CN102833047B - Multi-user precoding method in coordinated multi-point system - Google Patents

Abstract

The invention discloses a multi-user precoding method in a coordinated multi-point system and mainly solves the problems of large average feedback cost and average return cost in the prior art. The multi-user precoding method comprises the following implementing steps that: a random vector is utilized to carry out quantization and generate a codebook and the codebook is prestored at user sides and a base station side; each user obtains a downlink channel vector of the user by channel estimation and calculates a single-link channel direction; each user selects a codeword serial number of the single-link channel direction and a corresponding channel quality indication according to the minimum distance criterion; each user compares the channel quality indication of the user with a preset threshold value to obtain a downlink feedback index vector of the user; a base station carries out packet scheduling on the users according to the feedback index vectors to obtain a final set of the users to be served; and the base station reconstructs a downlink channel matrix of the set of the users to implement the zero-forcing precoding of a transmitting terminal. According to the invention, the selective limiting feedback and the packet scheduling are combined, the average feedback cost and the average return cost of the system are greatly reduced and the multi-user precoding method can be used for the coordinated multi-point system with low cost.

Description

Method for multi-user pre-coding in Synergistic multi-point system

Technical field

The invention belongs to communication technical field, particularly relate to the method for multi-user pre-coding in a kind of Synergistic multi-point system.

Background technology

In order to satisfied enhancing Long Term Evolution lte-a system is in the new demand of frequency, bandwidth, peak rate, average throughput, edge throughput, time delay and compatible each side, third generation partner program 3GPP proposes much new key technology in enhancing Long Term Evolution lte-a system, and cooperative multipoint transmission technology is exactly one of them.

Strengthening Long Term Evolution lte-a system adopts orthogonal frequency-time multiple access OFDMA as its descending multiple access technique, namely utilizes orthogonal subcarrier to distinguish different user, thus makes almost do not have user to disturb in community.But be that in the multi-cell system of 1, presence of intercell interference ICI still exists at frequency duplex factor as one, and limit the further raising of community entire throughput and edge user throughput to a great extent.In order to address this problem, the tissue such as third generation partner program 3GPP successively proposes and passes multiple technologies through discussion, as power control, channeling, random disturbances elimination etc. flexibly.Although these technology can improve edge user throughput, often to sacrifice community entire throughput be cost.

Coordinate multipoint CoMP is by the dynamic cooperative between cooperative base station, share the information of some necessity, as schedule information, channel condition information CSI, data message etc., effectively can suppress presence of intercell interference ICI, change interference for useful signal, improve cell throughout on the whole, especially edge user throughput.If but all cooperative base station serve multiple user on a running time-frequency resource, will produce presence of intercell interference ICI simultaneously, therefore base station end needs to use associating precoding to suppress presence of intercell interference ICI.Although precoding can suppress presence of intercell interference ICI more effectively, prerequisite is the channel information that all cooperative base station can obtain multiple user exactly.Under FDD pattern, base station can obtain descending channel information by user feedback, thus effectively suppresses presence of intercell interference ICI.

In " Asymptotic Capacity of Beamforming with Limited Feedback " that W.Santipach in 2004 etc. deliver at " IEEE International Symposium on Information Theory " " information theory international symposium of International Electrical Electronic Engineering Association " (in June, 2004-July) the progressive capacity of beam forming " in the Limited Feedback ", the limited feedback method that a kind of random vector quantizes RVQ is proposed; A kind of limited feedback method utilizing the packing of Jim Glassman subspace of " Limited feedback unitary precoding for spatial multiplexing systems " " the Limited Feedback unitary matrice precoding of space multiplexing system " middle proposition that D.J.Love in 2005 etc. deliver at " IEEE Trans.on Information Theory " " International Electrical Electronic Engineering Association information theory transactions " (volume August the 51st in 2005).Above-mentioned two kinds of methods are all utilize size for N=2 ^lcode book, vector quantization is carried out to down channel vector, wherein L bit number needed for code book.So, user does not need direct feedback of down channel vector, only needs L bit to feed back corresponding code word sequence number, although the method brings certain performance loss, but greatly reduces average feedback expense.The code book of design is larger, then performance is better, but average feedback expense is also larger; On the contrary, if code book is less, then feed back less, but performance is also poorer.

