CN101895892A - Multi-district dynamic limiting feedback method in LTE-A (Long Term Evolution-Advanced) - Google Patents

Abstract

The invention discloses a multi-district dynamic limiting feedback method in LTE-A (Long Term Evolution-Advanced), mainly aiming at overcoming the defects of no full consideration of practical requirements of different users and poor performance of users of district edges in the traditional feedback method. The multi-district dynamic limiting feedback method comprises the steps: establishing optimization models among districts, and solving the optimization models among the districts to allocate feedback bits to each district; establishing optimization models in the districts, and solving the optimization models in the districts to allocate feedback bits of each district to each user of the district; and finally reallocating the feedback bits of each district to each user when the position of the user in the district changes. The invention can meet the practical requirements of each user, greatly improve the performance of the users of district edges, avoid the problem of being unable to obtain good service by the poorest user in a practical system and well ensure the fairness of users. The invention can be used for communication between base stations of CoMP and users in the field of mobile communication.

Description

Many district dynamic limited feedback method among the LTE-A

Technical field

The invention belongs to field of mobile communication, be specifically related to many district dynamic limited feedback method among a kind of LTE-A, to improve the throughput of Cell Edge User.

Background technology

In existing advanced long-term evolution system LTE-A (Long Term Evolution-Advanced), by using multi-antenna technology to go to improve user's performance, but the use of multi-antenna technology only can improve the performance of Cell Center User, but is difficult to improve the transmission rate of Cell Edge User.Third generation partner program 3GPP releases multipoint cooperative CoMP technology for this reason, and its main purpose is the cooperation by the minizone, and promptly a plurality of sub-districts are detected the performance that the user improves edge customer jointly.

The Limited Feedback that how to realize the CoMP system effectively also is the key issue that mobile communication system need be considered, therefore the feedback of CoMP and the combination of raising community marginal user performance is had great importance.

Method commonly used at present is that Cell Center User is used different number of bits of feedback with edge customer, and this method generally includes static bit feedback and two kinds of schemes of dynamic bit feedback.The number of bits of feedback of center of housing estate and edge customer is fixed in the wherein static bit feedback method, does not take into full account the actual demand of different user, can not satisfy user's needs well.The dynamic bit feedback method is according to the feedback bits of the dynamic distributing user of requirements of different users, this method is a target to optimize user's total capacity, though can obviously improve the total capacity of system, it is worthless in real system that but the mis-behave that but can cause certain customers, this performance by the sacrificial section user reach the method for whole system performance optimization.

Therefore, above-mentioned two kinds of methods all can not go effectively to improve the performance of Cell Edge User by the feedback of CoMP.

Summary of the invention

The objective of the invention is to overcome the deficiency of above-mentioned prior art, propose many district dynamic limited feedback method among a kind of LTE-A,, effectively raise the performance of Cell Edge User to go to satisfy the actual demand of different user by the feedback of CoMP.

The technical thought that realizes above-mentioned purpose of the present invention is to give the distribution feedback bits of a plurality of district dynamic to the link circuit condition of sub-district according to the user, and according to the actual needs of different user, the number of bits of feedback of each sub-district is distributed to each user in this sub-district dynamically, and concrete steps comprise as follows:

(1) feedback bits is distributed in the minizone:

(1a) set up the minizone Optimization Model:

maxmize?min(r _k)

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k}{M-1}}}\right)\≥r_k$

$\underset{k\∈A,B}{\mathrm{\Σ}}{C}_k\≤C$

C _kBe nonnegative integer

Wherein, r _kBe the capacity that each sub-district can reach, target function min (r _k) the expression capacity that the poorest sub-district reached, w _kBe the pre-coding matrix of sub-district, k is a cell number, C _kBe each sub-district optimal solution, C is total number of bits of feedback, and M is the antenna number of each sub-district, H _kBe the channel of each sub-district, A and B are meant two sub-districts;

(1b) total number of bits of feedback C is relaxed into nonnegative real number, with the protruding optimization tool of standard Optimization Model is found the solution, after the sub-optimal result of trying to achieve rounded downwards, again with the neighbor searching algorithm to searching for by rounding the data that obtain downwards, obtain optimal solution C _AAnd C _B, respectively as the number of bits of feedback of distributing to sub-district A and sub-district B;

(2) feedback bits of each sub-district is distributed to each user in this sub-district:

(2a) set up Optimization Model in the sub-district:

maxmize?min(r _i)

