CN103118436A - User scheduling algorithm for MU-MIMO (multi-user multiple input multiple output) down link based on interference pre-evaluation - Google Patents

Abstract

Disclosed is a user scheduling algorithm for an MU-MIMO (multi-user multiple input multiple output) down link based on interference pre-evaluation. User scheduling is converted into space component channel selection, and singular values are decomposed according to a learned channel matrix H between a user and a base station, namely, MIMO signals are equalized into rank (H) parallel component channels, wherein the rank (H) indicates a derived matrix rank. By constructing sandwich matrixes and correlation matrixes, interference pre-evaluation factors are set, mutual interference of space component channels and transmission gain of the component channels in the scheduling process are comprehensively considered, user scheduling of an MU-MIMO system is realized, and better space component channel sets are selected for communication.

Description

A kind of MU-MIMO down link is based on user's dispatching algorithm of disturbing Pre-Evaluation

Technical field

The invention belongs to communication technical field, be specially a kind of MU-MIMO down link based on user's dispatching algorithm of disturbing Pre-Evaluation.

Background technology

MIMO(Multiple input multiple output) technology can be improved transmission rate, promote the advantage aspect link reliability with it under the prerequisite that does not increase transmitting power and bandwidth, be subject to extensive concern in more than ten years in the past, become the key technology of multiple wideband wireless mobile communication standard, as LTE-A and 802.16m.Than Single User MIMO, multiuser MIMO (Multiuser MIMO, MU-MIMO) is the elevator system performance further.In the MU-MIMO system, due to the restriction of base station ability, need to select one group from a plurality of users and serve, reasonably user's scheduling can obtain multi-user diversity gain (Multiuser diversity, MUD), realizes taking full advantage of of the communication resource.

In MU-MIMO user's scheduling, the data of different excited users are while, same concurrent transmission frequently, have road interference (Co-channel interference, CCI) altogether between them, and this factor also becomes the design considerations of many dispatching algorithms.When channel condition information (Channel State Information, CSI) was known, the base station can utilize exhaustive search to select one group of user that can obtain maximum system and speed, but the method complexity is high, is difficult in practice realize.In order to reduce complexity, the method for some suboptimums is suggested in succession.As the suboptimum greediness user selection algorithm based on channel relevancy, at first the method determines the user that communication quality is best, then turns to target with system and speed maximum and activates successively other users; Quasi-orthogonal user scheduling method, based on user's spatial relationship select one group preferably the user communicate; , only the user who is better than thresholding is dispatched from reducing feedback overhead based on the user scheduling method of presetting thresholding; Foundation and the phase mutual interference of selecting between the family are realized the distributed user dispatching algorithm of dispatching jointly by user and base station.Above research mainly is optimized for target with power system capacity and carries out the dispatching algorithm design.Also there is a few thing to adopt other design criterion, as based on the error rate or chordal distance dispatching algorithm.

It should be noted that in the process of scheduling, the user normally adds one by one.Above-mentioned algorithm carries out user's selection based on candidate user with selecting the phase mutual interference between the user, ignored user's potential, in the future may be selected impact (can be called the scheduling of posteriority formula (Reactive)), the interference between excited users is little may to cause the user that selects for the k time and front, but with remaining candidate user, larger phase mutual interference is arranged all, thereby affect system and speed.Therefore, the present invention is to candidate user and selected the interference between the user that may be activated in family and potential future to consider, the scheduling mechanism of a kind of priori formula (Proactive) is proposed, by comprehensive interference assessment, obtain one group of good user of the degree of correlation each other, realize the improvement of system and speed.

Existing subscriber's dispatching algorithm has following two limitation: (1) has only considered the degree of correlation of spatial sub-channel, has ignored the transmission gain of spatial sub-channel under different signal to noise ratios to the impact of communication performance; (2) ignored the potential spatial sub-channel that may be selected and selected interference between spatial sub-channel on the impact of system communication performance;

The content of invention

The object of the invention is to by considering the candidate sub-channel transmission gain, and candidate sub-channel with select that subchannel and a part are potential, phase mutual interference between subchannel that in the future may be selected, obtain one group and disturb each other less subchannel, propose a kind of MU-MIMO down link based on user's dispatching algorithm of disturbing Pre-Evaluation.

