CN104039005A - SLNR (signal-to-leakage-and-noise ratio) beam forming based user need considered power distribution method - Google Patents

Abstract

The invention discloses an SLNR (signal-to-leakage-and-noise ratio) beam forming based user need considered power distribution method, and belongs to the technical field of beam forming in a wireless communication system. The problems of low fairness among different users and low satisfaction corresponding to each user when system performance is improved by an existing beam forming method are solved. The method includes: acquiring a feasible initial power value through an initial point selection algorithm; solving an approximate convex optimization problem of an established power optimization problem, and acquiring power distribution values of the users; acquiring a temporary optimum power distribution value; updating an SLNR beam forming method to obtain a corresponding beam forming matrix, and acquiring an optimum power distribution result. Or, an expression of a corresponding power distribution matrix when the SLNR received by a user side is in minimum need is obtained by adopting suboptimization and utilizing an expression of the SLNR and matrix transformation, and a power distribution value is acquired so as to obtain a final power distribution result. System performance, fairness among the different users and the satisfaction corresponding to each user achieve best compromise.

Description

The power distribution method of the consideration user's request based on SLNR wave beam forming

Technical field

The invention belongs to the wave beam forming technical field in wireless communication system, particularly a kind of power distribution method of the consideration user's request based on SLNR wave beam forming.

Background technology

How in antagonism channel fading and interference, improving the availability of frequency spectrum, and improve power system capacity in ensureing communication quality, is problem demanding prompt solution in LTE-A system.Multiuser MIMO and wave beam forming technology can reach and strengthen desired signal and suppress inter-user interference, improve the object of message capacity and quality, and therefore, in LTE-A standard, the technology such as multiuser MIMO and beam shaping have also obtained further development and application.In the down link of TD-LTE-A, non-code book beam shaping method receives more concern owing to can utilizing the reciprocity of tdd mode lower channel to reduce the signaling consumption of channel estimating in recent years.

From Release 10 versions of 3GPP, start to support multi-user's multi-flow beam excipient, the non-code book wave beam forming of multi-user scheme is divided into the non-linear wave beam figuration scheme taking dirty paper code as representative, and linear arrangement, comprise block diagonalization, ZF, least mean-square error and SLNR (Signal-to-leakage-and-noise Ratio leaks signal to noise ratio) algorithm.Dirty paper code algorithm can reach the channel capacity upper bound of multi-user MIMO system, but it realizes extremely complexity, is difficult to realize in actual hardware.And SLNR algorithm can greatly reduce the complexity of algorithm, and compare other linear predictive coding schemes in the error rate and with the performance of capacity better.In addition, can improve the performance of wave beam forming by the power distribution algorithm in conjunction with suitable, transmitting terminal carries out power division to user terminal, can make system reach heap(ed) capacity.Power division also can meet the demand for services of user for business, such as, the channel conditions of a certain user terminal is poor, but its service rate has minimum limit value, and so now can be by distributing larger power that it is satisfied the demands for it.Have scholar to propose the water injection power allocation algorithm of consideration capacity optimum, the power that the bad user of channel condition is assigned to is just little, because water-filling algorithm is pursued optimized capacity, can abandon the poor especially user's of channel condition transmission, and this is irrational.There is scholar to propose the power distribution algorithm of channel self-adapting, considered the fairness between user's difference and the user of channel condition, but do not reach the optimization of capacity, also do not consider user's business demand.

Summary of the invention

The object of the invention is in order to solve current beam form-endowing method in improving systematic function, also there is fairness and the low problem of satisfaction corresponding to each user between different user, the invention provides a kind of power distribution method of the consideration user's request based on SLNR wave beam forming.

The present invention solves the problems of the technologies described above the technical scheme of taking to be:

The power distribution method of the consideration user's request based on SLNR wave beam forming of the present invention, provides two kinds of technical schemes:

Technical scheme one: optimized power distribution method, comprises following steps:

Step 1: under the condition of distributing at constant power, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power;

Step 2: under the wave beam formed matrix obtaining in step 1, obtain feasible power initial value by initial point selection algorithm;

Step 3: under the power initial value obtaining in step 2, solve the approximate protruding optimization problem of the power optimization problem of foundation, obtain user's power division value;

Step 4: the power distribution result that step 3 is obtained is as the initial value of step 3, repeated execution of steps three, before and after judging, whether the difference of the power division value of twice acquisition is enough little, satisfied perform step five, now, obtains temporary transient optimal power allocation value;

Step 5: under the optimal power allocation value obtaining in step 4, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power, execution step two is to step 4, if the difference of the optimization aim that front and back obtain for twice (system and capacity) value is enough little, finish algorithm, obtain final optimal power allocation result.

Technical scheme one is further qualified:

The process that in described step 1, employing obtains corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power is:

In the time that user side configurations match filtering receiver and user's transmitting power has been normalized, the SLNR of user k is expressed as:

$\begin{array}{c}{\mathrm{SLNR}}_k=\frac{{\left|\right|P_k{H}_k{w}_k\left|\right|}_F^2}{N_r{\mathrm{\σ}}^2+{\mathrm{\Σ}}_{i=1,i\≠k}^K{\left|\right|P_k{H}_i{w}_k\left|\right|}_F^2}\\ =\frac{\mathrm{trace}\left(P_k^H{w}_k^H{H}_k^H{H}_k{w}_k{P}_k\right)}{\mathrm{trace}\left(w_k^H(P_k^H{\stackrel{\‾}{H}}_k^H{\stackrel{\‾}{H}}_k{P}_k+N_r{\mathrm{\σ}}^2{I}_{N_t})w_k\right)}\end{array}$