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|h_i\left|\right|}^2{\left|{\stackrel{\text{~}}{h}}_i^H{w}_i\right|}^2}{1+{\left|\right|h_i\left|\right|}^2{2}^{-\frac{b_i}{M-1}}}\right)\≥r_i,$

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{b}_i\≤C_A,C_B$

b _iBe nonnegative integer

Wherein, r _iBe the capacity that each user can reach, target function min (r _i) the expression capacity that the poorest user reached, i is a Customs Assigned Number, b _iBe each user's optimal solution, N is a user's number in the sub-district, h _iBe i user's channel, w _iBe the pre-coding matrix of sub-district, C _AAnd C _BBe the A sub-district that obtains after the minizone Optimization Model is found the solution and the optimal solution of B sub-district, M is the antenna number of each sub-district;

(2b) with the optimal solution C of A sub-district _AOptimal solution C with the B sub-district _BRelax into nonnegative real number, Optimization Model is found the solution, after the sub-optimal result of trying to achieve is rounded, to searching for, obtain optimal solution b downwards by rounding the data that obtain downwards with the neighbor searching algorithm with the protruding optimization tool of standard ₁, b ₂B _N, as distributing to user 1, user 2 ... the number of bits of feedback of user N, after over-allocation, the remaining bits that total number of bits of feedback is deducted each user's number of bits of feedback gained compensates that user to the capacity minimum that can reach;

(3) repeating step (1) and step (2) are redistributed each user's number of bits of feedback.

At having the present situation that feedback method is not considered user's actual needs and worsened certain customers' performances now, the present invention is a target with the performance that optimization can reach the minimum user of capacity, set up Optimization Model in minizone Optimization Model and the sub-district, the minizone Optimization Model is according to the link condition of user to each sub-district, optimized the performance that can reach the sub-district of capacity minimum, feedback bits can be distributed to each sub-district dynamically, Optimization Model is further according to user channel quality in the sub-district, find the solution by optimization tool, improved the performance that can reach the user of capacity minimum, feedback bits can be distributed to dynamically each user in the sub-district, like this, by finding the solution to Optimization Model in minizone Optimization Model and the sub-district, just can satisfy each user's actual demand, and improved the performance of Cell Edge User greatly, in the system of reality, avoided the poorest user can not obtain the problem of good service, guaranteed user's fairness better, very big application value has been arranged.

Description of drawings

Fig. 1 is a flow chart of the present invention;

Fig. 2 is the illustraton of model for 2 used CoMP sub-districts of checking the present invention;

Fig. 3 is the poorest user performance comparison diagram of the inventive method and existing method.

Embodiment

Fig. 1 is a FB(flow block) of the present invention, and Fig. 2 is the model of used simple 2 the CoMP sub-districts of the present invention, and these two sub-districts are carried out joint transmission and received by a center processing management.Solid line among Fig. 1 is represented the transmission between inner base station, sub-district and the user, and dotted line is represented the transmission between the user of the base station of different districts and other sub-districts.

With reference to Fig. 1, specific implementation step of the present invention is as follows:

Step 1 is distributed feedback bits in the minizone.

(1.1) set up the minizone Optimization Model

With reference to Fig. 2, the Optimization Model that the present invention sets up the minizone is as follows:

maxmize?min(r _k)

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k}{M-1}}}\right)\≥r_k,$

$\underset{k\∈A,B}{\mathrm{\Σ}}{C}_k\≤C$

C _kBe nonnegative integer

Wherein, r _kBe the capacity that each sub-district can reach, target function min (r _k) the expression capacity that the poorest sub-district reached, w _kBe the pre-coding matrix of sub-district, k is a cell number, C _kBe each sub-district optimal solution, C is total number of bits of feedback, and M is the antenna number of each sub-district, H _kBe the channel of each sub-district, A and B are meant two sub-districts;

(1.2) total number of bits of feedback C is relaxed into nonnegative real number, Optimization Model is found the solution with the protruding optimization tool of standard:

At first, the minizone Optimization Model is converted into the new Optimization Model in following minizone:

$\max \mathrm{mize}\underset{k\∈A,B}{\mathrm{\Σ}}\left(r_k\right)$

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k^{*}}{M-1}}}\right)=r_k$

r _A＝r _B

$\underset{k\∈A,B}{\mathrm{\Σ}}{C}_k^{*}=C$

C∈R ⁺，

Wherein, r _AAnd r _BBe respectively the capacity that A sub-district and B sub-district can reach, R ⁺Be non-negative set of real numbers,

It is the suboptimal solution that satisfies k sub-district of minizone Optimization Model;