It comprises, a kind of MU-MIMO down link comprises the steps: based on user's dispatching algorithm of disturbing Pre-Evaluation

(1) spatial sub-channel set and the set of candidate's spatial sub-channel have been selected in initialization, wherein;

Selected spatial sub-channel set A=Φ, candidate's spatial sub-channel set C=Ω, scheduling times t=0 arranges and disturbs Pre-Evaluation factor η, and Φ represents null set, and Ω represents the universal class of all subchannels, and singular value decomposition H is carried out to the channel information matrix that obtains in the base station _k=U _kΛ _k(V _k) ^H, wherein, ${\mathrm{\Λ}}_k=\mathrm{diag}\left[\begin{array}{cccc}{\mathrm{\λ}}_{k,1}& {\mathrm{\λ}}_{k,2}& \·\·\·& {\mathrm{\λ}}_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ $U_k=\left[\begin{array}{cccc}{u}_{k,1},& u_{k,2},& \·\·\·,& u_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ $V_k=\left[\begin{array}{cccc}{v}_{k,1},& v_{k,2},& \·\·\·,& v_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ λ _k,jThe j singular value after channel information matrix between k user and base station carries out singular value decomposition, u _k,jThe j row that represent k user's U matrix, v _k,jRepresent the j row of k user's V matrix, rank of matrix is asked in rank () expression, and diag () expression diagonalization is processed.Carry out after singular value decomposition, the mimo channel equivalence being rank (H _k) individual decoupling parallel sub-channels;

(2) structure intermediary matrix

Wherein, V _kIn column vector with The mapping relations of middle column vector are

Λ _kMain diagonal element with The mapping relations of main diagonal element be

Step 3: structure correlation matrix $R=\left[\begin{array}{ccc}{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HDA00002846132100021.png"\; state="\{\{state\}\}">r</figure-callout>}}_1& \&CenterDot;\&CenterDot;\&CenterDot;& r_{{\mathrm{\&Sigma;}}_{k=1}^K\mathrm{rank}\left(H_k\right)}\end{array}\right];$

In formula, || expression is asked modular arithmetic,＜a, b〉inner product operation of expression vectorial a and b, r _m,nM is capable for the expression correlation matrix, the element of n row;

Step 4: calculate objective matrix;

(1) when t=0, A=Φ is arranged, namely card (A)=0, directly do ascending order to each row element of correlation matrix R and arrange, and obtains objective matrix R ₀, its element r _{0, (m, n)}Expression.

(2) when t＞0, card (A)=t is arranged, choose the t row corresponding with the t that has a selected spatial sub-channel and consist of matrix from R

With remaining part in R (altogether

Row) respectively row element is carried out the ascending order arrangement and obtain matrix

The structure objective matrix

Step 5: calculate object vector;

To R _tFront η element of every delegation sue for peace respectively,

Obtain object vector ${\mathrm{\&Psi;}}_t={\left[\begin{array}{ccc}{\mathrm{\&psi;}}_{t,1}& \&CenterDot;\&CenterDot;\&CenterDot;& {\mathrm{\&psi;}}_{t,{\mathrm{\&Sigma;}}_{k=1}^K\mathrm{rank}\left(H_k\right)}\end{array}\right]}^T;$

Step 6: definite user's subchannel that will communicate;

Determine the sequence number of t spatial sub-channel by following formula according to the transmission gain of object vector and spatial sub-channel: ${\tilde{k}}_t=\arg \underset{k\&Element;C}{\min }({\mathrm{\&psi;}}_{t,k}+{\mathrm{\&delta;}}_k),$ Wherein,

SNR equals

Step 7: new variables more;

T=t+1, A=A ∪ { k _t, if C=Ω-A is card (A)=N _T, algorithm finishes; Otherwise, return to step 4.

Relative prior art, the present invention changes into spatial sub-channel with user's scheduling and selects, according to the channel matrix H between the user of knowing and base station, it is carried out singular value decomposition, being about to the mimo channel equivalence is the individual parallel sub-channels of rank (H), and wherein rank of matrix is asked in rank () expression.By structure intermediary matrix and correlation matrix, arrange and disturb the Pre-Evaluation factor, consider the transmission gain of spatial sub-channel, realize the scheduling to multi-user MIMO system, select better spatial sub-channel set and communicate.Step comprises as follows.

