CN105703812A - A downlink precoding and base station power control method in a pilot frequency time shifting large-scale MIMO system - Google Patents

Abstract

The invention provides a downlink precoding and base station power control method in a pilot frequency time shifting large-scale MIMO system. Through design of downlink precoding and base station power control, performance tradeoff of downlink data transmission and uplink channel estimation in a time shifting and pilot frequency structure is realized so as to raise frequency spectrum performance of the system. The idea of a maximum SLNR is applied to a time shifting pilot frequency structure, interferences leaked by a user to other users are suppressed, and interferences leaked to a base station which is carrying out channel estimation are suppressed because of introduction of conversion factors. On the basis of the above maximized SLNR precoding, through utilization of an idea of truncated polynomial PTE, a gradual downlink SINR expression only containing large scale information in a condition of a limited number of antennas is derived; and then relatively good base station downlink transmitting power and weighting factors are obtained through solving with maximization of graduality and the rate as targets, so that the performance tradeoff of the downlink data transmission and the uplink channel estimation are realized; and the frequency spectrum performance of the system can be improved while the complexity degree is lowered.

Description

Downlink precoding in a kind of extensive mimo system of pilot tone time shift and base station power control method

Technical field

The present invention relates to the interference coordination field in the extensive mimo system in radio communication, be specially the downlink precoding in a kind of extensive mimo system of pilot tone time shift and base station power control method。

Background technology

In order to improve the utilization rate of limited spectrum resources, meet the wireless communication needs that people are growing, one of extensive MIMO (LargeScaleMIMO or MassiveMIMO) key technology becoming new generation of wireless communication system。It uses user's service of identical running time-frequency resource by being equipped with the base station of hundreds of antennas simultaneously for dozens of。Research shows, along with the increase of antenna for base station number, noise and incoherent presence of intercell interference will be negligible；And adopt simple signal processing method, such as maximum-ratio combing (MaximumRatioCombining, MRC), ZF (Zero-Forcing, ZF) power system capacity just can being made to significantly improve, what reduce hardware in real system realizes difficulty。But, owing to channel coherency time length is limited, therefore it cannot be guaranteed that in extensive mimo system all use use orthogonal pilot frequency sequence per family, be subject to using the interference of the user of same pilot resource when user being carried out channel estimating in base station, this phenomenon is referred to as pilot pollution。

The method of pilot pollution is suppressed mainly to have at present: channel estimation method accurately, the precoding algorithms of robust, pilot configuration design and pilot frequency sequence allocation algorithm etc.。Wherein time shift pilot configuration is the one in pilot configuration algorithm for design, it is by dividing into some groups by little, and the pilot signal of different groups staggers in time, when number of antennas tends to infinite, pilot pollution only by with using the user of same pilot to be formed in group, greatly reduces pilot pollution level。But, the number of antennas of real system can not reach infinity, the uplink channel estimation of time shift pilot configuration is subject to the interference of the downlink data that other group base stations send and can not ignore, data sending power is more big, channel estimating will be more inaccurate, and data sending power reduces the reduction that can cause data SNR, it is therefore desirable to downlink data transmit power is carried out compromise and considers。Existing document have studied MRT and ZF precoding performance in the extensive mimo system of pilot tone time shift when antenna number is limited, and simulation result shows that antenna number has in limited time, and speed the trend of first increases and then decreases occurs along with the increase of base station power。But MRT and ZF precoding is only simple maximization available signal power or minimizes jamming power, and their performance in the extensive mimo system of pilot tone time shift is unsatisfactory。

Summary of the invention

For problems of the prior art, the present invention provides the downlink precoding in a kind of extensive mimo system of pilot tone time shift and base station power control method, controlled by the design of downlink precoding and base station power, realize the performance compromise of the downlink data transmission under time shift pilot configuration and uplink channel estimation, thus promoting the spectral performance of system。

The present invention is achieved through the following technical solutions:

Downlink precoding in a kind of extensive mimo system of pilot tone time shift and base station power control method, comprise the steps,

Step 1, obtains the SLNR of each user in extensive mimo system as follows,

Wherein, U_kiSLNR, α for i-th community kth user are conversion factor, P_fIt is base-station transmitting-power,Represent available signal power,Representing that user is leaked to the interference of other users of this community, Ar represents that user is sending that group community of uplink pilot signal,Represent that user is leaked to the interference of the base station carrying out uplink channel estimation；G_ikiChannel vector between kth user and i-th cell base station in i-th community,Conjugate transpose for this channel vector；W_kiFor the precoding of kth user in i-th community；G_ilRepresent the channel matrix between i-th cell base station and the l cell base station；K is the number of users of every community；I, l and k are positive integer；

Step 2, under the restriction maximizing SLNR precoding, the precoding obtaining each user according to the SLNR of each user isWherein,For the base station channel estimation vector to this community user k in i-th community,Conjugate transpose for this channel estimation vector；Variance for this channel estimation errors；The conjugate transpose of the channel matrix between expression i-th cell base station and the l cell base station；I_MUnit matrix is tieed up for M × M；

Step 3, utilizes the precoding that step 2 is obtained by the method for Representation theorem to simplify, obtains the progressive descending Signal to Interference plus Noise Ratio containing only large scale information；

Step 4, the progressive descending Signal to Interference plus Noise Ratio obtained in the precoding obtained according to step 2 and step 3, and speed progressive with users all in extensive mimo system is optimization aim to the maximum, obtains the downlink transmission power P of optimum_fWith conversion factor α；

Step 5, controls user and adopts the maximization letter leakage noise ratio precoding obtained by the optimum translation factor, control base station and adopt the optimum downlink transmission power obtained to carry out data transmission。

Preferably, in step 2, maximize SLNR precoding and be expressed as optimization problem:

s.t.||w_ki||²=1, k=1 ..., K, i=1 ..., L

Wherein,For the conjugate transpose of the precoding of kth user, M in i-th community_ki,N_kiRespectively,

Further, in step 3, Representation theorem is adopted to simplify precoding, in precodingCarry out Taylor expansion, and intercept finite term, namely

Wherein, N is the length of Representation theorem, b_ki,nFor Representation theorem coefficient, it makes beamforming vector meet normalization restriction。

Further, in step 4, adopt optimum downlink transmission power and the conversion factor of particle cluster algorithm Optimization Solution system。

Preferably, in step 4, and speed progressive with users all in extensive mimo system is optimization aim to the maximum and is expressed as follows,

s.t.0≤P_f≤ Ρ；

0≤α≤A

Wherein,For the progressive Signal to Interference plus Noise Ratio of kth user in i-th community, L is the community number in system, and Ρ and A is Ρ respectively_fThe empirical value upper bound with α。

Compared with prior art, the present invention has following useful technique effect:

The thought maximizing letter leakage noise ratio SLNR is applied in time shift pilot configuration by the present invention, not only inhibits user to be leaked to the interference of other users, and inhibits, by introducing conversion factor, the interference being leaked to carry out the base station of channel estimating；On the basis of above-mentioned maximization SLNR precoding, the present invention utilizes the thought of Representation theorem TPE, the progressive descending SINR expression formula containing only large scale information when number of antennas of having derived is limited, then it is object solving to the maximum with progressive and speed and obtains preferably base station down transmit power and weight factor, realize the performance compromise of downlink data transmission and uplink channel estimation, thus promoting the spectral performance of system while reducing complexity。

Accompanying drawing explanation

Fig. 1 is the extensive mimo system scene schematic diagram described in present example。

Fig. 2 is the pilot tone time shift transmission mechanism schematic diagram described in present example。

Fig. 3 places under the scene of 100 antennas and total seven cell the every user's Mean Speed of different schemes with the change curve of downlink data signal to noise ratio in the base station described in present example。

