CN105227222A - A kind ofly utilize the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information - Google Patents

Abstract

The invention discloses and a kind ofly utilize the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information.The optimization problem that first the present invention utilizes the virtual up-link power of statistical channel state information process to distribute, and utilize the transfer problem of uplink downlink duality process uplink downlink power division, thus obtain the downlink power allocation vector optimized; Secondly, utilize instantaneous channel state information to calculate downlink beamforming vector, thus obtain the downlink beamforming vector optimized.The method that the present invention proposes can avoid power renewal frequency too high, reduces computation complexity.Meanwhile, its system energy efficiency performance is close to its system energy efficiency performance of method utilizing instantaneous channel state information in prior art.Therefore, the method that the present invention proposes can weigh system energy efficiency performance and computation complexity effectively.

Description

A kind ofly utilize the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information

Technical field

The present invention relates to wireless communication technology field, particularly extensive multiple-input and multiple-output (multiple-inputmul-tiple-output, MIMO) efficiency transmission technology.

Background technology

In recent years, with the fast development of the universal of smart mobile phone with mobile broadband, wireless data traffic presented explosive growth.In order to meet the demand of user, operator will take to increase the measure such as base station deployment and antenna configuration, and this will bring the sharply increase of energy ezpenditure, how to weigh the research emphasis that power system capacity and energy resource consumption become wireless communication technology.

In order to weigh power system capacity and energy resource consumption, prior art proposes with energy efficiency (i.e. the ratio of system weighted sum rate and total power consumption) as optimization aim, such as, utilize a multiuser MIMO efficiency beamforming design for uplink downlink duality, its weak point is to sacrifice spectrum efficiency for cost.

Extensive MIMO technology is one of key technology in the 5th third-generation mobile communication technology (5G), and this technology can improve spectrum efficiency and energy efficiency simultaneously, and has very large Improvement in robustness and reliability.If continue to use prior art to design extensive MIMO efficiency beam forming solutions, then the operand of a power renewal is comparatively large and power renewal frequency is too high, and its computation complexity limits the application of prior art in extensive mimo system.Therefore, extensive MIMO efficiency transmission technology will face the problem how reducing computation complexity.

Summary of the invention

In order to overcome the deficiencies in the prior art, the present invention proposes one and utilizes statistical channel state information (channelstateinformation, CSI) the extensive MIMO beam-forming method of high energy efficiency, the method can reduce computation complexity from two aspects: the operand reducing by a power renewal, and reduces power renewal frequency.

Utilize the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information, its step comprises:

Obtain the statistical channel state information between single base station and collaboration user;

The optimization problem that single base station utilizes the virtual up-link power of statistical channel state information process to distribute, obtains the virtual up-link power allocation vector optimized;

Utilize the transfer problem of uplink downlink duality process uplink downlink power division according to the virtual up-link power allocation vector of described optimization, obtain the downlink power allocation vector optimized;

Obtain the instantaneous channel state information between single base station and collaboration user;

Utilize instantaneous channel state information to calculate downlink beamforming vector according to the virtual up-link power allocation vector of described optimization, obtain the downlink beamforming vector optimized.

What the present invention carried a kind ofly utilizes the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information, be characterized in the optimization problem first utilizing the virtual up-link power of statistical channel state information process to distribute, and utilize the transfer problem of uplink downlink duality process uplink downlink power division, thus obtain the downlink power allocation vector optimized; Secondly, utilize instantaneous channel state information to calculate downlink beamforming vector, thus obtain the downlink beamforming vector optimized.Can find thus, upgrading downlink power allocation vector is utilize statistical channel state information instead of instantaneous channel state information, thus can reduce the operand of a power renewal.In addition, in duration time interval T, downlink power allocation is upgraded, and the downlink power allocation vector once upgraded before maintaining in duration time interval T; Further, in duration time interval τ (T > > τ), upgrade downlink beamforming vector, power renewal frequency effectively can be avoided so too high, reduce operand further.Therefore, a kind of extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information that utilizes that the present invention proposes can reduce computation complexity significantly.