Description of drawings

Fig. 1 is single residential quarter MU-MIMO system model;

Fig. 2 is based on testing formula user dispatching algorithm flow chart before the interference Pre-Evaluation factor;

Fig. 3 is that different signal to noise ratio lower channel capacity are with the change curve analogous diagram of disturbing Pre-Evaluation factor η;

Fig. 4 is L=8, N _T=4, N _k=2, η=N _T=4 o'clock, the channel capacity analogous diagram of different user dispatching method.

Specific implementation method

With reference to Fig. 1, the present invention studies single residential quarter MU-MIMO downlink broadcast communication system, and the user both can adopt beam forming (Beamforming, BF) mode, also can adopt space division multiplexing (Spatial Multiplexing, SM) mode.Total number of users L in the residential quarter, antenna for base station is counted N _T, the antenna number N of mobile subscriber k _k, the base station therefrom selects the K sub-channels to communicate by letter simultaneously.Usually travelling carriage has stricter size restrictions than the base station, without loss of generality, supposes N _T＞N _kSuppose to obtain channel information accurately, H _kN between expression user k and base station _k* N _TRank channel matrix, its element are 0 average, and the multiple Gaussian random variable of the independent same distribution of unit variance supposes that all users experience the identical and separate frequency-flat decline of statistical property.For guaranteeing fairness, the activation space subchannel is divided equally base station transmitting power P _T

The specific embodiment of the present invention is described in further detail, and below narration arranges N _T=4, disturb Pre-Evaluation factor η=N _T

Step 1: spatial sub-channel set and the set of candidate's spatial sub-channel have been selected in initialization;

Initialization is carried out to following parameters in the base station, has selected spatial sub-channel set A=Φ and candidate's spatial sub-channel set C=Ω, and scheduling times t=0 arranges and disturbs Pre-Evaluation factor η=N _TWherein Φ represents null set, and Ω represents the universal class of all subchannels.User's side adopts the method for combined channel parameter Estimation can obtain channel matrix information, i.e. base station emission N _TWay signal stream adds the data block that the known training symbol of user's side forms, the estimation that the user realizes the channel state information matrix H between user and base station according to the signal of receiving and known training data before every way signal stream.The user sends to the base station with the channel information matrix.

Step 2: structure intermediary matrix;

Singular value decomposition H is carried out to the channel information matrix H in the base station _k=U _kΛ _k(V _k) ^H(k represents k user).Wherein $U_k=\left[\begin{array}{cccc}{u}_{k,1},& u_{k,2},& \&CenterDot;\&CenterDot;\&CenterDot;,& u_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ $V_k=\left[\begin{array}{cccc}{v}_{k,1},& v_{k,2},& \&CenterDot;\&CenterDot;\&CenterDot;,& v_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ ${\mathrm{\&Lambda;}}_k=\mathrm{diag}\left[\begin{array}{cccc}{\mathrm{\&lambda;}}_{k,1}& {\mathrm{\&lambda;}}_{k,2}& \&CenterDot;\&CenterDot;\&CenterDot;& {\mathrm{\&lambda;}}_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right].$

Wherein, u _k,jThe j row that represent k user's U matrix, v _k,jThe j row that represent k user's V matrix, λ _k,jJ the characteristic value that represents k user's channel information matrix, reflection spatial sub-channel transmission gain.Carry out after singular value decomposition, the mimo channel equivalence being rank (H _k) individual decoupling parallel sub-channels.

Base station structure intermediary matrix

Wherein, intermediary matrix

In v _l,jThe channel information matrix H that represents l user _kCarry out the j column vector of the V matrix after singular value decomposition, corresponding to Wherein, mapping relations are

Here all users' subchannel being placed on the status that is equal to processes, which user does not distinguish this subchannel when therefore selecting is, therefore a plurality of subchannels that choose at last might belong to respectively different user, also may belong to identical user, the user who namely activates might adopt the BF mode, also might select the SM mode.User's scheduling is converted to subchannel and selects.

Step 3: structure correlation matrix;

Correlation matrix is constructed according to intermediary matrix in the base station

In formula, || expression is asked modular arithmetic,＜a, b〉inner product operation of expression vectorial a and b, r _m,nM is capable for the expression correlation matrix, the element of n row, the degree of correlation between reflection subchannel m and n.