In formula, w _kthe wave beam formed matrix of user k, N _rthe reception antenna number of user side, N _tthe number of transmit antennas of base station end, wherein K is the number of users of base station service, H _ithe channel matrix of user i, wherein I _nfor N _tthe unit matrix of dimension, p _krepresent that user k distributes the performance number obtaining, wherein the Frobenius norm of representing matrix A, the mark that trace (A) is matrix A, σ ²for the noise power of user k;

In the time that SLNR maximizes, the wave beam formed matrix of optimized user k is:

$w_k^{\mathrm{opt}}=\underset{w_k}{\arg \max }\left({\mathrm{SLNR}}_k\right).$

In step 2, the process of initial point selection algorithm is:

Step 2 one: by constant power apportioning cost initial value P the most _m, m=0, i constraints of optimization problem is made as c _i(P);

Step 2 two: calculate non-effective collection J, set J={j:c _j(P _k) >0}, if represent empty set, algorithm finishes, and obtains rational power initial value vector; Otherwise, continue execution step two or three;

Step 2 three: get j ∈ J, with P _mfor initial value, utilize interior point method to solve following problem:

min?c _j(P)?j∈J

s.t.c _i(P)≤0?

Step 2 four: if the optimal solution P of this subproblem _m+1meet c _j(P _m+1)≤0, m=m+1, execution step two or two.

The described power optimization Target Modeling in step 3 is:

$\max C\left(P\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\log (1+{\mathrm{SINR}}_i)$

s.t.SINR _i≥SINR _imin

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\≤P_T$

In formula, ${\mathrm{SINR}}_i=\left(p_i{w}_i^H{H}_i^H{H}_i{w}_i\right)/(\underset{j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w}_j^H{H}_j^H{H}_i{w}_j+{\mathrm{\σ}}^2),$ Optimization aim be make system reach maximization with capacity, top n constraints represents that the SINR value that user i receives is greater than the minimum SINR limit value of user i SINR _imin, last constraints limit is distributed to each user's power and must not be exceeded the transmission power level of base station place maximum, wherein P _tfor the gross power of base station transmit antennas; One total N+1 restrictive condition;

For target function, maximize be equal to maximization due to SINR _ibe greater than zero, therefore maximize be equivalent to minimize target function is converted into:

$\begin{array}{c}F\left(P\right)=\underset{i=1}{\overset{N}{\mathrm{\Π}}}\frac{1}{1+{\mathrm{SINR}}_i}\\ =\underset{i=1}{\overset{N}{\mathrm{\Π}}}\left(\frac{\underset{j\≠i}{\mathrm{\Σ}}{w}_j^H{H}_j^H{H}_i{w}_j{p}_j+{\mathrm{\σ}}^2}{\underset{j\∈U}{\mathrm{\Σ}}{w}_j^H{H}_i^H{H}_i{w}_j{p}_j+{\mathrm{\σ}}^2}\right)=\underset{i=1}{\overset{N}{\mathrm{\Π}}}\left(\frac{\underset{j\≠i}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2}{\underset{j\∈U}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2}\right)\end{array}$

In formula, u={1,2 ..., N}; F (P) is approximately:

$F\left(P\right)\≈d^{\left(L\right)}{p}_1^{{\mathrm{\β}}_1^{\left(L\right)}}...p_N^{{\mathrm{\β}}_N^{\left(L\right)}}$

In formula, ${\mathrm{\β}}_i^{\left(L\right)}=\frac{p_i^{(L-1)}}{F\left(P^{(L-1)}\right)}\frac{\∂F\left(P^{(L-1)}\right)}{\∂p_i^{(L-1)}},d^{\left(L\right)}=\frac{F\left(P^{(L-1)}\right)}{p_1^{(L-1){\mathrm{\β}}_1^{\left(L\right)}}...p_N^{(L-1){\mathrm{\β}}_N^{\left(L\right)}}};$ L represents iterations, and approximate function can approach original target function in certain iterations, obtains the new target function that can be solved;

Carry out the conversion of form for constraints, target function and constraints removed to logarithm, obtain the approximate protruding optimization problem of the power optimization problem described in step 2:

$\min \log (d^{\left(L\right)}{p}_1^{{\mathrm{\β}}_1^{\left(L\right)}}...p_N^{{\mathrm{\β}}_N^{\left(L\right)}})$

$s.t.\frac{{\mathrm{SINR}}_{i\min }}{{\mathrm{SINR}}_i}=\log \left(\frac{{\mathrm{SINR}}_{i\min }}{G_{\mathrm{ii}}{p}_i}(\underset{j\≠i}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2)\right)\≤0$

$\log \left(\frac{1}{P_T}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\right)\≤0$

Above formula is the convex formula of geometric programming, and geometric programming can utilize interior point method to solve, and obtains power division value.

Describedly perform step five condition judging whether described in step 4 and be:

$|p_i^{\left(L\right)}-p_i^{(L-1)}|\≤\mathrm{\ϵ},\∀i;$

In formula, represent the power of the user i that the L time approximate iteration obtain.

The described condition finishing in the evaluation algorithm described in step 5 be:

|C(P _n)-C(P _n-1)|≤ε ₂；

In formula, P _nrepresent to repeat for the n time the power distribution result that the step 2 to four in claim 2 obtains, C (P _n) to be illustrated in power division matrix be P _ntime and capacity.