Then, the new Optimization Model in minizone is found the solution, this solution procedure is converted into finds the solution following equation group:

$\left\{\begin{array}{c}\frac{{\left|\right|H_A\left|\right|}^2{\left|{\tilde{H}}_A^H{w}_A\right|}^2}{1+{\left|\right|H_A\left|\right|}^2{2}^{-\frac{C_A^{*}}{M-1}}}=\frac{{\left|\right|H_B\left|\right|}^2{\left|{\tilde{H}}_B^H{w}_B\right|}^2}{1+{\left|\right|H_B\left|\right|}^2{2}^{-\frac{C_B^{*}}{M-1}}}\\ C_A^{*}+C_B^{*}=C\end{array}\right.$

Wherein, H _AAnd H _BBe respectively the channel of A sub-district and B sub-district, w _AAnd w _BBe respectively the pre-coding matrix of A sub-district and B sub-district,

With It is respectively the suboptimal solution that satisfies the A sub-district and the B sub-district of minizone Optimization Model;

At last, the solving equation group obtains the suboptimal solution of minizone Optimization Model

With

(1.3) will round downwards by the sub-optimal result that the protruding optimization tool of standard is tried to achieve after, with the neighbor searching algorithm to searching for by rounding the data that obtain downwards:

(1.3.1) set initial value: to each sub-district, order

K=A, B,

$f={\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k}{M-1}}}\right),$

Wherein,

Be the suboptimal solution that step (1.2) is tried to achieve, μ _kBe value of rounding downwards of the capacity suboptimal solution of sub-district k, Ω is a sub-district remaining bits number;

(1.3.2) select k, make all j are satisfied

Wherein, j is a cell number, C _kAnd C _jIt is respectively the optimal solution of sub-district k and sub-district j;

(1.3.3) make μ _k← μ _k+ 1, Ω ← Ω-1, if Ω 〉=1, repeating step (1.3.1) and step (1.3.2); Otherwise stop;

(1.3.4) with the μ that tries to achieve in the step (1.3.3) _AAnd μ _BValue compose optimal solution C to the sub-district _AAnd C _B

Step 2 is distributed to each user in this sub-district with the feedback bits of each sub-district.

(2.1) set up Optimization Model in the sub-district:

maxmize?min(r _i)

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|h_i\left|\right|}^2{\left|{\stackrel{\text{~}}{h}}_i^H{w}_i\right|}^2}{1+{\left|\right|h_i\left|\right|}^2{2}^{-\frac{b_i}{M-1}}}\right)\≥r_i,$

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{b}_i\≤C_A,C_B$

b _iBe nonnegative integer

Wherein, r _iBe the capacity that each user can reach, target function min (r _i) the expression capacity that the poorest user reached, i is a Customs Assigned Number, b _iBe each user's optimal solution, N is a user's number in the sub-district, h _iBe i user's channel, w _iBe the pre-coding matrix of sub-district, C _AAnd C _BBe the A sub-district that obtains after the minizone Optimization Model is found the solution and the optimal solution of B sub-district, M is the antenna number of each sub-district;

(2.2) with the optimal solution C of A sub-district _AOptimal solution C with the B sub-district _BRelax into nonnegative real number, Optimization Model found the solution with the protruding optimization tool of standard:

At first, Optimization Model in the sub-district is converted into new Optimization Model in the following sub-district:

$\max \mathrm{mize}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\left(r_i\right)$

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_i\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_i^H{w}_i\right|}^2}{1+{\left|\right|H_i\left|\right|}^2{2}^{-\frac{b_i^{*}}{M-1}}}\right)=r_i$

r _i＝r _m， $\∀i\≠m$

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{b}_i^{*}=C_A,C_B$

C _A，C _B∈R ⁺，

Wherein, m is a Customs Assigned Number,

It is the suboptimal solution that satisfies the user i correspondence of Optimization Model in the sub-district;

Then, new Optimization Model in the sub-district is found the solution, this solution procedure is converted into finds the solution following equation group:

$\left\{\begin{array}{c}\frac{{\left|\right|H_1\left|\right|}^2{\left|{\tilde{H}}_1^H{w}_1\right|}^2}{1+{\left|\right|H_1\left|\right|}^2{2}^{-\frac{b_1^{*}}{M-1}}}=\frac{{\left|\right|H_2\left|\right|}^2{\left|{\tilde{H}}_2^H{w}_2\right|}^2}{1+{\left|\right|H_2\left|\right|}^2{2}^{-\frac{b_2^{*}}{M-1}}}=...=\frac{{\left|\right|H_N\left|\right|}^2{\left|{\tilde{H}}_N^H{w}_N\right|}^2}{1+{\left|\right|H_N\left|\right|}^2{2}^{-\frac{b_N^{*}}{M-1}}}\\ b_1^{*}+b_2^{*}+...+b_N^{*}=C_A,C_B\end{array}\right.,$