A kind of method of user's selectivity feedback of down channel vector is proposed in " the selectivity limited feedback method for cooperative multicast system " that Li Jing in 2011 delivers at Chinese patent.The method is by presetting a threshold value γ _th, then user i is according to the single-link channel vector h estimated _ij, calculate the signal to noise ratio of base station j to its single-link channel gain then with threshold value γ _thcompare, if γ _ij< γ _th, then corresponding single-link channel vector h is not fed back _ij; Otherwise, then corresponding single-link channel vector h is fed back _ij, wherein i=1 ..., K, j=1 ..., B, K are number of users, and B is base station number, and ρ is signal to noise ratio, and υ is the quantization error factor.The method does quantification treatment to down channel vector, and optionally feedback link channel vector, although the average feedback expense of system can be reduced to a certain extent, with sacrificial system performance for cost.In addition, owing to still needing the data message of mutual all users to be serviced between cooperative base station, therefore the average passback expense of the method is very large.

Summary of the invention

The object of the invention is to, for above-mentioned the deficiencies in the prior art, propose the method for multi-user pre-coding in a kind of Synergistic multi-point system, to realize, while guarantee throughput of system is higher, reduce further the average feedback expense of system and on average returning expense.

The technical scheme realizing the object of the invention is: selectivity Limited Feedback and packet scheduling are combined, the single-link channel quality first being calculated it by user indicates, again it is compared with predetermined threshold value, obtain its downstream feedback indices vector and send to base station, divide into groups to user according to the feedback indices vector received in base station, in each subgroup, carry out user scheduling respectively, obtain user to be serviced, and then treat service-user and carry out precoding.Concrete steps comprise as follows:

(1) random vector is utilized to quantize to generate code book Ω={ w ₁... w _p..., w _n, and be pre-stored in user side and base station end, wherein w _pfor p the code word of code book Ω, p=1 ..., N, N=16 are codebook size;

(2) user i carries out channel estimating to all base stations to its down link, obtains its down channel vector: h _i=[h _i1 ^t..., h _ij ^t..., h _iB ^t] ^t, wherein h _ijfor the single-link channel vector of base station j to user i, i=1 ..., U, j=1 ..., B, U are number of users, and B is base station number, () ^tfor transpose operation;

(3) user i is according to single-link channel vector h _ijcalculate corresponding single-link channel direction: wherein, ‖ h _ij‖ ₂for the single-link channel gain of base station j to user i, ‖ ‖ ₂for the computing of mould 2 norm;

(4) user i is according to minimum distance criterion, and the sequence number selecting inner product maximum is as single-link channel direction code word sequence number: ${\mathrm{index}}_{\mathrm{ij}}=\underset{p=1,...,N}{\arg }\max \left|{{\tilde{h}}_{\mathrm{ij}}}^H{w}_p\right|,$ The code word that now this sequence number is corresponding is $c_{\mathrm{ij}}=w_p{|}_{{p=\mathrm{index}}_{\mathrm{ij}}},$ Wherein for maximum operation;

(5) user i is according to single-link channel gain ‖ h _ij‖ ₂, single-link channel direction with code word c _ij, calculate corresponding channel quality instruction:

${\mathrm{\&gamma;}}_{\mathrm{ij}}=\frac{\mathrm{\&rho;}{{\left|\right|h_{\mathrm{ij}}\left|\right|}_2}^2\mathrm{\&upsi;}}{1+\mathrm{\&rho;}{{\left|\right|h_{\mathrm{ij}}\left|\right|}_2}^2(1-\mathrm{\&upsi;})},$

Wherein, ρ is signal to noise ratio, for the quantization error factor, () ^hfor conjugate transpose operation;

(6) channel quality is indicated γ by user i _ijwith predetermined threshold value γ _threlatively, its downstream feedback indices vector v is obtained _i=[v _i1 ^t..., v _ij ^t..., v _iB ^t] ^t, wherein v _ijfor the single-link feedback indices vector of base station j to user i;