Technical scheme two: also proposed the power distribution method of suboptimization in order to reduce complexity, comprised following steps:

Step 1: under the condition of distributing at constant power, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power;

Step 2: utilize expression formula and the matrixing of SINR, when the SINR that obtains receiving when user side is minimum SINR demand, the expression formula of corresponding power division matrix, under the wave beam formed matrix obtaining, and then obtains the power division value of just meeting consumers' demand in step 1.

Step 3: by remaining power proportionally the factor distribute, obtain final power distribution result.

Technical scheme two is further qualified:

When the SINR receiving when user side in described step 2 is minimum SINR demand, acquisition user power expression formula is:

$p_i=\frac{\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w_j}^H{H_i}^H{H}_i{w}_j{\mathrm{SINR}}_{i\min }+{\mathrm{\σ}}^2{\mathrm{SINR}}_{i\min }}{{w_i}^H{H_i}^H{H}_i{w}_i}$

For simplifying expression formula, introduce matrix A and B, the method that obtains A and B is:

$A_{\mathrm{ij}}=\left\{\begin{array}{cc}{w_j}^H{H_j}^H{H}_i{w}_j& i\≠j\\ 0& i=j\end{array}\right.i,j=\mathrm{1,2,3},...,K$

$B_{\mathrm{ij}}=\left\{\begin{array}{cc}0& i\≠j\\ \frac{{\mathrm{SINR}}_{i\min }}{{w_i}^H{H_i}^H{H}_i{w}_i}& i=j\end{array}\right.i,j=\mathrm{1,2,3},...,K$

User power expression formula is converted into expression matrix form:

In formula, P=[p ₁, p ₂..., p _n] ^t, it is N × 1 dimension matrix; Through simple matrixing, power division matrix notation is:

In formula, P ⁽¹⁾power division matrix while being minimum SINR demand for the SINR receiving when user side, I is that N × N ties up unit matrix.

The formula that scale factor described in step 3 obtains is:

${\mathrm{\ρ}}_i=\frac{P_T}{\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{p}_i}i=\mathrm{1,2},...,K$

The formula that obtains final power division matrix is:

$P_{\mathrm{all}}=\mathrm{\ρ}{P}^{\left(1\right)}=(P_T/\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{p}_i)P^{\left(1\right)}$

In formula, ρ=[ρ ₁, ρ ₂..., ρ _k] be equitable proportion factor matrix.

The present invention has following beneficial effect:

Effect of the present invention is, optimum iterative power allocation algorithm can elevator system and demand and matched beam figuration matrix and the power division matrix of capacity, guarantee customer service.Suboptimization algorithm, with respect to optimum allocation algorithm, reduces complexity greatly, and capacity only has 10% decline, can meet consumers' demand preferably.In other words, the present invention, in improving systematic function, can take into account fairness and satisfaction corresponding to each user between different user, makes fairness and satisfaction corresponding to each user between systematic function, different user reach optimal compromise.

Because SLNR beamforming algorithm has the complexity that can greatly reduce algorithm, and compare the better advantage of other linear predictive coding schemes in the error rate and with the performance of capacity, the present invention is based on this algorithm research power distribution algorithm, owing to distributing to certain user's the not only communication quality of impact of power itself, and affect its interference to other, the performance of wave beam forming can get a promotion by the power division to user like this, and good power distribution strategies can elevator system with capacity and ensure user's demand.

That optimum iterative algorithm in the present invention can maximize system under the prerequisite of meeting consumers' demand and capacity, in addition, consider that the result of power division can cause the change of wave beam formed matrix, in the present invention, design the iterative process of upgrading wave beam formed matrix, wave beam formed matrix and power division matrix are matched.Meanwhile, in order to reduce the complexity in physical device, the present invention proposes suboptimization power distribution algorithm.

Brief description of the drawings

Fig. 1 is the principle schematic of the power distribution method based on SLNR wave beam forming described in embodiment one;

Fig. 2 be power distribution method of the present invention and existing power distribution algorithm with capacity comparison curve synoptic diagram;

Fig. 3 be the user that is not satisfied of the removal demand of power distribution method of the present invention and existing power distribution algorithm with capacity comparison curve synoptic diagram;

Fig. 4 is the average interrupt probability correlation curve schematic diagram of power distribution method of the present invention and existing power distribution algorithm;

Fig. 5 is the contrast block diagram of the SINR that receives of the user of power distribution method of the present invention and existing power distribution algorithm.

Embodiment

Embodiment one: in conjunction with Fig. 1, present embodiment is described, the power distribution method of the consideration user's request based on SLNR wave beam forming described in present embodiment, wherein optimum iterative power distribution method comprises the steps:

Step 1: under the condition of distributing at constant power, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power;

Step 2: under the wave beam formed matrix obtaining in step 1, obtain feasible power initial value by initial point selection algorithm;

Step 3: under the power initial value obtaining in step 2, solve the approximate protruding optimization problem of the power optimization problem of foundation, obtain user's power division value;

Step 4: the power distribution result that step 3 is obtained is as the initial value of step 3, repeated execution of steps three, before and after judging, whether the difference of the power division value of twice acquisition is enough little, satisfied perform step five, now, obtains temporary transient optimal power allocation value;

Step 5: under the optimal power allocation value obtaining in step 4, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power, execution step two is to step 4, if the difference of the optimization aim that front and back obtain for twice (system and capacity) value is enough little, finish algorithm, obtain final optimal power allocation result.