Accompanying drawing explanation

Fig. 1 is the system model figure of the embodiment of the present invention;

Fig. 2 is a kind of flow chart utilizing the high energy efficiency of statistical channel state information extensive MIMO velocity of wave manufacturing process that the present invention proposes;

Fig. 3 is the flow chart that in the present invention, single base station utilizes the virtual up-link power allocation optimization problems of statistical channel state information process;

Fig. 4 is the graph of relation that in the present invention, single base station utilizes efficiency optimization target values and the power consumption factor in virtual its process of up-link power allocation optimization problems of statistical channel state information process;

Fig. 5 is the system energy efficiency performance of institute of the present invention extracting method and the graph of relation of single base station transmit antennas number.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail.Should understand these embodiments to be only not used in for illustration of the present invention and to limit the scope of the invention, after reading this disclosure, the amendment of those skilled in the art to the various equivalent form of value of the present invention all falls within right appended by the present patent application.

Application scenarios: the multi-user downlink system of a single base station, the number of users of serving is K, and wherein, base-station node configuration N transmit antennas and user node configuration single received antenna, system model figure is see Fig. 1.

Optimization aim: through-put power and QoS of customer (qualityofservice to greatest extent in single base station, QoS) under the constraints such as demand, maximize system energy efficiency performance (energyefficiency, EE), the i.e. ratio of system weighted sum rate and total power consumption, its mathematic(al) representation is

$s.t.{\stackrel{\→}{\mathrm{\γ}}}_k\≥{\mathrm{\γ}}_k\∀k,\left|\right|w_k|{|}^2=1\∀k,{\mathrm{\Σ}}_{k=1}^K{p}_k\≤P$

Wherein, ${\stackrel{\→}{\mathrm{\γ}}}_k=\frac{p_k\left|\right|h_k^H{w}_k|{|}^2}{{\mathrm{\Σ}}_{\underset{m\≠k}{m=1}}^K,p_m\left|\right|h_k^H{w}_m|{|}^2+1};$

W=[w ₁..., w _k], represent the beam vector of collaboration user k;

P=[p ₁..., p _k] ^t, p _krepresent the downlink power allocation value of collaboration user k;

represent the normalization channel coefficients vector between single base station and collaboration user k;

represent speed weighted factor and the target Signal to Interference plus Noise Ratio of collaboration user k respectively;

P,P _c, P ₀represent single base station through-put power to greatest extent respectively, the circuit-level power consumption that single antenna is fixed and single base station maintain the base power consumption of normal work; represent the poor efficiency of single base station power amplifier.

Utilize the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information, its flow chart is see Fig. 2, and concrete implementation step comprises:

Step 201, obtain statistical channel state information between single base station and collaboration user;

The optimization problem that step 202, single base station utilize the virtual up-link power of statistical channel state information process to distribute, obtains the virtual up-link power allocation vector optimized;

Step 203, utilize the transfer problem of uplink downlink duality process uplink downlink power division according to the virtual up-link power allocation vector of described optimization, obtain the downlink power allocation vector optimized;

Step 204, obtain instantaneous channel state information between single base station and collaboration user;

Step 205, to utilize according to the virtual up-link power allocation vector of described optimization instantaneous channel state information to calculate downlink beamforming vector, obtain the downlink beamforming vector optimized.

It should be noted that, step 201 to step 203 is the optimization problems utilizing the virtual up-link power of statistical channel state information process to distribute, and utilize the transfer problem of uplink downlink duality process uplink downlink power division, the downlink power allocation be optimized vector; Step 204 and step 205 utilize instantaneous channel state information to calculate downlink beamforming vector, the downlink beamforming be optimized vector.

To step 202 be illustrated, step 203 below, the concrete feasible implementation method of step 205,

Such as, the concrete feasible implementation method of step 202, its flow chart, see Fig. 3, comprising:

Step 301, initialization of virtual up-link power allocation vector;

It can be specifically minimum power under single base station to greatest extent constraints such as through-put power and QoS demand, namely minimum power (powerminimization in prior art is utilized, PM) the virtual up-link power allocation vector of algorithm initialization, thus obtain virtual up-link power allocation vector initial value q ⁽⁰⁾, wherein, represent the virtual up-link power apportioning cost of collaboration user k.