Wherein, H ₁, H ₂H _NBe respectively user 1, the user 2 ... the channel of user N, w ₁, w ₂W _NBe respectively user 1, the user 2 ... the pre-coding matrix of user N, Be respectively the user 1 who satisfies Optimization Model in the sub-district, the user 2 ... the suboptimal solution of user N;

At last, the solving equation group obtains the suboptimal solution of Optimization Model in the sub-district

(2.3) sub-optimal result of trying to achieve is rounded downwards after, to searching for, obtain optimal solution b with the neighbor searching algorithm by rounding the data that obtain downwards ₁, b ₂B _N, as distributing to user 1, user 2 ... the number of bits of feedback of user N, after over-allocation, the remaining bits that total number of bits of feedback is deducted each user's number of bits of feedback gained compensates that user to the capacity minimum that can reach:

(2.3.1) set initial value: to each user, order

I=1,2 ... N,

${\mathrm{\Ω}}_{A,B}\←C_{A,B}-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_i,$ $f={\log }_2\left(\frac{{\left|\right|h_i\left|\right|}^2{\left|{\stackrel{\text{~}}{h}}_i^H{w}_i\right|}^2}{1+{\left|\right|h_i\left|\right|}^2{2}^{-\frac{b_i}{M-1}}}\right),$

Wherein,

Be the suboptimal solution of the user i that tries to achieve of step (2.2), μ _iBe

The value of rounding downwards, Ω _{A, B}It is user's remaining bits number;

(2.3.2) select m, all n are satisfied $\frac{\∂f}{\∂b_m}\≤\frac{\∂f}{\∂b_n},$

Wherein, n is a Customs Assigned Number, b _mAnd b _nIt is respectively the optimal solution of user m and user n;

(2.3.3) make μ _m← μ _m+ 1, Ω _{A, B}← Ω _{A, B}-1,

Wherein, μ _mBe value of rounding downwards of the suboptimal solution of user m,

If Ω _{A, B}〉=1, repeating step (2.3.1) and step (2.3.2), otherwise stop;

(2.3.4) with μ ₁, μ ₂μ _NValue compose optimal solution b to the user ₁, b ₂B _N, calculate then and make user capacity r _iObtain minimum i value, with Ω _{A, B}Value compose to optimal solution b _i

Step 3, repeating step 1 and step 2 are redistributed each user's number of bits of feedback.

Effect of the present invention can further specify by following simulation result.

1. simulation parameter:

Code modulation mode: coding+QPSK not;

Sub-district number and user's number: totally two sub-districts, two users in each sub-district;

Antenna configurations: two antennas of each base station configuration, each user disposes an antenna;

Method for precoding: based on the associating precoding of characteristic value decomposition;

Channel type: multiple Gaussian channel, average is 0, variance is 1;

2. simulation result

Emulation content of the present invention is among original feedback method and the LTE-A of the present invention under many district dynamic limited feedback method, and the poorest user is the user's of capacity minimum a performance, as shown in Figure 3.As can be seen from Figure 3, optimize the user can reach the capacity minimum and can make each user's performance average, its poorest user's performance be better than the performance of the poorest user when being target to the maximum with total speed.

Claims (5)

1. many district dynamic limited feedback method among the LTE-A comprises the steps:

(1) feedback bits is distributed in the minizone:

(1a) set up the minizone Optimization Model:

maxmize min(r _k)

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k}{M-1}}}\right)\≥r_k,$

$\underset{k\∈A,B}{\mathrm{\Σ}}{C}_k\≤C$

C _kBe nonnegative integer

R wherein _kBe the capacity that each sub-district can reach, target function min (r _k) the expression capacity that the poorest sub-district reached, w _kBe the pre-coding matrix of sub-district, k is a cell number, C _kBe each sub-district optimal solution, C is total number of bits of feedback, and M is the antenna number of each sub-district, H _kBe the channel of each sub-district, A and B are meant two sub-districts;