Step 4: the acquisition of objective matrix;

Following operation is carried out according to the information that above-mentioned steps obtains in the base station:

(1) when t=0, card (A)=0 is arranged.Directly each row element of correlation matrix R is done ascending order and arrange, obtain objective matrix R ₀, its element r _{0, (m, n)}Expression.

(2) when t＞0, card (A)=t is arranged.Choose the t row corresponding with the t that has a selected spatial sub-channel and consist of matrix from R

With remaining part in R (altogether

Row) respectively row element is carried out the ascending order arrangement and obtain matrix

The structure objective matrix

Because the degree of correlation between the reflection of the element in correlation matrix subchannel, the degree of correlation between the user is less shows that mutual interference is lower, therefore easier being chosen.The structure of objective matrix is the degree of correlation between each sub-channels to be done ascending order from small to large arrange.

Step 5: calculate object vector;

The base station is according to disturbing Pre-Evaluation factor η=N _T, to R _tFront η element of every delegation sue for peace respectively, ${\mathrm{\&psi;}}_{t,m}=\underset{n=1}{\overset{\mathrm{\&eta;}}{\mathrm{\&Sigma;}}}{r}_{t,(m,n)},$ Obtain object vector ${\mathrm{\&Psi;}}_t={\left[\begin{array}{ccc}{\mathrm{\&psi;}}_{t,1}& \&CenterDot;\&CenterDot;\&CenterDot;& {\mathrm{\&psi;}}_{t,{\mathrm{\&Sigma;}}_{k=1}^K\mathrm{rank}\left(H_k\right)}\end{array}\right]}^T.$

The interference Pre-Evaluation factor η that herein arranges is a fixing value, when the number of subchannels of having selected during less than η, to R _tFront η element of every delegation sue for peace respectively, will select in fact subchannel and the degree of correlation potential, between may selecteed subchannel all to take into account.Can obtain like this to disturb each other less sets of sub-channels.

Step 6: select spatial sub-channel;

The sequence number of t spatial sub-channel is determined according to the transmission gain of object vector and spatial sub-channel in the base station by following formula: ${\tilde{k}}_t=\arg \underset{k\&Element;C}{\min }({\mathrm{\&psi;}}_{t,k}+{\mathrm{\&delta;}}_k).$ Wherein,

SNR equals

The choice criteria of subchannel has considered the impact of phase mutual interference between subchannel and subchannel transmission gain herein, and is more comprehensive.

Step 7: new variables more;

The base station is new data more: t=t+1, C=Ω-A.If card (A)=N _T, activating and selected subchannel to communicate, algorithm finishes; Otherwise, return to step 4, continue the selection of follow-up subchannel.

Emulation experiment

Effect of the present invention can further illustrate by following emulation:

Simulated conditions: the total number of users L=8 in the MU-MIMO system, N _T=4, N _k=2, η span is (1, Lmin (N _T, N _k)] integer.The user scheduling of research comprises: 1. exhaustive scheduling (Exhaustive scheduling, ES) can obtain the subchannel of maximum and speed by one group of exhaustive selection; 2. posteriority formula scheduling (Reactive scheduling, RS), original subchannel only determine that according to transmission gain all the other subchannels based on selecting with the interference of selecting part, are equivalent to disturb Pre-Evaluation factor η=card (A)+1; 3. priori formula scheduling (Proactive scheduling, PS), this paper method; 4. do not consider δ _kPriori formula scheduling (PS without δ _k, PS w/o δ _k).

Fig. 3 has provided under different signal to noise ratios, the impact of different η on power system capacity.

As can be seen from the figure.Under specific SNR, η chooses adopting system that PS obtains and the impact of speed.Use solid five-pointed star to best η in figure _optMark.Can find, get η=N _TMaximum be can obtain or maximum system and speed approached.Due at η _optNear, and speed and optimum value very nearly the same, from reducing the angle of complexity, get η=N _TRational.Change N _TThe time, also can lead to the same conclusion.

Fig. 4 has provided at η=N _TUnder condition, the power system capacity simulation curve figure under above-mentioned four kinds of different user dispatching algorithms.