Step 302, external iteration upgrade the power consumption factor;

Can be specifically determine according to linear search method

$\{{\mathrm{\α}}_{\min }^{\left(n\right)},{\mathrm{\α}}_{\mathrm{mi}d1}^{\left(n\right)}=\frac{2{\mathrm{\α}}_{\min }^{\left(n\right)}+{\mathrm{\α}}_{\max }^{\left(n\right)}}{3},{\mathrm{\α}}_{\mathrm{mi}d2}^{\left(n\right)}=\frac{{\mathrm{\α}}_{min}^{\left(n\right)}+2{\mathrm{\α}}_{\max }^{\left(n\right)}}{3},{\mathrm{\α}}_{\max }^{\left(n\right)}\},$

Wherein, n represents external iteration number of times, initial value

Step 303, statistical channel state information internal layer iteration is utilized to upgrade virtual up-link power allocation vector;

Concrete feasible implementation method comprises the steps:

According to the power consumption factor-alpha of current external iteration ⁽ⁿ⁾initialization of virtual up-link power allocation vector q ^(m), wherein, m represents internal layer iterations, initial value m=0;

According to the virtual up-link power allocation vector q of current internal layer iteration ^(m)calculate

$e_k=\frac{1}{N}Tr\left({\mathrm{\Θ}}_{k}T\right),T={(\frac{1}{N}\underset{n=1}{\overset{K}{\Σ}}\frac{q_n{\mathrm{\Θ}}_n}{1+e_n}+I)}^{-1},$

Obtain wherein, Θ _krepresent the normalization spatial correlation matrix between single base station and collaboration user k;

According to the power consumption factor-alpha of current external iteration ⁽ⁿ⁾broad sense water-filling algorithm is utilized to upgrade virtual up-link power allocation vector, namely

Or, search optimum level of water line meet power constraints ${\mathrm{\Σ}}_{k=1}^K{q}_k=\mathrm{\αP},$ Obtain the virtual up-link power allocation vector q after upgrading ^(m+1), wherein,

Judge whether to stop internal layer iteration, if then by described virtual up-link power allocation vector q ^(m+1)as the virtual up-link power allocation vector q of current external iteration ^{* (n)}, otherwise return the virtual up-link power allocation vector calculating e performed according to current internal layer iteration, wherein ζ is a predetermined threshold value.

If when step 303 also can be understood as external iteration frequency n=0, then according to the power consumption factor of current external iteration utilize statistical channel state information internal layer iteration to upgrade virtual up-link power allocation vector, obtain the virtual up-link power allocation vector of current external iteration, be designated as otherwise according to the power consumption factor of current external iteration utilize statistical channel state information internal layer iteration to upgrade virtual up-link power allocation vector, obtain the virtual up-link power allocation vector of current external iteration, be designated as

Step 304, calculating efficiency optimization target values, and utilize interval elimination approach to shorten the region of search;

If when can be specifically external iteration frequency n=0, then according to the power consumption factor of current external iteration with the virtual up-link power allocation vector of current external iteration calculate the efficiency optimization target values of current external iteration, be designated as otherwise according to the power consumption factor of current external iteration with the virtual up-link power allocation vector of current external iteration calculate the efficiency optimization target values of current external iteration, be designated as wherein, the computing formula of efficiency optimization target values is

$EE={\mathrm{\Σ}}_{k=1}^K{\stackrel{\‾}{\mathrm{\θ}}}_k{\log }_2(1+q_k{e}_k).$

Interval elimination approach is utilized to shorten the region of search, if then ${\mathrm{EE}}_{\max }^{\left(n\right)}={\mathrm{EE}}_{mid2}^{\left(n\right)},$ Otherwise, ${\mathrm{\α}}_{\min }^{\left(n\right)}={\mathrm{\α}}_{mid1}^{\left(n\right)},q_{\min }^{*\left(n\right)}=q_{mid1}^{*\left(n\right)},{\mathrm{EE}}_{\min }^{\left(n\right)}={\mathrm{EE}}_{mid1}^{\left(n\right)};$

Step 305, judge whether to meet external iteration stop condition, if then perform step 306, otherwise return execution step 302;

Step 306, general or as the virtual up-link power allocation vector q optimized ^*.