Embodiment two: present embodiment is the further restriction of the power distribution method to the consideration user's request based on SLNR wave beam forming described in embodiment one, and the method that the SLNR beam form-endowing method of the combination user power adopting in described step 1 obtains corresponding wave beam formed matrix is:

The SLNR of user k is expressed as:

$\begin{array}{c}{\mathrm{SLNR}}_k=\frac{{\left|\right|P_k{H}_k{w}_k\left|\right|}_F^2}{N_r{\mathrm{\σ}}^2+{\mathrm{\Σ}}_{i=1,i\≠k}^K{\left|\right|P_k{H}_i{w}_k\left|\right|}_F^2}\\ =\frac{\mathrm{trace}\left(P_k^H{w}_k^H{H}_k^H{H}_k{w}_k{P}_k\right)}{\mathrm{trace}\left(w_k^H(P_k^H{\stackrel{\‾}{H}}_k^H{\stackrel{\‾}{H}}_k{P}_k+N_r{\mathrm{\σ}}^2{I}_{N_t})w_k\right)}\end{array}$

In formula, w _kthe wave beam formed matrix of user k, N _rthe reception antenna number of user side, N _tthe number of transmit antennas of base station end, wherein K is the number of users of base station service, H _ithe channel matrix of user i, wherein for N _tthe unit matrix of dimension, p _krepresent that user k distributes the performance number obtaining, wherein the Frobenius norm of representing matrix A, the mark that trace (A) is matrix A, σ ²for the noise power of user k;

In the time that SLNR maximizes, the wave beam formed matrix of optimized user k is:

$w_k^{\mathrm{opt}}=\underset{w_k}{\arg \max }\left({\mathrm{SLNR}}_k\right).$

Embodiment three: present embodiment is the further restriction of the optimum iterative power distribution method to the consideration user's request based on SLNR wave beam forming described in embodiment one or two,

In step 2, the step of initial point system of selection is:

Step 1: by constant power apportioning cost initial value P the most _m, m=0, i constraints of optimization problem is made as c _i(P).

Step 2: calculate non-effective collection J, set J={j:c _j(P _k) >0}, if algorithm finishes, and obtains rational power initial value vector.Otherwise, continue execution step three;

Step 3: get j ∈ J, with P _mfor initial value, utilize interior point method to solve following problem:

min?c _j(P)?j∈J

s.t.c _i(P)≤0?

Step 4: if the optimal solution P of this subproblem _m+1meet c _j(P _m+1)≤0, m=m+1, execution step two.

Embodiment four: present embodiment is the further restriction of the optimum iterative power distribution method to the consideration user's request based on SLNR wave beam forming described in embodiment one, two or three, is modeled as in the power optimization problem of step 3:

$\max C\left(P\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\log (1+{\mathrm{SINR}}_i)$

s.t.SINR _i≥SINR _imin

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\≤P_T$

In formula, ${\mathrm{SINR}}_i=\left(p_i{w}_i^H{H}_i^H{H}_i{w}_i\right)/(\underset{j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w}_j^H{H}_j^H{H}_i{w}_j+{\mathrm{\σ}}^2),$ Optimization aim be make system reach maximization with capacity, top n constraints represents that the SINR value that user i receives is greater than the minimum SINR limit value of user i SINR _imin, last constraints limit is distributed to each user's power and must not be exceeded the transmission power level of base station place maximum, wherein P _tfor the gross power of base station transmit antennas.One total N+1 restrictive condition.

Above-mentioned optimization problem is Nonlinear Nonconvex optimization problem, is difficult to obtain globally optimal solution, therefore adopts approximate method to be translated into protruding optimization problem.

For target function, maximize be equal to maximization due to SINR _ibe greater than zero, therefore maximize be equivalent to minimize target function is converted into:

$\begin{array}{c}F\left(P\right)=\underset{i=1}{\overset{N}{\mathrm{\Π}}}\frac{1}{1+{\mathrm{SINR}}_i}\\ =\underset{i=1}{\overset{N}{\mathrm{\Π}}}\left(\frac{\underset{j\≠i}{\mathrm{\Σ}}{w}_j^H{H}_j^H{H}_i{w}_j{p}_j+{\mathrm{\σ}}^2}{\underset{j\∈U}{\mathrm{\Σ}}{w}_j^H{H}_i^H{H}_i{w}_j{p}_j+{\mathrm{\σ}}^2}\right)=\underset{i=1}{\overset{N}{\mathrm{\Π}}}\left(\frac{\underset{j\≠i}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2}{\underset{j\∈U}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2}\right)\end{array}$

In formula, u={1,2 ..., N}.F (P) is approximately:

$F\left(P\right)\≈d^{\left(L\right)}{p}_1^{{\mathrm{\β}}_1^{\left(L\right)}}...p_N^{{\mathrm{\β}}_N^{\left(L\right)}}$

In formula, ${\mathrm{\β}}_i^{\left(L\right)}=\frac{p_i^{(L-1)}}{F\left(P^{(L-1)}\right)}\frac{\∂F\left(P^{(L-1)}\right)}{\∂p_i^{(L-1)}},d^{\left(L\right)}=\frac{F\left(P^{(L-1)}\right)}{p_1^{(L-1){\mathrm{\β}}_1^{\left(L\right)}}...p_N^{(L-1){\mathrm{\β}}_N^{\left(L\right)}}}.$ L represents iterations, and approximate function can approach original target function in certain iterations, so just can obtain the new target function that can be solved.