Can find from figure, ES can obtain optimum system and speed, and though RS original subchannel determine or follow-up subchannel interpolation aspect, to disturbing and subchannel transmission gains and all lacks comprehensive consideration in common road, so and rate capability far below ES.Owing to considering candidate sub-channel and the phase mutual interference of selecting between subchannel and potential possible selected subchannel, priori formula dispatching algorithm significantly is better than the scheduling of posteriority formula.Should be understood that, for those of ordinary skills, can be improved according to the above description or conversion, and all these improve and conversion all should belong to the protection range of claims of the present invention.

Claims (3)

1. a MU-MIMO down link is based on user's dispatching algorithm of disturbing Pre-Evaluation, it is applicable to cordless communication network, described cordless communication network comprises and selects spatial sub-channel and candidate's spatial sub-channel, it is characterized in that: selected the equal participating user scheduling of spatial sub-channel and candidate's spatial sub-channel to calculate.

2. a kind of MU-MIMO down link as claimed in claim 1 is based on user's dispatching algorithm of disturbing Pre-Evaluation, and it is characterized in that: it comprises the following steps:

(1) spatial sub-channel set and the set of candidate's spatial sub-channel have been selected in initialization, wherein;

If selected spatial sub-channel set A=Φ, candidate's spatial sub-channel set C=Ω, scheduling times t=0, arrange and disturb Pre-Evaluation factor η, Φ represents null set, and Ω represents the universal class of all subchannels, and singular value decomposition H is carried out to the channel information matrix that obtains in the base station _k=U _kΛ _k(V _k) ^H, wherein, ${\mathrm{\&Lambda;}}_k=\mathrm{diag}\left[\begin{array}{cccc}{\mathrm{\&lambda;}}_{k,1}& {\mathrm{\&lambda;}}_{k,2}& \&CenterDot;\&CenterDot;\&CenterDot;& {\mathrm{\&lambda;}}_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ $U_k=\left[\begin{array}{cccc}{u}_{k,1},& u_{k,2},& \&CenterDot;\&CenterDot;\&CenterDot;,& u_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ $V_k=\left[\begin{array}{cccc}{v}_{k,1},& v_{k,2},& \&CenterDot;\&CenterDot;\&CenterDot;,& v_{k,\mathrm{rank}\left(H_k\right)}\end{array}\right],$ λ _k,jThe j singular value after channel information matrix between k user and base station carries out singular value decomposition, u _k,jThe j row that represent k user's U matrix, v _k,jRepresent the j row of k user's V matrix, rank of matrix is asked in rank () expression, and diag () expression diagonalization is processed;

(2) structure intermediary matrix

Wherein, mapping relations are

Step 3: structure correlation matrix $R=\left[\begin{array}{ccc}{r}_1& \&CenterDot;\&CenterDot;\&CenterDot;& r_{{\mathrm{\&Sigma;}}_{k=1}^K\mathrm{rank}\left(H_k\right)}\end{array}\right];$

In formula, || expression is asked modular arithmetic,＜a, b〉inner product operation of expression vectorial a and b, r _m,nM is capable for the expression correlation matrix, the element of n row;

Step 4: calculate objective matrix;

(1) when t=0, A=Φ is arranged, namely card (A)=0, directly do ascending order to each row element of correlation matrix R and arrange, and obtains objective matrix R ₀, the element number of set A is asked in card (A) expression;

(2) when t＞0, card (A)=t is arranged, choose the t row corresponding with the t that has a selected spatial sub-channel and consist of matrix from R

Respectively the row element of remainder in R being carried out ascending order arranges and obtains matrix

Described remainder is candidate's spatial sub-channel, the structure objective matrix

Step 5: calculate object vector;

To R _tFront η element of every delegation sue for peace respectively, Obtain object vector

Wherein, η needs the subchannel number of the phase mutual interference of investigation for when dispatching;

Step 6: new variables more;

T=t+1,

If C=Ω-A is card (A)=N _T, algorithm finishes; Otherwise, return to step 4.

3. a kind of MU-MIMO down link as claimed in claim 1 based on user's dispatching algorithm of disturbing Pre-Evaluation, is characterized in that: further comprising the steps of between described step 5 and step 6:

Determine the sequence number of t spatial sub-channel by following formula according to the transmission gain of object vector and spatial sub-channel: ${\tilde{k}}_t=\arg \underset{k\&Element;C}{\min }({\mathrm{\&psi;}}_{t,k}+{\mathrm{\&delta;}}_k),$ Wherein,

SNR equals

P _TBe base station transmitting power,

The expression noise power.

Images