(7) base station is according to the downstream feedback indices vector v received _i, will satisfy condition v _i=d _kall users be divided into a kth subgroup E _kin, obtain a kth subgroup E _kcontained user's set:

$U^{\left(k\right)}=\{{u_1}^{\left(k\right)},...,{u_m}^{\left(k\right)},...,{u_{U_k}}^{\left(k\right)}\},$

Wherein, user gathers U ^(k)m element u _m ^(k)satisfy condition: k=1 ..., 2 ^b, m=1 ..., U _k, d _kfor a kth subgroup E _kindices vector, U _kfor a kth subgroup E _kthe number of contained user;

(8) base station first weeds out the 1st subgroup E ₁interior user U ⁽¹⁾, then according to indices vector produce orthogonal set G={g ₁..., g _q..., g _c, and then from orthogonal set G Stochastic choice q element g _qas treating scheduling subgroup collection, wherein q=1 ..., the number of C, C element contained by orthogonal set G, for kth _sindividual subgroup, k _s∈ 2 ..., 2 ^b, s=1 ..., Q, Q are q element g _qthe number of contained subgroup;

(9) base station calculates kth _sindividual subgroup allow the maximum number of user of scheduling and to kth _sindividual subgroup contained user's set carry out packet scheduling, thus obtain final user's set to be serviced wherein M is the number of transmit antennas of base station, and D is kth _sindividual subgroup indices vector the number of contained nonzero value, for kth _sindividual subgroup the number of contained user,

(10) base station first utilizes the code book Ω={ w prestored ₁... w _p..., w _nand the code word sequence number index that receives _ijrecover single-link channel direction: again according to this single-link channel direction gather with user reconstruct down channel matrix obtain final pre-coding matrix T ', wherein for the single-link channel direction of base station j to user i, $i=1,...,U_1\&Element;{\stackrel{\&OverBar;}{U}}^{\left(k_s\right)}.$

Tool of the present invention has the following advantages:

1. the present invention is owing to employing the Limited Feedback based on code book, and user does not directly feed back its down channel vector, but feedback code sequence number, thus reduce the average feedback expense of system;

2. the present invention is owing to employing selectivity feedback, by presetting the threshold value of a channel quality instruction, making user's only feedback fraction code word sequence number, thus reducing the average feedback expense of system;

3. the present invention is owing to defining a kind of feedback judgement index newly, i.e. channel quality instruction with conventional method compare, reduce further the average feedback expense of system;

4. selectivity Limited Feedback and packet scheduling combine by the present invention, to realize, while guarantee throughput of system is higher, reduce further the average feedback expense of system and on average returning expense.

Accompanying drawing explanation

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

Fig. 2 is selectivity feedback stream journey figure of the present invention;

Fig. 3 is packet scheduling sub-process figure of the present invention;

Fig. 4 is ZF pre-coded sub-streams journey figure of the present invention;

Fig. 5 is the illustraton of model of the Synergistic multi-point system that the present invention uses;

Fig. 6 is the simulation comparison figure that the average feedback duty ratio of the present invention and conventional method changes with predetermined threshold value;

Fig. 7 is the simulation comparison figure that the average passback duty ratio of the present invention and conventional method changes with average feedback duty ratio;

Embodiment

The invention will be further described with reference to the accompanying drawings.

The present invention use cooperative multicast system model as shown in Figure 5, wherein base station BS ₁, BS ₂, BS ₃configure many antennas; Each user, due to the restriction of volume, power consumption and hardware complexity factor, can only configure single antenna.Each user first calculate single-link feedback indices vector, according to this feedback indices vector by the information feed back of correspondence to base station BS ₁, BS ₂, BS ₃, then base station BS ₁, BS ₂, BS ₃feedback indices vector according to receiving carries out packet scheduling to each user, obtains final user's set to be serviced, then reconstructs the down channel matrix of this user set, to realize the associating precoding of transmitting terminal.

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

Step 1, utilizes random vector to quantize to generate code book Ω={ w ₁... w _p..., w _n, and be pre-stored in user side and base station end, wherein w _pfor p the code word of code book Ω, p=1 ..., N, N=16 are codebook size.