Again such as, the concrete feasible implementation method of step 203 comprises the steps:

According to the virtual up-link power allocation vector q optimized ^*utilize the transfer problem of uplink downlink duality process uplink downlink power division, i.e. p=(A ⁰) ^-11, obtain the downlink power allocation vector p optimized ^*, wherein, 1 represents that all elements is all the K dimensional vector of 1,

$A_{k,m}^o=\left\{\begin{array}{c}\frac{|e_k{|}^2}{{\stackrel{\←}{\mathrm{\γ}}}_k|e_k^{\′}{|}^2},k=m\\ -\frac{e_{k,m}^{\′}}{{(1+e_k)}^2|e_m^{\′}{|}^2},k\≠m\end{array}\right.,$ Wherein, ${\stackrel{\←}{\mathrm{\γ}}}_k=q_k{e}_k,$

Make e '=[e ' ₁..., e ' _k] ^t, e ' _m=[e ' _{1, m}..., e ' _{k, m}] ^t, then e '=(I _k-J) ^-1v, e ' _m=(I _k-J) ^-1v _m,

Wherein, $\left\{\begin{array}{c}{\[J\]}_{k,n}=\frac{Tr\left({\mathrm{\Θ}}_k{\mathrm{Tq}}_n{\mathrm{\Θ}}_{n}T\right)/N}{N{(1+e_n)}^2}\\ v={\[\frac{1}{N}Tr\left({\mathrm{\Θ}}_1{T}^2\right),...,\frac{1}{N}Tr\left({\mathrm{\Θ}}_K{T}^2\right)\]}^T\\ v_m={\[\frac{1}{N}Tr({\mathrm{\Θ}}_1{\mathrm{T\Θ}}_{m}T,...,{\mathrm{\Θ}}_K{\mathrm{T\Θ}}_{m}T)\]}^T\end{array}\right..$

Again such as, the concrete feasible implementation method of step 205 comprises the steps:

Can be specifically the virtual up-link power allocation vector q according to optimizing ^*utilize instantaneous channel state information to calculate downlink beamforming vector, obtain the downlink beamforming vector W optimized ^*, wherein, computing formula is as follows:

$w_k=\frac{{({\mathrm{\Σ}}_{\underset{m\≠k}{m=1}}^K,q_m{h}_m{h}_m^H+I)}^{-1}{h}_k}{\left|\right|{({\mathrm{\Σ}}_{\underset{m\≠k}{m=1}}^K,q_m{h}_m{h}_m^H+I)}^{-1}{h}_k\left|\right|}.$

It should be noted that, simultaneously single base stations and multiuser MIMO efficiency beam-forming method that prior art proposes utilizes instantaneous channel state information to upgrade downlink power allocation vector sum downlink beamforming vector, and its duration time interval once upgraded is τ.But, a kind of high energy efficiency of statistical channel state information extensive MIMO beam-forming method that utilizes in this paper is that single base station utilizes statistical channel state information updating downlink power allocation vector and utilizes instantaneous channel state information updating downlink beamforming vector, the duration time interval that its downlink power allocation vector once upgrades is T, the duration time interval that downlink beamforming vector once upgrades is τ, wherein T > > τ.Also can be understood as, in duration time interval T, the method carried herein only needs renewal downlink power allocation vector, and prior art needs to upgrade T/ τ downlink power allocation vector, namely illustrate that institute's extracting method obviously can reduce the renewal frequency of downlink power allocation vector herein, thus reduce computation complexity further.

A kind of performance utilizing the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information will proposed by the present invention of MATLAB simulating, verifying below.

Simulated environment optimum configurations is as follows, and the number of users of single base station service is K, base-station node configuration N transmit antennas, user node configuration single received antenna.Coverage radius of cell is 500m, user's random distribution and and distance at least 35 between base station _m.Adopt the flat fading model under the suburb non line of sight scene in 3GPP, its path loss is 38log ₁₀d ()+34.5, shadow fading obeys the logarithm normal distribution of zero-mean standard deviation 8dB, and multipath fading coefficient obeys the multiple gaussian random distribution of zero mean unit variance.The circuit-level power consumption that single antenna is fixed is P _c=30dBm, the base power consumption that single base station maintains normal work is P ₀=40dBm, and single base station to greatest extent transmitting power be P.Signal bandwidth W=10MHz, user's receiving terminal noise factor is the poor efficiency of 9dB, base station power amplifier and user rate weighted factor is set to simultaneously, a kind of method that united beam is shaped and power controls is had in prior art, its beam forming adopts high specific transmission (maximumratiotransmission, MRT) technology, power controls to adopt mean allocation strategy, using corresponding for the method user's Signal to Interference plus Noise Ratio as ownership goal Signal to Interference plus Noise Ratio