(1b) total number of bits of feedback C is relaxed into nonnegative real number, with the protruding optimization tool of standard Optimization Model is found the solution, after the sub-optimal result of trying to achieve rounded downwards, again with the neighbor searching algorithm to searching for by rounding the data that obtain downwards, obtain optimal solution C _AAnd C _B, respectively as the number of bits of feedback of distributing to sub-district A and sub-district B;

(2) feedback bits of each sub-district is distributed to each user in this sub-district:

(2a) set up Optimization Model in the sub-district:

maxmize?min(r _i)

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|h_i\left|\right|}^2{\left|{\stackrel{\text{~}}{h}}_i^H{w}_i\right|}^2}{1+{\left|\right|h_i\left|\right|}^2{2}^{-\frac{b_i}{M-1}}}\right)\≥r_i,$

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{b}_i\≤C_A,C_B$

b _iBe nonnegative integer

R wherein _iBe the capacity that each user can reach, target function min (r _i) the expression capacity that the poorest user reached, i is a Customs Assigned Number, b _iBe each user's optimal solution, N is a user's number in the sub-district, h _iBe i user's channel, w _iBe the pre-coding matrix of sub-district, C _AAnd C _BBe the A sub-district that obtains after the minizone Optimization Model is found the solution and the optimal solution of B sub-district, M is the antenna number of each sub-district;

(2b) with the optimal solution C of A sub-district _AOptimal solution C with the B sub-district _BRelax into nonnegative real number, Optimization Model is found the solution, after the sub-optimal result of trying to achieve is rounded, to searching for, obtain optimal solution b downwards by rounding the data that obtain downwards with the neighbor searching algorithm with the protruding optimization tool of standard ₁, b ₂B _N, as distributing to user 1, user 2 ... the number of bits of feedback of user N, after over-allocation, the remaining bits that total number of bits of feedback is deducted each user's number of bits of feedback gained compensates that user to the capacity minimum that can reach;

(3) repeating step (1) and step (2) are redistributed each user's number of bits of feedback.

2. many district dynamic limited feedback method among the LTE-A according to claim 1, wherein step (1b) is described relaxes into nonnegative real number with total number of bits of feedback C, with the protruding optimization tool of standard Optimization Model is found the solution, and carries out as follows:

2a) the minizone Optimization Model is converted into:

$\max \mathrm{mize}\underset{k\∈A,B}{\mathrm{\Σ}}\left(r_k\right)$

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k^{*}}{M-1}}}\right)=r_k$

r _A＝r _B

$\underset{k\∈A,B}{\mathrm{\Σ}}{C}_k^{*}=C$

C∈R ⁺，

Wherein, r _AAnd r _BBe respectively the capacity that A sub-district and B sub-district can reach, R ⁺Be non-negative set of real numbers,

It is the suboptimal solution that satisfies k sub-district of minizone Optimization Model.

2b) solution procedure of new model be converted into find the solution following equation group:

$\left\{\begin{array}{c}\frac{{\left|\right|H_A\left|\right|}^2{\left|{\tilde{H}}_A^H{w}_A\right|}^2}{1+{\left|\right|H_A\left|\right|}^2{2}^{-\frac{C_A^{*}}{M-1}}}=\frac{{\left|\right|H_B\left|\right|}^2{\left|{\tilde{H}}_B^H{w}_B\right|}^2}{1+{\left|\right|H_B\left|\right|}^2{2}^{-\frac{C_B^{*}}{M-1}}}\\ C_A^{*}+C_B^{*}=C\end{array}\right.$

Wherein, H _AAnd H _BBe respectively the channel of A sub-district and B sub-district, w _AAnd w _BBe respectively the pre-coding matrix of A sub-district and B sub-district,

With

It is respectively the suboptimal solution that satisfies the A sub-district and the B sub-district of minizone Optimization Model;

2c) solution procedure 2b) equation group in obtains suboptimal solution

With

3. many district dynamic limited feedback method among the LTE-A according to claim 1, wherein step (1b) is described will round downwards by the sub-optimal result that the protruding optimization tool of standard is tried to achieve after,, carry out as follows searching for the neighbor searching algorithm by rounding the data that obtain downwards:

3a) set initial value: to each sub-district, order

K=A, B,

$f={\log }_2\left(\frac{{\left|\right|H_k\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_k^H{w}_k\right|}^2}{1+{\left|\right|H_k\left|\right|}^2{2}^{-\frac{C_k}{M-1}}}\right),$

Wherein,

Be step 2c) suboptimal solution of trying to achieve, μ _kBe value of rounding downwards of the capacity suboptimal solution of sub-district k, Ω is a sub-district remaining bits number;