Fig. 4 places under the scene of 100 antennas and total seven cell the CDF curve of every user's Mean Speed in different schemes in the base station described in present example。

Fig. 5 arranges, described in present example, the curve that under the scene of seven cell, every user's Mean Speed of different schemes changes with antenna for base station number。

Detailed description of the invention

Below in conjunction with specific embodiment, the present invention is described in further detail, and the explanation of the invention is not limited。

The precoding based on SLNR that the present invention relates in extensive mimo system under time shift pilot configuration and base station power control。First, the thought maximizing letter leakage noise ratio SLNR is applied in time shift pilot configuration, not only allow for user and be leaked to the interference of other users, and by introducing conversion factor, the interference being leaked to carry out the base station of channel estimating is taken into account；Secondly, and speed progressive with system is target to the maximum, solves conversion factor and the downlink data transmit power of optimum, thus realizing the performance compromise of downlink data transmission and uplink channel estimation。

The present invention proposes the downlink precoding under a kind of time shift pilot configuration and base station power control program。The core concept of the program is: the first step, the thought maximizing letter leakage noise ratio (SLNR) is applied in time shift pilot configuration, not only allow for user and be leaked to the interference of other users, and by introducing conversion factor, the interference being leaked to carry out the base station of channel estimating is taken into account；Second step, and speed progressive with system is target to the maximum, solves conversion factor and the downlink data transmit power of optimum, thus realizing the performance compromise of downlink data transmission and uplink channel estimation。

Consider an extensive mimo system of multiple cell, as shown in Figure 1, it is assumed that having L community, each community to comprise a base station being equipped with M root antenna and K single-antenna subscriber (K≤M), adopt the transmission means of TDD, up-downgoing channel has reciprocity。In jth community, the channel vector between kth user and the l cell base station is designated asWherein β_lkjRepresent large scale path loss and shadow fading coefficient, h_lkjRepresenting little yardstick channel vector, its element is separate and obeys multiple Gauss CN (0,1) distribution。Represent the channel matrix between base station i and base station l, comprise direct-view footpath (LOS) and non-straight visual path (NLOS), wherein c_ilFor large scale coefficient, H_ilTieing up little yardstick channel matrix for (M × M), its element obeys L-S distribution。

Consider pilot tone time shift transmission mechanism, as shown in Figure 2。In this mechanism, intrasystem L base station be divided into Γ group (A1 ..., A Γ) (Fig. 2 has divided 3 groups), when the user in a certain group send uplink pilot signal time, the base station in all the other groups send downlink data signal。Obtaining the base station i ∈ Ar pilot signal received is,

In formula:

ρ_k'jKth in the j of community ' the pilot tone transmit power of individual user；

ψ_k'The pilot frequency sequence that user k' uses, ψ_k=[ψ_k1,…,ψ_kτ], | ψ_kj|=1。Make τ=K, then

P_fDownlink data transmit power；

w_k'lKth in the l of community ' the transmission beamforming vector of individual user；

s_k'lKth in the l of community ' the downlink data vector that sends of individual user；

z_iNoise item, element is separate and obeys multiple Gauss CN (0,1) distribution。

The MMSE channel estimating of community j user k is by base station l

Wherein

It is the form estimating channel plus estimation difference by Channel Modeling, namely

Due to separate between MMSE channel estimation value and estimation difference, therefore the covariance matrix of channel estimation errors is

OrderThen may be assumed that channel estimation errors vector meetsIt can be seen that data sending power is more big, the variance of channel estimation errors is more big, and channel estimating is more inaccurate。But, data sending power reduces the reduction that can cause again data SNR。Therefore, different from conventional transmission mechanism, in the mode of pilot tone time shift, every user's average spectral efficiency (ase) no longer continues to increase with the increase of downlink data signal to noise ratio, but presents the trend of first increases and then decreases。Therefore under this transmission mechanism, it is necessary to downlink data transmit power is carried out compromise and considers。

The basic step of downlink precoding and power control scheme is:

The design of 1 precoding

Maximize SLNR precoding and can reduce the interference to other users while ensureing useful signal quality, therefore consider to be applied in the pilot configuration of time shift by SLNR precoding, not only allow for user and be leaked to the interference of other users, and the interference being leaked to carry out the base station of channel estimating is taken into account。Owing to extensive MIMO every community user number is more, between intra-cell users, interference is the most serious, therefore it is considered herein that ignore the interference being leaked to other community users。The SLNR of the i-th (i ∈ At, At are the community group sending downlink data) individual community kth user is

Wherein, molecule is available signal power, and denominator is the interference revealed。In denominator, Section 1 is be leaked to the interference of other users of this community, and Section 2 represents the interference being leaked to the base station carrying out uplink channel estimation, and α is conversion factor, and 1 is noise item。Due to interference between user when Section 1 represents downlink data transmission in denominator, and Section 2 represents the interference that ascending pilot frequency is transmitted by downlink data transmission, and two kinds of interference are different on the form that affects of system transfers performance, it is therefore necessary to introduce conversion factor α。

Wherein available signal powerCan be approximately

Then maximize SLNR precoding and can be expressed as optimization problem:

s.t.||w_ki||²=1, k=1 ..., K, i=1 ..., L

Wherein M_ki,N_kiRespectively

Variance for channel estimation errors。Formula (7) is typical Rayleigh entropy form, and its closed solutions is

2 descending powers and conversion factor optimization

With and speedIt is optimization aim to the maximum, carrys out combined optimization downlink transmission power P_fWith conversion factor α。The descending Signal to Interference plus Noise Ratio expression formula of i-th (i ∈ At) individual community kth user is

Wherein, At' represents that except At other are sending the community group of downlink data, and Ar is the community group sending uplink pilot signal。The molecule of formula (10) represents available signal power, in denominator, Section 1 represents from the interference using same pilot user in group, Section 2 represents with the interference using different pilot tone user in group, Section 3 represents the interference from difference group user, and Section 4 represents the interference from the user sending pilot group。By formula (10) it can be seen that object function is information-related with transient channel, if little yardstick channel information changes, it is necessary to re-execute optimization process, algorithm complex is still difficult to accept。On the other hand, optimized variable herein is base station down transmit power and conversion factor, from the angle of system realize, they should have certain stability, not should with little dimensional information Rapid Variable Design。

Consideration based on above two aspects, it is necessary to optimization problem is adjusted, will be adjusted to progressive with speed maximization problems and speed will be maximum, and make optimization object function only relevant with large-scale channel information。Concrete thought is: first, utilizes the thought of Representation theorem (TPE), derives containing only the SINR expression formula having large-scale channel information。Then, it is target search to the maximum with progressive and speed and solves preferably transmit power and conversion factor, the method that population (PSO) that what the present invention adopted is is searched for。

1) progressive SINR derives

The progressive descending Signal to Interference plus Noise Ratio lower bound of i-th (i ∈ At) individual community kth user is

Wherein, w_kiFor precoding, as shown in (9) formula。The method adopting TPE is namely rightCarry out Taylor expansion, and intercept finite term, namely

Wherein, b_ki,nFor Representation theorem coefficient, it must make beamforming vector meet normalization restriction。(12) formula is substituted into formula (9), then avoids the matrix inversion operation in precoding, while reducing complexity, also allow for the derivation of progressive SINR。

Representation theorem generally at most intercepts first five items, it is contemplated that the computation complexity present invention intercepts first three items, namely takes N=2, just can obtain close to optimum performance when can be seen that N=2 from simulation result。

Make S_kiRepresenting available signal power, namely the molecule in formula (11) is

WhereinFor the variance of little yardstick channel estimating, P_fDownlink transmission power, α is conversion factor, b_ki,nFor Representation theorem coefficient, β_lkjFor the large-scale channel information between base station to user, c_ilFor the large-scale channel information between base station。