Carry out the conversion of form for constraints, target function and constraints removed to logarithm, obtain the approximate protruding optimization problem of the power optimization problem described in step 2:

$\min \log (d^{\left(L\right)}{p}_1^{{\mathrm{\β}}_1^{\left(L\right)}}...p_N^{{\mathrm{\β}}_N^{\left(L\right)}})$

$s.t.\frac{{\mathrm{SINR}}_{i\min }}{{\mathrm{SINR}}_i}=\log \left(\frac{{\mathrm{SINR}}_{i\min }}{G_{\mathrm{ii}}{p}_i}(\underset{j\≠i}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2)\right)\≤0$

$\log \left(\frac{1}{P_T}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\right)\≤0$

Above formula is the convex formula of geometric programming, and geometric programming can utilize interior point method to solve, and obtains power division value.

Embodiment five: present embodiment is the further restriction of the optimum iterative power distribution method to the consideration user's request based on SLNR wave beam forming described in embodiment one, two, three or four, five the condition of performing step judging whether described in step 4 is:

$|p_i^{\left(L\right)}-p_i^{(L-1)}|\≤\mathrm{\ϵ},\∀i;$

Embodiment six: present embodiment is the further restriction of the power distribution method to the consideration user's request based on SLNR wave beam forming described in embodiment one, two, three, four or five, and the condition that the evaluation algorithm described in step 5 finishes is:

|C(P _n)-C(P _n-1)|≤ε ₂

In formula, P _nrepresent to repeat for the n time the power distribution result that the step 2 to four in execution mode one obtains, C (P _n) to be illustrated in power division matrix be P _ntime and capacity.

Embodiment seven: in conjunction with Fig. 1, present embodiment is described, the power distribution method of the consideration user's request based on SLNR wave beam forming described in present embodiment, wherein suboptimization power distribution method comprises the steps:

Step 1: under the condition of distributing at constant power, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power;

Step 2: utilize expression formula and the matrixing of SINR, when the SINR that obtains receiving when user side is minimum SINR demand, the expression formula of corresponding power division matrix, under the wave beam formed matrix obtaining, and then obtains the power division value of just meeting consumers' demand in step 1.

Step 3: by remaining power proportionally the factor distribute, obtain final power distribution result.

Embodiment eight: present embodiment is the further restriction of the power division second-rate optimization method to the consideration user's request based on SLNR wave beam forming described in embodiment seven, when the SINR receiving when user side in step 2 is minimum SINR demand, the method that obtains user power expression formula is:

$p_i=\frac{\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w_j}^H{H_i}^H{H}_i{w}_j{\mathrm{SINR}}_{i\min }+{\mathrm{\σ}}^2{\mathrm{SINR}}_{i\min }}{{w_i}^H{H_i}^H{H}_i{w}_i}$

For simplifying expression formula, introduce matrix A and B, the method that obtains A and B is:

$A_{\mathrm{ij}}=\left\{\begin{array}{cc}{w_j}^H{H_j}^H{H}_i{w}_j& i\≠j\\ 0& i=j\end{array}\right.i,j=\mathrm{1,2,3},...,K$

$B_{\mathrm{ij}}=\left\{\begin{array}{cc}0& i\≠j\\ \frac{{\mathrm{SINR}}_{i\min }}{{w_i}^H{H_i}^H{H}_i{w}_i}& i=j\end{array}\right.i,j=\mathrm{1,2,3},...,K$

Like this, user power expression formula is converted into the method for expression matrix form and is:

In formula, P=[p ₁, p ₂..., p _n] ^t, it is N × 1 dimension matrix.Power division matrix can be expressed as:

Embodiment nine: present embodiment is the further restriction of the power division second-rate optimization method to the consideration user's request based on SLNR wave beam forming described in embodiment seven or eight, the method that the scale factor described in step 3 obtains is:

${\mathrm{\ρ}}_i=\frac{P_T}{\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{p}_i}i=\mathrm{1,2},...,K$

The method that obtains final power division matrix is:

$P_{\mathrm{all}}=\mathrm{\ρ}{P}^{\left(1\right)}=(P_T/\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{p}_i)P^{\left(1\right)}$

In formula, ρ=[ρ ₁, ρ ₂..., ρ _k].

For making object of the present invention, technological means and advantage clearer, below in conjunction with accompanying drawing, the present invention is described in further detail.

Embodiment:

Fig. 1 is the principle schematic that multiuser downstream wave beam forming joint Power is distributed, in the present invention, setting channel condition information (CSI, Channel State Information) is known at eNodeB end, carries out power division according to the result of wave beam forming and CSI.Total number of users of system is set to K, and the number of transmit antennas of eNodeB is Nt, and the reception antenna number of UE is Nr.At eNodeB end, all users' data, by string modular converter, send to UE via beam shaping module and power division module.H _kbe the channel matrix between user k and eNodeB, in the time of accidental channel that channel is flat fading, its element is obeyed the multiple Gaussian Profile of zero-mean, unit variance; w _kthe wave beam formed matrix of user k, p _kthe performance number of distributing to family k, n _kthat zero-mean, variance are σ ²additive white Gaussian noise; the signal transmission of power normalization user k afterwards, m _kfor the data fluxion of user k..

