eta;}}_{j,i}),$ Here η _{J, i}=| (f ⁿ _i) ^Hf ⁿ _j| ²,

5. repeating step 4., up to k "=N _T,

E. construct N _TIndividual selecteed user's signal at first makes user k ", wherein k "=1 ..., N _T, pre-beam vectors be $f_{k^{\&prime;\&prime;}}=\sqrt{P/N_I}{f}^{n_{k^{\&prime;\&prime;}}},$ " the transmission signal is f to obtain user k _{K "}x _{K "}, x here _{K "}Be user's information symbol to be passed,

F. tectonic system always sends signal, always sends signal by the N that directly superposes _TIndividual selecteed user's signal obtains, i.e. Tx=f ₁x ₁+ f ₂x ₂+ ... + f _NTx _NT,

G. in the current transmission cycle, system is provided with feedback door limit value Γ, utilizes system broadcast channel to all users broadcastings, selects the user for next periodic system, and feedback door limit value Γ setting up procedure is as follows:

1. calculate the subchannel gains average h of all feedback users in the current period ₂,

2. calculating K ' with maximum and the difference of Kn, and calculated difference and peaked merchant be multiply by the adjustment step-length that the subchannel gains average obtains feeding back thresholding with this merchant,

3. next periodic feedback thresholding is that current feedback door limit value deducts the adjustment step-length.

Images