It is likewise possible to the interference derived between with the interior user using same pilot of group, namelyWherein have

With the inter-user interference using different pilot tone in group it isWherein

Interference between the different group of order is N_ki,3, due to separate between precoding and the channel estimating of difference group, therefore have

The Pilot Interference making other groups is N_ki,4, then have

Based on above derivation, it is possible to obtain the expression formula of descending progressive Signal to Interference plus Noise Ratio。

Therefore, optimization problem is:

s.t.0≤P_f≤Ρ(18)

0≤α≤A

Wherein

Ρ and A is Ρ respectively_fThe empirical value upper bound with α。

2) Representation theorem coefficient optimizes

Formula (14) introduces Representation theorem coefficient, and in order to reduce the impact of truncated error as far as possible, we are target to the maximum to choose Representation theorem coefficient b with progressive SINR_ki=[b_ki,0,b_ki,1,b_ki,2]^T。Owing to the optimization of the Representation theorem coefficient to all users is identical and mutually independent, therefore to statement is convenient, it is omitted here subscript " ki "。Optimization problem is

s.t.b^HCb=1

Matrix B in formula (20), the coefficient matrix of the corresponding parameter composition of D respectively formula (14) and (15), Matrix C makes beamforming vector meet normalization constraint, matrix A=aa^H, wherein a is

MatrixWherein:

Optimization problem (21) is broad sense Rayleigh entropy form, therefore can try to achieve Representation theorem coefficient and be

B=η (B+D+1/P_f(N₃+N₄+1)C)^-1a(23)

Wherein, η makes Representation theorem meet the restrictive condition in (20) formula, and its expression formula is

Formula (23) is substituted into formula (19), only relevant with large-scale channel information descending progressive SINR can be finally given, substituted into the optimization problem of formula (18), just establish and be the base station down transmit power of optimization aim and the combined optimization problem of conversion factor to the maximum with progressive and speed。

The present invention simulates the performance of downlink precoding and the power control scheme put forward, and contrasts with high specific transfer pre-coding, the maximization SLNR precoding being not optimised and high complexity precoding and power control scheme。Wherein, high complexity precoding and power control scheme, refer to adopt the maximization SLNR precoding carried, and be optimization aim to the maximum with instantaneous and speed and come the optimum transmit power of Optimization Solution and conversion factor, namely each TTI performs primary particle group's algorithm and comes optimum transmit power and the conversion factor of this TTI of Optimization Solution。The complexity of program Optimization Solution is significantly high, and the complexity of each Drop is o (ST (2L²K²·M²+LKM²+LM³))+o (2ST)。And the low complex degree precoding proposed and power optimization scheme, the complexity of each Drop is o (SN³)+o (2S), wherein T represents the TTI number of each Drop, and S is population, and N is the length of Representation theorem。(note, owing to the computation complexity of two scheme precoding step is identical, herein do not comprise precoding complexity in analysis of complexity)。It can be seen that the algorithm complex of the precoding of carried low complex degree and power joint prioritization scheme greatly reduces。Simulation parameter reference table 1。

Table 1MASSIVEMIMO system descending Propagation Simulation parameter

Fig. 3 gives and maximizes under time shift pilot configuration under SLNR precoding and high specific transfer pre-coding and conventional alignment pilot configuration that the spectrum efficiency of high specific transfer pre-coding is with the change curve of downlink data signal to noise ratio, and choose antenna number is 100 herein。It can be seen that system spectrum increases along with the increase of downlink data signal to noise ratio under conventional alignment pilot configuration, finally present a platform。But the spectrum efficiency under time shift pilot configuration is to present first to increase the trend subtracted afterwards。This is owing to when antenna for base station Limited Number, under the pilot configuration of time shift, the interference of the downlink data that base station is subject to other group transmissions when carrying out uplink channel estimation can not be ignored, and data sending power is more big, and channel estimating is more inaccurate。But, data sending power reduces the reduction that can cause again data SNR。Therefore, every user's average spectral efficiency (ase) no longer increases with the increase of downlink data signal to noise ratio, but presents the trend of first increases and then decreases。Therefore, under this transmission mechanism, suitable downlink transmission power is selected to be a need for。