Step 2, user i carries out channel estimating to all base stations to its down link, obtains its down channel vector: h _i=[h _i1 ^t..., h _ij ^t..., h _iB ^t] ^t, wherein h _ijfor the single-link channel vector of base station j to user i, i=1 ..., U, j=1 ..., B, U are number of users, and B is base station number, () ^tfor transpose operation.

The user that realizes of this step carries out channel estimating by following two kinds of algorithms to down link:

One is the channel estimation method based on training sequence, this algorithm idea is that base station sends known training sequence, user carries out initial channel estimating when receiving this training sequence, when base station sends useful information data again, user utilizes initial channel estimation results to carry out a judgement renewal, completes real-time channel estimating;

Two is the channel estimation methods based on frequency pilot sign, this algorithm idea is that known frequency pilot sign is inserted in base station in the useful data sent, user can obtain the channel estimation results of pilot frequency locations, then the channel estimation results of pilot frequency locations is utilized, obtained the channel estimation results of useful data position by interpolation, complete channel estimating.

Step 3, user i is according to single-link channel vector h _ijcalculate corresponding single-link channel direction: wherein, ‖ h _ij‖ ₂for the single-link channel gain of base station j to user i, ‖ ‖ ₂for the computing of mould 2 norm.

Step 4, user i is according to minimum distance criterion, and the sequence number selecting inner product maximum is as single-link channel direction code word sequence number: ${\mathrm{index}}_{\mathrm{ij}}=\underset{p=1,...,N}{\arg }\max \left|{{\tilde{h}}_{\mathrm{ij}}}^H{w}_p\right|,$ The code word that now this sequence number is corresponding is: $c_{\mathrm{ij}}=w_p{|}_{p={\mathrm{index}}_{\mathrm{ij}}},$ Wherein for maximum operation.

Step 5, user i is according to single-link channel gain ‖ h _ij‖ ₂, single-link channel direction with code word c _ij, calculate corresponding channel quality instruction:

${\mathrm{\&gamma;}}_{\mathrm{ij}}=\frac{\mathrm{\&rho;}{{\left|\right|h_{\mathrm{ij}}\left|\right|}_2}^2\mathrm{\&upsi;}}{1+\mathrm{\&rho;}{{\left|\right|h_{\mathrm{ij}}\left|\right|}_2}^2(1-\mathrm{\&upsi;})},$

Wherein, ρ is signal to noise ratio, for the quantization error factor, () ^hfor conjugate transpose operation.

Step 6, channel quality is indicated γ by user i _ijwith predetermined threshold value γ _threlatively, its downstream feedback indices vector v is obtained _i, wherein, predetermined threshold value γ _thaverage feedback expense according to system requirements is arranged.

With reference to Fig. 2, the realization of this step is as follows:

(6a) channel quality is indicated γ by user i _ijwith predetermined threshold value γ _threlatively:

If γ _ij>=γ _th, then the single-link feedback indices vector of base station j to user i now user i is by this single-link feedback indices vector v _ij, code word sequence number index _ijwith single-link channel quality instruction γ _ijall send to base station;

If γ _ij< γ _th, then the single-link feedback indices vector of base station j to user i now user i is only by this single-link feedback indices vector v _ijsend to base station;

(6b) user i is according to single-link feedback indices vector v _ijobtain its downstream feedback indices vector:

V _i=[v _ij ^t..., v _ij ^t..., v _iB ^t] ^t, () ^tfor transpose operation.

Step 7, base station is according to the downstream feedback indices vector v received _i, will satisfy condition v _i=d _kall users be divided into a kth subgroup E _kin, obtain a kth subgroup E _kcontained user's set:

$U^{\left(k\right)}=\{{u_1}^{\left(k\right)},...,{u_m}^{\left(k\right)},...,{u_{U_k}}^{\left(k\right)}\},$

Wherein, user gathers U ^(k)m element u _m ^(k)satisfy condition: k=1 ..., 2 ^b, m=1 ..., U _k, d _kfor a kth subgroup E _kindices vector, U _kfor a kth subgroup E _kthe number of contained user.

Step 8, base station selection treats scheduling subgroup collection.