In the present invention, suppose that transmission mode is single-stream transmission, receiving terminal, the reception signal of user k can be expressed as:

$r_k=\sqrt{p_k}{H}_k{w}_k{S}_k+\underset{j=1,j\≠k}{\overset{K}{\mathrm{\Σ}}}\sqrt{p_j}{H}_k{w}_j{S}_j+n_k$

Can find out, receive signal r _kin except the desired signal of user k, also exist the interference of other user to users k, if can not eliminate or reduce this interference, can greatly reduce the Signal to Interference plus Noise Ratio of receiving terminal, and then cause the decline of systematic function.First, the corresponding wave beam formed matrix of application SLNR beam form-endowing method primary Calculation.In the time of UE end configurations match filtering receiver, the SLNR of user k can be expressed as:

$\begin{array}{c}{\mathrm{SLNR}}_k=\frac{{\left|\right|P_k{H}_k{w}_k\left|\right|}_F^2}{N_r{\mathrm{\σ}}^2+{\mathrm{\Σ}}_{i=1,i\≠k}^K{\left|\right|P_k{H}_i{w}_k\left|\right|}_F^2}\\ =\frac{\mathrm{trace}\left(P_k^H{w}_k^H{H}_k^H{H}_k{w}_k{P}_k\right)}{\mathrm{trace}\left(w_k^H(P_k^H{\stackrel{\‾}{H}}_k^H{\stackrel{\‾}{H}}_k{P}_k+N_r{\mathrm{\σ}}^2{I}_{N_t})w_k\right)}\end{array}$

In formula, w _kthe wave beam formed matrix of user k, N _rthe reception antenna number of user side, N _tthe number of transmit antennas of base station end, wherein K is the number of users of base station service, H _ithe channel matrix of user i, wherein for N _tthe unit matrix of dimension, p _krepresent that user k distributes the performance number obtaining, wherein the Frobenius norm of representing matrix A, the mark that trace (A) is matrix A, σ ²for the noise power of user k.SLNR beamforming algorithm is to maximize above formula as target, and wave beam formed matrix can be obtained by formula below:

$w_k^{\mathrm{opt}}=\underset{w_k}{\arg \max }\left({\mathrm{SLNR}}_k\right)$

Can find by above formula, in the time adopting SLNR algorithm, the optimal beam figuration matrix of user k choose that only the wave beam formed matrix with itself is relevant, and there is no direct relation with other users' wave beam formed matrix.Existing document is verified, being solved to of wave beam formed matrix:

$w_k^{\mathrm{opt}}\∝\max \mathrm{eigenvector}\left({({\stackrel{\‾}{H}}_k^H{\stackrel{\‾}{H}}_k+N_r{\mathrm{\σ}}^2{I}_{N_t}/p_k)}^{-1}{H}_k^H{H}_k\right)$

According to wave beam formed matrix and power distribution result, calculate each user's SINR, specific formula for calculation is as follows:

${\mathrm{SINR}}_i=\frac{p_i{w_i}^H{H_i}^H{H}_i{w}_i}{\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w}_j{w_j}^H{H_i}^H{H}_i{w}_j+{\mathrm{\σ}}^2}$

System as follows with computing formula capacity, wherein B is system bandwidth:

$C=B\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\log }_2(1+{\mathrm{SINR}}_i)$

In Fig. 2,3,4 emulation, the minimum SINR demand of each user equates, its settings under unjust SNR are as shown in table 1:

Table 1 is minimum SINR demand under different SNR

SNR(dB)	0	5	10	15	20	25
							SINR min(dB)	-0.97	1.80	4.77	7.00	12.05	15.56

According to the parameter request of LTE system, simulation parameter arranges as follows: the transmitting antenna sum of eNB is set to 4, user side configuration single antenna, the total number of users K=4 of intrasystem work; System bandwidth is 1.4MHz; System adopts extended cyclic prefix, i.e. CP=32; Channel is the accidental channel of smooth multiple Gaussian Profile.

Fig. 2 has provided system and the capacity comparison figure of power distribution method of the present invention and existing power distribution algorithm, can find out, water-filling algorithm with volumetric properties optimum, it is that to sacrifice fairness and user's request between user be that cost reaches higher and capacity, optimum iterative power allocation algorithm in the present invention comes second, in the time that SNR is greater than 10dB, the two with capacity be close, in addition, the suboptimization algorithm and the channel self-adapting power distribution algorithm that propose than the present invention, have respectively 10% and 15% performance boost.

Fig. 3 provided user that the removal demand of power distribution method of the present invention and existing power distribution algorithm is not satisfied with capacity comparison figure, can find out, in the time that SNR is greater than 7dB, the best performance of the optimum iterative power allocation algorithm in the present invention.Comparison diagram 2, water-filling algorithm and channel self-adapting power distribution algorithm are owing to having ignored user's request, performance significantly reduces, and for example, works as SNR=15dB, water-filling algorithm be reduced to 21Mbps with capacity by 26Mbps.In the performance of two algorithms provided by the invention and figure bis-, be close, because the present invention has considered user's request and then promoted fairness between user.

Fig. 4 has provided the average interrupt probability comparison diagram of power distribution method of the present invention and existing power distribution algorithm, when the SINR receiving is during lower than minimum SINR requirements, interrupts occurring.Can find out, the outage probability of two algorithms provided by the invention is all lower than 0.1, and channel self-adapting power distribution algorithm and water-filling algorithm are respectively higher than 0.4 and 0.3.

Fig. 5 has provided user and has had under different minimum SINR demands, the comparison diagram of the SINR that the user of power distribution method of the present invention and existing power distribution algorithm receives, can find out, two algorithms provided by the invention can meet each user's minimum SINR restriction, and for UE4, the SINR of water-filling algorithm is much smaller than SINR _min, channel self-adapting power distribution algorithm cannot meet the demand of UE1 and UE3.

Can be found out by above embodiment, the power distribution method of the consideration user's request based on SLNR wave beam forming of the present invention can reach good laser propagation effect.Wherein, optimum iterative power allocation algorithm is demand and matched beam figuration matrix and the power division matrix of elevator system and capacity, guarantee customer service effectively, suboptimization algorithm is with respect to optimum allocation algorithm, greatly reduce complexity, only there is 10% decline with capacity, can meet consumers' demand preferably.