Fig. 4, Fig. 5 simulate the spectral performance of put forward precoding and power control scheme, and contrast with the high specific transfer pre-coding of power optimization and precoding and the power control scheme of the maximization SLNR precoding being not optimised and high complexity, the power that the power ratio of the SLNR scheme selection that wherein power is not optimised optimizes out is big, for P_f=30, conversion factor is chosen identical with suggested plans；The conversion factor of the SLNR scheme selection that conversion factor is not optimised is α=0.5, power with suggested plans identical。Fig. 4 is the CDF curve of every user's Mean Speed in these five kinds of situations, and Fig. 5 gives the curve that in these five kinds of situations, every user's Mean Speed changes with antenna number。Can be seen that time shift pilot configuration lower carried maximization SLNR precoding spectrum efficiency performance be better than traditional MRT precoding, this be due to carried maximization SLNR precoding ensure useful signal quality while can suppress the interference to other users and the interference to the base station carrying out channel estimating。Simultaneously, although the power that power ratio selected by the maximization SLNR pre-coding scheme that power is not optimised is suggested plans is big, the average per-user speed suggested plans still maximizes the high of SLNR precoding than what be not optimised, this is owing to downlink transmission power is too big, can cause that precision of channel estimation declines, and carried algorithm can select suitable down transmitting power and conversion factor, it is possible to realize the better compromise of uplink channel estimation and downlink user Signal to Interference plus Noise Ratio, thus promote system and rate capability；What conversion factor was not optimised maximizes conversion factor selected by SLNR pre-coding scheme is not optimum, it is impossible to realizing the performance compromise between uplink channel estimation and downlink data transmission, therefore average per-user speed does not have carried algorithm height。It addition, the performance of carried precoding and power control scheme is close to the scheme solving optimal power and conversion factor based on instantaneous and speed, but being suggested plans and only need to just perform optimized algorithm when channel large scale information changes, complexity is substantially reduced。

Claims (5)

1. the downlink precoding in the extensive mimo system of pilot tone time shift and base station power control method, it is characterised in that comprise the steps,

Step 1, obtains the SLNR of each user in extensive mimo system as follows,

$U_{ki}=\frac{P_f{\left|g_{iki}^H{w}_{ki}\right|}^2}{\underset{k^{\′}=1,k^{\′}\≠k}{\overset{K}{\Σ}}{P}_f{\left|g_{{\mathrm{ik}}^{\′}i}^H{w}_{ki}\right|}^2+\mathrm{\α}\underset{l\∈Ar}{\Σ}{P}_f\left|\right|G_{il}{w}_{ki}|{|}^2+1};$

Wherein, U_kiSLNR, α for i-th community kth user are conversion factor, P_fIt is base-station transmitting-power,Represent available signal power,Representing that user is leaked to the interference of other users of this community, Ar represents that user is sending that group community of uplink pilot signal,Represent that user is leaked to the interference of the base station carrying out uplink channel estimation；G_ikiChannel vector between kth user and i-th cell base station in i-th community,Conjugate transpose for this channel vector；W_kiFor the precoding of kth user in i-th community；G_ilRepresent the channel matrix between i-th cell base station and the l cell base station；K is the number of users of every community；I, l and k are positive integer；