(8a) base station weeds out the 1st subgroup E ₁indices vector d ₁;

(8b) base station is according to residue subgroup indices vector produce orthogonal set:

G={g ₁，...,g _q，...,g _C}；

The wherein number of q=1 .., C, C element contained by orthogonal set G, any subgroup with indices vector with satisfy condition: for kth _sindividual subgroup, for kth _tindividual subgroup, k _s, k _t∈ 2 ..., 2 ^b, s ≠ t, s, t=1 ..., Q, Q are q element g _qthe number of contained subgroup.

(8c) base station Stochastic choice q element g from orthogonal set G _qas treating scheduling subgroup collection.

Step 9, base station calculates waits to dispatch subgroup allow the maximum number of user of scheduling and packet scheduling is carried out to them, obtain final user's set to be serviced.

With reference to Fig. 3, the realization of this step is as follows:

(9a) base station is by kth _sindividual subgroup interior contained number of users with the maximum number of user allowing to dispatch compare, if then return step (8); If then calculate kth _sindividual subgroup contained user's set $U^{\left(k_s\right)}=\{{u_1}^{\left(k_s\right)},...,{u_m}^{\left(k_s\right)},...,{u_{U_{k_s}}}^{\left(k_s\right)}\}$ Down channel quality instruction:

Wherein, for user's set m element, r is element number contained by set Φ, ∑ () is summation operation;

(9b) sort to the size that this down channel quality indicates in base station, namely then selective channel quality instruction maximum before individual user is as final user's set to be serviced ${\stackrel{\&OverBar;}{U}}^{\left(k_s\right)}=\{{u_{m_1}}^{\left(k_s\right)},...,{u_{m_n}}^{\left(k_s\right)},...,{u_{m_{A_{k_s}}}}^{\left(k_s\right)}\}.$

Step 10, base station recovers single-link channel direction obtain final pre-coding matrix T '.

With reference to Fig. 4, the realization of this step is as follows:

(10a) base station is according to single-link channel direction gather with user reconstruct down channel matrix:

Wherein, for final total number of users to be serviced;

(10b) base station is by down channel matrix obtain pre-coding matrix: wherein t _rfor the column vector of pre-coding matrix, r=1 ..., U ₁, () ^-1for inversion operation;

(10c) base station is to column vector t _rnew column vector is obtained as normalized: thus obtain final pre-coding matrix: $T^{\&prime;}=[{t_1}^{\&prime;},...,{t_r}^{\&prime;},...,{t_{U_1}}^{\&prime;}].$

Advantage of the present invention is further illustrated by following theory analysis and simulation result:

1) theory analysis

By single-link channel direction with p code word w _pseparate isotropism vector, known stochastic variable obey β (M-1,1) distribution, wherein M is the number of transmit antennas of base station, and obtaining cumulative distribution function is F _x(x)=x ^m-1, x ∈ [0,1], the thus quantization error factor probability density function f _υ(x) be:

f _υ(x)＝N(M-1)(1-x) ^M-2(1-(1-x) ^M-1) ^N-1,x∈[0,1]；

By single-link channel vector h _ijit is the known stochastic variable of multiple Gaussian random vector obeying the degree of freedom is the χ of 2M ²distribution, thus the probability density function of stochastic variable h is wherein Γ (M+1)=M × Γ (M);

The single-link channel quality instruction γ of base station j to user i _ijobey same distribution, therefore unification represents with γ.According to above-mentioned f _υ(x) and f _hy the expression formula of () obtains the cumulative distribution function F of channel quality instruction γ _γ(z) be

$F_{\mathrm{\&gamma;}}\left(z\right)=P(\mathrm{\&gamma;}\&le;z)=P(\frac{\mathrm{\&rho;h\&upsi;}}{1+\mathrm{\&rho;h}(1-\mathrm{\&upsi;})}\&le;z)$

$=\frac{1}{\mathrm{\&Gamma;}\left(M\right)}\{{\&Integral;}_0^{\frac{z}{\mathrm{\&rho;}}}{e}^{-y}{y}^{M-1}\mathrm{dy}+\underset{i=0}{\overset{N}{\mathrm{\&Sigma;}}}{C}_N^i{(-1)}^i\frac{z(1+z)}{\mathrm{\&rho;}}{\&Integral;}_0^1\frac{{(1-x)}^{i(M-1)}{e}^{-\frac{z}{\mathrm{\&rho;}[x(1+z)-z]}}{\left(\frac{z}{\mathrm{\&rho;}[x(1+z)-z]}\right)}^{M-1}}{{[x(1+z)-z]}^2}\mathrm{dx}\};$