The present invention also can have other various embodiments; in the situation that not deviating from spirit of the present invention and essence thereof; those skilled in the art are when making according to the present invention various corresponding changes and distortion, but these corresponding changes and distortion all should belong to the protection range of the appended claim of the present invention.

Claims (9)

1. a power distribution method for the consideration user's request based on SLNR wave beam forming, is characterized in that, the implementation procedure of described method is:

Step 1: under the condition of distributing at constant power, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power;

Step 2: under the wave beam formed matrix obtaining in step 1, obtain feasible power initial value by initial point selection algorithm;

Step 3: under the power initial value obtaining in step 2, solve the approximate protruding optimization problem of the power optimization target of foundation, obtain user's power division value;

Step 4: the power distribution result that step 3 is obtained is as the initial value of step 3, repeated execution of steps three, and before and after judging, whether the difference of the power division value of twice acquisition is enough little, satisfied perform step five, now, obtain temporary transient optimal power allocation value;

Step 5: under the optimal power allocation value obtaining in step 4, adopt SLNR beam form-endowing method to obtain corresponding wave beam formed matrix, execution step two is to step 4, if the difference of the optimization aim system that front and back obtain for twice and capability value is enough little, finish algorithm, obtain final optimal power allocation result.

2. the optimum iterative power distribution method of the consideration user's request based on SLNR wave beam forming according to claim 1, it is characterized in that, the process that in described step 1, employing obtains corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power is:

In the time that user side configurations match filtering receiver and user's transmitting power has been normalized, the SLNR of user k is expressed as:

$\begin{array}{c}{\mathrm{SLNR}}_k=\frac{{\left|\right|P_k{H}_k{w}_k\left|\right|}_F^2}{N_r{\mathrm{\σ}}^2+{\mathrm{\Σ}}_{i=1,i\≠k}^K{\left|\right|P_k{H}_i{w}_k\left|\right|}_F^2}\\ =\frac{\mathrm{trace}\left(P_k^H{w}_k^H{H}_k^H{H}_k{w}_k{P}_k\right)}{\mathrm{trace}\left(w_k^H(P_k^H{\stackrel{\‾}{H}}_k^H{\stackrel{\‾}{H}}_k{P}_k+N_r{\mathrm{\σ}}^2{I}_{N_t})w_k\right)}\end{array}$

In formula, w _kthe wave beam formed matrix of user k, N _rthe reception antenna number of user side, N _tthe number of transmit antennas of base station end, wherein K is the number of users of base station service, H _ithe channel matrix of user i, wherein I _nfor N _tthe unit matrix of dimension, p _krepresent that user k distributes the performance number obtaining, wherein the Frobenius norm of representing matrix A, the mark that trace (A) is matrix A, σ ²for the noise power of user k;

In the time that SLNR maximizes, the wave beam formed matrix of optimized user k is:

$w_k^{\mathrm{opt}}=\underset{w_k}{\arg \max }\left({\mathrm{SLNR}}_k\right).$

3. the optimum iterative power distribution method of the consideration user's request based on SLNR wave beam forming according to claim 1 and 2, is characterized in that, in step 2, the process of initial point selection algorithm is:

Step 2 one: by constant power apportioning cost initial value P the most _m, m=0, i constraints of optimization problem is made as c _i(P);

Step 2 two: calculate non-effective collection J, set J={j:c _j(P _k) >0}, if represent empty set, algorithm finishes, and obtains rational power initial value vector; Otherwise, continue execution step two or three;

Step 2 three: get j ∈ J, with P _mfor initial value, utilize interior point method to solve following problem:

min?c _j(P)?j∈J

s.t.c _i(P)≤0?

Step 2 four: if the optimal solution P of this subproblem _m+1meet c _j(P _m+1)≤0, m=m+1, execution step two or two.

4. the optimum iterative power distribution method of the consideration user's request based on SLNR wave beam forming according to claim 3, is characterized in that, the described power optimization Target Modeling in step 3 is:

$\max C\left(P\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\log (1+{\mathrm{SINR}}_i)$

s.t.SINR _i≥SINR _imin

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\≤P_T$

In formula, ${\mathrm{SINR}}_i=\left(p_i{w}_i^H{H}_i^H{H}_i{w}_i\right)/(\underset{j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w}_j^H{H}_j^H{H}_i{w}_j+{\mathrm{\σ}}^2),$ Optimization aim be make system reach maximization with capacity, top n constraints represents that the SINR value that user i receives is greater than the minimum SINR limit value of user i SINR _imin, last constraints limit is distributed to each user's power and must not be exceeded the transmission power level of base station place maximum, wherein P _tfor the gross power of base station transmit antennas; One total N+1 restrictive condition;

For target function, maximize be equal to maximization due to SINR _ibe greater than zero, therefore maximize be equivalent to minimize target function is converted into:

$\begin{array}{c}F\left(P\right)=\underset{i=1}{\overset{N}{\mathrm{\Π}}}\frac{1}{1+{\mathrm{SINR}}_i}\\ =\underset{i=1}{\overset{N}{\mathrm{\Π}}}\left(\frac{\underset{j\≠i}{\mathrm{\Σ}}{w}_j^H{H}_j^H{H}_i{w}_j{p}_j+{\mathrm{\σ}}^2}{\underset{j\∈U}{\mathrm{\Σ}}{w}_j^H{H}_i^H{H}_i{w}_j{p}_j+{\mathrm{\σ}}^2}\right)=\underset{i=1}{\overset{N}{\mathrm{\Π}}}\left(\frac{\underset{j\≠i}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2}{\underset{j\∈U}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2}\right)\end{array}$