Fig. 4 is the graph of relation that in the present invention, single base station utilizes efficiency optimization target values and the power consumption factor in virtual its process of up-link power allocation optimization problems of statistical channel state information process, according to different power consumption factor-alpha ∈ [0,1] statistical channel state information iteration is utilized to upgrade virtual up-link power allocation vector, thus obtain corresponding virtual up-link power allocation vector and calculate its efficiency optimization target values, wherein, community user number K=2, base station transmit antennas number N=16.In figure, (a), (b), (c), (d) are single base station transmitting power P=26dBm to greatest extent respectively, P=34dBm, P=42dBm, (H1 is realized at four accidental channels during P=46dBm, H2, H3, H4) under efficiency optimization target values and the graph of relation of the power consumption factor.Simulation result shows, when transmission power region is 26dBm≤P < 42dBm to greatest extent, and the optimal power consumption factor-alpha that maximum efficiency optimization target values is corresponding ^*=1, also just mean and make full use of through-put power when transmission power region is 42dBm≤P≤46dBm to greatest extent, efficiency optimization target values and the power consumption factor (0≤α≤1) present the relation first increasing and reduce afterwards.Like this, there is a optimal power consumption factor 0≤α ^*≤ 1 makes efficiency optimization target values maximum, and linear search method therefore can be adopted to find α ^*, then obtain the virtual up-link power allocation vector of optimization.

Fig. 5 is the system energy efficiency performance of institute of the present invention extracting method and the graph of relation of single base station transmit antennas number, wherein, and community user number K=4, single base station transmitting power P=46dBm to greatest extent.In figure, " instantaneous CSI " curve refers to and adopts the single base stations and multiuser MIMO efficiency beam-forming method utilizing instantaneous channel state information of the prior art, the average system performance efficiency under 1000 accidental channels realize and the relation curve of single base station transmit antennas number; In figure, " statistical CSI " curve refers to that adopt the present invention to propose a kind of utilizes the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information, the average system performance efficiency under 1000 accidental channels realize and the relation curve of single base station transmit antennas number; In figure, " percentage " curve refers to " statistical CSI " and " instantaneous CSI " both ratio of average system efficiency and relation curves of single base station transmit antennas number.Simulation result shows, single base station user number is fixed (K=4), utilizes the efficiency beam-forming method of instantaneous CSI or statistical CSI all to there is best antenna for base station configuration (N ^*=8) _.when antenna for base station number N=5, the ratio of " statistical CSI " and " instantaneous CSI " both average system efficiencies is 89.46%; When antenna for base station number N=64, the ratio of the two average system efficiency is 98.68%.What also just mean that the present invention carries utilizes its system energy efficiency performance of the beam-forming method of statistical channel state information close to its system energy efficiency performance of beam-forming method utilizing instantaneous channel state information in prior art, almost can reach more than 90%.

In sum, the a kind of of the present invention's proposition utilizes the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information, be characterized in the optimization problem first utilizing the virtual up-link power of statistical channel state information process to distribute, and utilize the transfer problem of uplink downlink duality process uplink downlink power division, thus obtain the downlink power allocation vector optimized; Secondly, utilize instantaneous channel state information to calculate downlink beamforming vector, thus obtain the downlink beamforming vector optimized.The advantage of institute of the present invention extracting method is, statistical channel state information is utilized to replace the optimization problem of instantaneous channel state information processing downlink power allocation vector, reduce the operand once upgrading downlink power allocation vector, the renewal frequency of downlink power allocation vector effectively can be avoided too high simultaneously, thus reduce computation complexity further.In addition, what the present invention carried utilizes its system energy efficiency performance of the beam-forming method of statistical channel state information close to its system energy efficiency performance of beam-forming method utilizing instantaneous channel state information in prior art.Therefore, a kind of extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information that utilizes that the present invention proposes can weigh system energy efficiency performance and computation complexity.

Claims (7)

1. utilize the extensive MIMO beam-forming method of the high energy efficiency of statistical channel state information, it is characterized in that, comprising:

Obtain the statistical channel state information between single base station and collaboration user;

The optimization problem that single base station utilizes the virtual up-link power of statistical channel state information process to distribute, obtains the virtual up-link power allocation vector optimized;

According to the virtual up-link power allocation vector of described optimization, utilize uplink downlink duality, the transfer problem of process uplink downlink power division, obtain the downlink power allocation vector optimized;

Obtain the instantaneous channel state information between single base station and collaboration user;

According to the virtual up-link power allocation vector of described optimization, utilize instantaneous channel state information, calculate downlink beamforming vector, obtain the downlink beamforming vector optimized.