Definition average feedback duty ratio is

$\stackrel{\&OverBar;}{\mathrm{\&xi;}}=\frac{1}{\mathrm{BU}}{E}_t\left[N\left(t\right)\right],\stackrel{\&OverBar;}{\mathrm{\&xi;}}\&Element;\left[\mathrm{0,1}\right];$

Wherein, N (t) represents whole code word number of all user feedbacks of t, the down channel matrix namely reconstructed the number of contained non-zero column vector, E _t() is mean operation.Obviously, N (t) obeys (BU, p ₁) binomial distribution, by probability $p_1=P(N\left(t\right)=n)=C_{\mathrm{BU}}^n{[1-F_{\mathrm{\&gamma;}}\left(z\right)]}^n{\left[F_{\mathrm{\&gamma;}}\left(z\right)\right]}^{\mathrm{BU}-n},$ Obtain average feedback duty ratio with threshold value γ _thoccluding relation formula:

$\stackrel{\&OverBar;}{\mathrm{\&xi;}}=\frac{1}{\mathrm{BU}}{E}_t\left[N\left(t\right)\right]=\frac{1}{\mathrm{BU}}\underset{n=0}{\overset{\mathrm{BU}}{\mathrm{\&Sigma;}}}\mathrm{nP}(N\left(t\right)=n)=1-F_{\mathrm{\&gamma;}}\left({\mathrm{\&gamma;}}_{\mathrm{th}}\right);$

Because 0≤F _γ(γ _th)≤1, so average feedback duty ratio and if only if all base stations are to the single-link channel quality instruction γ of user _ij>=γ _thtime, equal sign is set up.It can thus be appreciated that, predetermined threshold value γ _thhigher, channel quality instruction cumulative distribution function F _γ(γ _th) larger, thus average feedback duty ratio lower, that is the present invention is fed back by selectivity, reduces the average feedback expense of system.

Definition average passback duty ratio is

$\stackrel{\&OverBar;}{\mathrm{\&eta;}}=\frac{1}{\mathrm{BU}}{E}_t\left[Z\left(t\right)\right],\stackrel{\&OverBar;}{\mathrm{\&eta;}}\&Element;\left[\mathrm{0,1}\right];$

Wherein Z (t) represents the number of users of t all base stations service, namely pre-coding matrix W comprise the number of non-zero column vector.The present invention, by user is carried out packet scheduling, makes the down channel matrix reconstructed for block diagonal matrix, then final pre-coding matrix T ' is also block diagonal matrix, and due to pre-coding matrix T ' comprise non-zero column vector number just equal reconstruct down channel matrix comprise the number of non-zero column vector, i.e. Z (t)=N (t), thus and for conventional method, be constantly equal to 1.It can thus be appreciated that the present invention, by packet scheduling, reduces the average passback expense of system.

Above-mentioned theory analysis shows, the present invention is by arranging suitable threshold value γ _th, average feedback duty ratio can be reduced simultaneously on average return duty ratio and

2) emulation experiment

2.1) simulated conditions: adopt flat Rayleigh bulk nanometer materials, the reception antenna number R=1 of the number of transmit antennas M=2 of base station number B=3, number of users U=3 × 30=90, each base station, each user, codebook size N=16.

2.2) content is emulated

Emulation 1, under the condition of signal to noise ratio snr=10dB, with predetermined threshold value situation of change, Monte-Carlo Simulation is carried out to the average feedback duty ratio of the present invention and conventional method, obtain corresponding simulation comparison figure, as shown in Figure 6, wherein abscissa represents predetermined threshold value, and ordinate represents average feedback duty ratio.As can be seen from Figure 6, when predetermined threshold value increases to 30dB from-10dB, corresponding average feedback duty ratio is reduced to 0 from 3, illustrates that the present invention can reduce the average feedback expense of system by improving predetermined threshold value.In addition, be 10dB place in predetermined threshold value, the present invention decreases 0.3 relative to its average feedback duty ratio of conventional method.