In formula, u={1,2 ..., N}; F (P) is approximately:

$F\left(P\right)\≈d^{\left(L\right)}{p}_1^{{\mathrm{\β}}_1^{\left(L\right)}}...p_N^{{\mathrm{\β}}_N^{\left(L\right)}}$

In formula, ${\mathrm{\β}}_i^{\left(L\right)}=\frac{p_i^{(L-1)}}{F\left(P^{(L-1)}\right)}\frac{\∂F\left(P^{(L-1)}\right)}{\∂p_i^{(L-1)}},d^{\left(L\right)}=\frac{F\left(P^{(L-1)}\right)}{p_1^{(L-1){\mathrm{\β}}_1^{\left(L\right)}}...p_N^{(L-1){\mathrm{\β}}_N^{\left(L\right)}}};$ L represents iterations, and approximate function can approach original target function in certain iterations, obtains the new target function that can be solved;

Carry out the conversion of form for constraints, target function and constraints removed to logarithm, obtain the approximate protruding optimization problem of the power optimization problem described in step 2:

$\min \log (d^{\left(L\right)}{p}_1^{{\mathrm{\β}}_1^{\left(L\right)}}...p_N^{{\mathrm{\β}}_N^{\left(L\right)}})$

$s.t.\frac{{\mathrm{SINR}}_{i\min }}{{\mathrm{SINR}}_i}=\log \left(\frac{{\mathrm{SINR}}_{i\min }}{G_{\mathrm{ii}}{p}_i}(\underset{j\≠i}{\mathrm{\Σ}}{G}_{\mathrm{ij}}{p}_j+{\mathrm{\σ}}^2)\right)\≤0$

$\log \left(\frac{1}{P_T}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\right)\≤0$

Above formula is the convex formula of geometric programming, and geometric programming can utilize interior point method to solve, and obtains power division value.

5. according to the optimum iterative power distribution method of the consideration user's request based on SLNR wave beam forming described in claim 1 or 4, it is characterized in that, describedly perform step five condition judging whether described in step 4 and be:

$|p_i^{\left(L\right)}-p_i^{(L-1)}|\≤\mathrm{\ϵ},\∀i;$

In formula, represent the power of the user i that the L time approximate iteration obtain.

6. the optimum iterative power distribution method of the consideration user's request based on SLNR wave beam forming according to claim 5, is characterized in that, the described condition finishing in the evaluation algorithm described in step 5 be:

|C(P _n)-C(P _n-1)|≤ε ₂；

In formula, P _nrepresent to repeat for the n time the power distribution result that the step 2 to four in claim 2 obtains, C (P _n) to be illustrated in power division matrix be P _ntime and capacity.

7. a power distribution method for the consideration user's request based on SLNR wave beam forming, is characterized in that, the implementation procedure of described method is:

Step 1: under the condition of distributing at constant power, adopt and obtain corresponding wave beam formed matrix in conjunction with the SLNR beam form-endowing method of user power;

Step 2: utilize expression formula and the matrixing of SINR, when the SINR that obtains receiving when user side is minimum SINR demand, the expression formula of corresponding power division matrix, under the wave beam formed matrix obtaining, and then obtains the power division value of just meeting consumers' demand in step 1;

Step 3: by remaining power proportionally the factor distribute, obtain final power distribution result.

8. the suboptimization power distribution method of the consideration user's request based on SLNR wave beam forming according to claim 7, is characterized in that, when the SINR receiving when user side in described step 2 is minimum SINR demand, acquisition user power expression formula is:

$p_i=\frac{\underset{j=1,j\≠i}{\overset{K}{\mathrm{\Σ}}}{p}_j{w_j}^H{H_i}^H{H}_i{w}_j{\mathrm{SINR}}_{i\min }+{\mathrm{\σ}}^2{\mathrm{SINR}}_{i\min }}{{w_i}^H{H_i}^H{H}_i{w}_i}$

For simplifying expression formula, introduce matrix A and B, the method that obtains A and B is:

$A_{\mathrm{ij}}=\left\{\begin{array}{cc}{w_j}^H{H_j}^H{H}_i{w}_j& i\≠j\\ 0& i=j\end{array}\right.i,j=\mathrm{1,2,3},...,K$

$B_{\mathrm{ij}}=\left\{\begin{array}{cc}0& i\≠j\\ \frac{{\mathrm{SINR}}_{i\min }}{{w_i}^H{H_i}^H{H}_i{w}_i}& i=j\end{array}\right.i,j=\mathrm{1,2,3},...,K$

User power expression formula is converted into expression matrix form:

In formula, P=[p ₁, p ₂..., p _n] ^t, it is N × 1 dimension matrix; Through simple matrixing, power division matrix notation is:

In formula, P ⁽¹⁾power division matrix while being minimum SINR demand for the SINR receiving when user side, I is that N × N ties up unit matrix.

9. according to the suboptimization power distribution method of the consideration user's request based on SLNR wave beam forming described in claim 7 or 8, it is characterized in that, the formula that the scale factor described in step 3 obtains is:

${\mathrm{\ρ}}_i=\frac{P_T}{\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{p}_i}i=\mathrm{1,2},...,K$

The formula that obtains final power division matrix is:

$P_{\mathrm{all}}=\mathrm{\ρ}{P}^{\left(1\right)}=(P_T/\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{p}_i)P^{\left(1\right)}$

In formula, ρ=[ρ ₁, ρ ₂..., ρ _k] be equitable proportion factor matrix.