2. method according to claim 1, is characterized in that, the optimization problem that described single base station utilizes the virtual up-link power of statistical channel state information process to distribute comprises:

Initialization of virtual up-link power allocation vector q ⁽⁰⁾, meet target Signal to Interference plus Noise Ratio and power constraints and utilize minimum power method to obtain q ⁽⁰⁾, wherein, represent the virtual up-link power apportioning cost of collaboration user k, K represents the number of users that single base station is served;

External iteration upgrades power consumption factor-alpha ⁽ⁿ⁾, determine according to linear search method wherein, ${\mathrm{\&alpha;}}_{\min }^{\left(n\right)}<{\mathrm{\&alpha;}}_{mid1}^{\left(n\right)}<{\mathrm{\&alpha;}}_{mid2}^{\left(n\right)}<{\mathrm{\&alpha;}}_{\max }^{\left(n\right)},$ N represents external iteration number of times, initial value ${\mathrm{\&alpha;}}_{min}^{\left(0\right)}={\mathrm{\&Sigma;}}_{k=1}^K{q}_k^{\left(0\right)}/P,$ ${\mathrm{\&alpha;}}_{\max }^{\left(0\right)}=1,$ P refers to the transmission power value to greatest extent of single base station;

According to the power consumption factor of current external iteration, utilize statistical channel state information, internal layer iteration upgrades virtual up-link power allocation vector, obtains the virtual up-link power allocation vector of current external iteration, is designated as

According to the virtual up-link power allocation vector of the current external iteration of power consumption Summing Factor of current external iteration, calculate the efficiency optimization target values of current external iteration, be designated as

Shorten the region of search, if ${\mathrm{EE}}_{mid1}^{\left(n\right)}>{\mathrm{EE}}_{mid2}^{\left(n\right)},$ Then ${\mathrm{\&alpha;}}_{max}^{\left(n\right)}={\mathrm{\&alpha;}}_{mid2}^{\left(n\right)},$ $q_{\max }^{*\left(n\right)}=q_{mid2}^{*\left(n\right)},$ ${\mathrm{EE}}_{\max }^{\left(n\right)}={\mathrm{EE}}_{mid2}^{\left(n\right)},$ Otherwise, ${\mathrm{\&alpha;}}_{\min }^{\left(n\right)}={\mathrm{\&alpha;}}_{mid1}^{\left(n\right)},$ $q_{\min }^{*\left(n\right)}=q_{mid1}^{*\left(n\right)},$ ${\mathrm{EE}}_{min}^{\left(n\right)}={\mathrm{EE}}_{mid1}^{\left(n\right)};$

Judge whether to stop external iteration, if then will or as the virtual up-link power allocation vector q optimized ^*, otherwise return the execution external iteration renewal power consumption factor, wherein ζ is a predetermined threshold value.

3. method according to claim 2, is characterized in that, the described power consumption factor according to current external iteration, utilizes statistical channel state information, and internal layer iteration upgrades virtual up-link power allocation vector, comprising:

According to the power consumption factor of current external iteration, initialization of virtual up-link power allocation vector q ^(m), wherein, m represents internal layer iterations, initial value m=0;

According to the virtual up-link power allocation vector of current internal layer iteration, calculate wherein $e_k=\frac{1}{N}Tr\left({\mathrm{\&Theta;}}_{k}T\right),$ $T={(\frac{1}{N}\underset{n=1}{\overset{K}{\&Sigma;}}\frac{q_n{\mathrm{\&Theta;}}_n}{1+e_n}+I)}^{-1},$

Θ _krepresent the normalization spatial correlation matrix between single base station and collaboration user k, N represents the transmit antenna number that single base station configures;

According to the power consumption factor of current external iteration, utilize broad sense water-filling algorithm, upgrade virtual up-link power allocation vector, namely

search optimum level of water line meet power constraints obtain the virtual up-link power allocation vector q after upgrading ^(m+1), wherein, θ _kand γ _krepresent speed weighted factor and the target Signal to Interference plus Noise Ratio of collaboration user k respectively, represent the poor efficiency of single base station power amplifier, P _crepresent the circuit-level power consumption that single antenna is fixing, P ₀represent that single base station maintains the base power consumption of normal work;

Judge whether to stop internal layer iteration, if then by described virtual up-link power allocation vector q ^(m+1)as the virtual up-link power allocation vector q of current external iteration ^{* (n)}, otherwise return the virtual up-link power allocation vector calculating e performed according to current internal layer iteration.