Step 2, under the restriction maximizing SLNR precoding, the precoding obtaining each user according to the SLNR of each user is $w_{ki}={\left(\underset{k^{\′}=1,k^{\′}\≠k}{\overset{K}{\Σ}}\right({\hat{g}}_{{\mathrm{ik}}^{\′}i}{\hat{g}}_{{\mathrm{ik}}^{\′}i}^H+{\mathrm{\σ}}_{{\mathrm{ik}}^{\′}i}^2{I}_M)+\underset{l\∈Ar}{\Σ}{G}_{il}{G}_{il}^H+\frac{1}{P_f}{I}_M)}^{-1}{\hat{g}}_{iki},$ Wherein,For the base station channel estimation vector to this community user k in i-th community,Conjugate transpose for this channel estimation vector；Variance for this channel estimation errors；The conjugate transpose of the channel matrix between expression i-th cell base station and the l cell base station；I_MUnit matrix is tieed up for M × M；

Step 3, utilizes the precoding that step 2 is obtained by the method for Representation theorem to simplify, obtains the progressive descending Signal to Interference plus Noise Ratio containing only large scale information；

Step 4, the progressive descending Signal to Interference plus Noise Ratio obtained in the precoding obtained according to step 2 and step 3, and speed progressive with users all in extensive mimo system is optimization aim to the maximum, obtains the downlink transmission power P of optimum_fWith conversion factor α；

Step 5, controls user and adopts the maximization letter leakage noise ratio precoding obtained by the optimum translation factor, control base station and adopt the optimum downlink transmission power obtained to carry out data transmission。

2. control method according to claim 1, it is characterised in that in step 2, maximizes SLNR precoding and is expressed as optimization problem:

$\underset{w_{ki}}{max}{U}_{ki}=\frac{w_{ki}^H{M}_{ki}{w}_{ki}}{w_{ki}^H{N}_{ki}{w}_{ki}};$

s.t.||w_ki||²=1, k=1 ..., K, i=1 ..., L

Wherein,For the conjugate transpose of the precoding of kth user, M in i-th community_ki,N_kiRespectively,

$M_{ki}={\hat{g}}_{iki}{\hat{g}}_{iki}^H+{\mathrm{\σ}}_{iki}^2{I}_M$

$N_{ki}=\underset{k^{\′}=1,k^{\′}\≠k}{\overset{K}{\Σ}}({\hat{g}}_{{\mathrm{ik}}^{\′}i}{\hat{g}}_{{\mathrm{ik}}^{\′}i}^H+{\mathrm{\σ}}_{{\mathrm{ik}}^{\′}i}^2{I}_M)+\mathrm{\α}\underset{l\∈Ar}{\Σ}{G}_{il}{G}_{il}^H+\frac{1}{P_f}{I}_M.$

3. control method according to claim 2, it is characterised in that in step 3, adopts Representation theorem to simplify precoding, in precodingCarry out Taylor expansion, and intercept finite term, namely

$N_{ki}^{-1}\≈\underset{n=0}{\overset{N}{\Σ}}{b}_{ki,n}{(\underset{k^{\′}=1,k^{\′}\≠k}{\overset{K}{\Σ}}{\hat{g}}_{{\mathrm{ik}}^{\′}i}{\hat{g}}_{{\mathrm{ik}}^{\′}i}^H+\mathrm{\α}\underset{l\∈Ar}{\Σ}{G}_{il}^H{G}_{il})}^n;$

Wherein, N is the length of Representation theorem, b_ki,nFor Representation theorem coefficient, it makes beamforming vector meet normalization restriction。

4. control method according to claim 3, it is characterised in that in step 4, adopts optimum downlink transmission power and the conversion factor of particle cluster algorithm Optimization Solution system。

5. control method according to claim 1, it is characterised in that in step 4, and speed progressive with users all in extensive mimo system is optimization aim to the maximum and is expressed as follows,

$\underset{P_f,\mathrm{\α}}{max}\underset{i=1}{\overset{L}{\Σ}}\underset{k=1,}{\overset{K}{\Σ}}log(1+{\stackrel{\‾}{SINR}}_{ki})$

s.t.0≤P_f≤ Ρ；

0≤α≤A

Wherein,For the progressive Signal to Interference plus Noise Ratio of kth user in i-th community, L is the community number in system, and Ρ and A is Ρ respectively_fThe empirical value upper bound with α。