3b) select k, make all j are satisfied $\frac{\∂f}{\∂C_k}\≤\frac{\∂f}{\∂C_j},$

Wherein, j is a cell number, C _kAnd C _jIt is respectively the optimal solution of sub-district k and sub-district j;

3c) make μ _k← μ _k+ 1, Ω ← Ω-1, if Ω 〉=1, repeating step 3a) and step 3b); Otherwise stop;

3d) with step 3c) in the μ that tries to achieve _AAnd μ _BValue compose optimal solution C to the sub-district _AAnd C _B

4. many district dynamic limited feedback method among the LTE-A according to claim 1, the wherein described optimal solution C of step (2b) with the A sub-district _AOptimal solution C with the B sub-district _BRelax into nonnegative real number, Optimization Model found the solution, carry out as follows with the protruding optimization tool of standard:

4a) Optimization Model in the sub-district is converted into:

$\max \mathrm{mize}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\left(r_i\right)$

$\mathrm{subject\; to}{\log }_2\left(\frac{{\left|\right|H_i\left|\right|}^2{\left|{\stackrel{\text{~}}{H}}_i^H{w}_i\right|}^2}{1+{\left|\right|H_i\left|\right|}^2{2}^{-\frac{b_i^{*}}{M-1}}}\right)=r_i$

r _i＝r _m， $\∀i\≠m$

$\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{b}_i^{*}=C_A,C_B$

C _A，C _B∈R ⁺，

Wherein, m is a Customs Assigned Number,

It is the suboptimal solution that satisfies the user i correspondence of Optimization Model in the sub-district;

4b) solution procedure of new model be converted into find the solution following equation group:

$\left\{\begin{array}{c}\frac{{\left|\right|H_1\left|\right|}^2{\left|{\tilde{H}}_1^H{w}_1\right|}^2}{1+{\left|\right|H_1\left|\right|}^2{2}^{-\frac{b_1^{*}}{M-1}}}=\frac{{\left|\right|H_2\left|\right|}^2{\left|{\tilde{H}}_2^H{w}_2\right|}^2}{1+{\left|\right|H_2\left|\right|}^2{2}^{-\frac{b_2^{*}}{M-1}}}=...=\frac{{\left|\right|H_N\left|\right|}^2{\left|{\tilde{H}}_N^H{w}_N\right|}^2}{1+{\left|\right|H_N\left|\right|}^2{2}^{-\frac{b_N^{*}}{M-1}}}\\ b_1^{*}+b_2^{*}+...+b_N^{*}=C_A,C_B\end{array}\right.,$

Wherein, H ₁, H ₂H _NBe respectively user 1, the user 2 ... the channel of user N, w ₁, w ₂W _NBe respectively user 1, the user 2 ... the pre-coding matrix of user N,

Be respectively the user 1 who satisfies Optimization Model in the sub-district, the user 2 ... the suboptimal solution of user N;

4c) solution procedure 4b) equation group in obtains suboptimal solution

5. many district dynamic limited feedback method among the LTE-A according to claim 1, wherein step (2b) is described will round downwards by the sub-optimal result that the protruding optimization tool of standard is tried to achieve after,, carry out as follows searching for the neighbor searching algorithm by rounding the data that obtain downwards:

5a) set initial value: to each user, order

I=1,2 ... N,

$f={\log }_2\left(\frac{{\left|\right|h_i\left|\right|}^2{\left|{\stackrel{\text{~}}{h}}_i^H{w}_i\right|}^2}{1+{\left|\right|h_i\left|\right|}^2{2}^{-\frac{b_i}{M-1}}}\right),$

Wherein,

Be step 4c in the claim 4) suboptimal solution of the user i that tries to achieve, μ _iBe

The value of rounding downwards, Ω _{A, B}It is user's remaining bits number;

5b) select m, all n are satisfied $\frac{\∂f}{\∂b_m}\≤\frac{\∂f}{\∂b_n},$

Wherein, n is a Customs Assigned Number, b _mAnd b _nIt is respectively the optimal solution of user m and user n;

5c) make μ _m← μ _m+ 1, Ω _{A, B}← Ω _{A, B}-1,

Wherein, μ _mBe value of rounding downwards of the suboptimal solution of user m,

If Ω _{A, B}〉=1, repeating step 5a) and step 5b), otherwise stop;

5d) with μ ₁, μ ₂μ _NValue compose optimal solution b to the user ₁, b ₂B _N, calculate then and make user capacity r _iObtain minimum i value, with Ω _{A, B}Value compose to optimal solution b _i

Images