4. method according to claim 2, is characterized in that, the virtual up-link power allocation vector of the described current external iteration of power consumption Summing Factor according to current external iteration, calculates the efficiency optimization target values of current external iteration, comprising:

According to the power consumption factor-alpha of current external iteration ⁽ⁿ⁾with the virtual up-link power allocation vector q of current external iteration ^{* (n)}, calculate the efficiency optimization target values EE of current external iteration ⁽ⁿ⁾, namely

5. method according to claim 2, is characterized in that, also comprises:

When external iteration frequency n>=1, then according to the power consumption factor of current external iteration, can utilize statistical channel state information, internal layer iteration upgrades virtual up-link power allocation vector, obtain the virtual up-link power allocation vector of current external iteration, be designated as further, according to the virtual up-link power allocation vector of the current external iteration of power consumption Summing Factor of current external iteration, calculate the efficiency optimization target values of current external iteration, be designated as

6. method according to claim 1, is characterized in that, the described virtual up-link power allocation vector according to described optimization, utilizes uplink downlink duality, and the transfer problem of process uplink downlink power division, comprising:

According to the virtual up-link power allocation vector q of described optimization ^*, utilize uplink downlink duality, the transfer problem of process uplink downlink power division, i.e. p=(A ^o) ^-11, obtain the downlink power allocation vector p optimized ^*, wherein, 1 represents that all elements is the K dimensional vector of 1, represent the downlink power allocation value of collaboration user k, $A_{k,m}^o=\left\{\begin{array}{c}\frac{|e_k{|}^2}{{\stackrel{\&LeftArrow;}{\mathrm{\&gamma;}}}_k|e_k^{\&prime;}{|}^2},k=m\\ -\frac{e_{k,m}^{\&prime;}}{{(1+e_k)}^2|e_m^{\&prime;}{|}^2},k\&NotEqual;m\end{array}\right.,$ Wherein, ${\stackrel{\&LeftArrow;}{\mathrm{\&gamma;}}}_k=q_k{e}_k,$

Make e'=[e ' ₁..., e' _k] ^t, e' _m=[e ' _{1, m}..., e' _k,m] ^t, then e'=(I _k-J) ^-1v, e' _m=(I _k-J) ^-1v _m,

Wherein, $\left\{\begin{array}{c}{\&lsqb;J\&rsqb;}_{k,n}=\frac{Tr\left({\mathrm{\&Theta;}}_k{\mathrm{Tq}}_n{\mathrm{\&Theta;}}_{n}T\right)/N}{N{(1+e_n)}^2}\\ v={\&lsqb;\frac{1}{N}Tr\left({\mathrm{\&Theta;}}_1{T}^2\right),...,\frac{1}{N}Tr\left({\mathrm{\&Theta;}}_K{T}^2\right)\&rsqb;}^T\\ v_m={\&lsqb;\frac{1}{N}Tr({\mathrm{\&Theta;}}_1{\mathrm{T\&Theta;}}_{m}T,...,{\mathrm{\&Theta;}}_K{\mathrm{T\&Theta;}}_{m}T)\&rsqb;}^T\end{array}\right..$

7. method according to claim 1, is characterized in that, the described virtual up-link power allocation vector according to described optimization, utilizes instantaneous channel state information, calculates downlink beamforming vector, comprising:

According to the virtual up-link power allocation vector q of described optimization ^*, utilize instantaneous channel state information, calculate downlink beamforming vector, obtain the downlink beamforming vector W optimized ^*, wherein, represent the downlink beamforming vector of collaboration user k, namely

$w_k=\frac{{({\mathrm{\&Sigma;}}_{\underset{m\&NotEqual;k}{m=1}}^K,q_m{h}_m{h}_m^H+I)}^{-1}{h}_k}{\left|\right|{({\mathrm{\&Sigma;}}_{\underset{m\&NotEqual;k}{m=1}}^K,q_m{h}_m{h}_m^H+I)}^{-1}{h}_k\left|\right|},$

Wherein, h _krepresent the normalization channel fading coefficient between single base station and collaboration user k.