Emulation 2, with average feedback duty ratio situation of change, Monte-Carlo Simulation is carried out to the average passback duty ratio of the present invention and conventional method, obtains corresponding simulation comparison figure, as shown in Figure 7, wherein abscissa represents average feedback duty ratio, and ordinate represents and on average returns duty ratio.As can be seen from Figure 7, for conventional method, average passback duty ratio is constantly equal to 1, and for user grouping dispatching method of the present invention, average passback duty ratio is less than or equal to 1, and with the linear growth of average feedback duty ratio, fits like a glove with above-mentioned theory analysis.

Simulation result shows, the present invention can reduce the average feedback expense of system simultaneously and on average return expense.

Claims (3)

1. the method for multi-user pre-coding in Synergistic multi-point system, comprises the steps:

(1) random vector is utilized to quantize to generate code book Ω={ w ₁... w _p..., w _n, and be pre-stored in user side and base station end, wherein w _pfor p the code word of code book Ω, p=1 ..., N, N=16 are codebook size;

(2) user i carries out channel estimating to all base stations to its down link, obtains its down channel vector: wherein h _ijfor the single-link channel vector of base station j to user i, i=1 ..., U, j=1 ..., B, U are number of users, and B is base station number, () ^tfor transpose operation;

(3) user i is according to single-link channel vector h _ijcalculate corresponding single-link channel direction: wherein, || h _ij|| ₂for the single-link channel gain of base station j to user i, || || ₂for the computing of mould 2 norm;

(4) user i is according to minimum distance criterion, and the sequence number selecting inner product maximum is as single-link channel direction code word sequence number: ${\mathrm{index}}_{\mathrm{ij}}=\underset{p=1,...,N}{\arg }\max \left|{{\tilde{h}}_{\mathrm{ij}}}^H{w}_p\right|,$ The code word that now this sequence number is corresponding is $c_{\mathrm{ij}}=w_p{|}_{p=\mathrm{inde}{x}_{\mathrm{ij}}},$ Wherein for maximum operation;

(5) user i is according to single-link channel gain || h _ij|| ₂, single-link channel direction with code word c _ij, calculate corresponding channel quality instruction:

${\mathrm{\&gamma;}}_{\mathrm{ij}}=\frac{\mathrm{\&rho;}{{\left|\right|h_{\mathrm{ij}}\left|\right|}_2}^2\mathrm{\&upsi;}}{1+\mathrm{\&rho;}{{\left|\right|h_{\mathrm{ij}}\left|\right|}_2}^2(1-\mathrm{\&upsi;})},$

Wherein, ρ is signal to noise ratio, for the quantization error factor, () ^hfor conjugate transpose operation;

(6) channel quality is indicated γ by user i _ijwith predetermined threshold value γ _threlatively, its downstream feedback indices vector is obtained wherein v _ijfor the single-link feedback indices vector of base station j to user i;

(7) base station is according to the downstream feedback indices vector v received _i, will satisfy condition v _i=d _kall users be divided into a kth subgroup E _kin, obtain a kth subgroup E _kcontained user's set:

$U^{\left(k\right)}=\{{u_1}^{\left(k\right)},...,{u_m}^{\left(k\right)},...,{u_{U_k}}^{\left(k\right)}\},$

Wherein, user gathers m the element u of U (k) _m ^(k)satisfy condition: k=1 ..., 2 ^b, m=1 ..., U _k, d _kfor a kth subgroup E _kindices vector, U _kfor a kth subgroup E _kthe number of contained user;

(8) base station first weeds out the 1st subgroup E ₁interior indices vector d ₁, then according to residue subgroup indices vector produce orthogonal set G={g ₁..., g _q..., g _c, and then from orthogonal set G Stochastic choice q element g _qas treating scheduling subgroup collection, wherein q=1 ..., the number of C, C element contained by orthogonal set G, for kth _sindividual subgroup, k _s∈ 2 ..., 2 ^b, s=1 ..., Q, Q are q element g _qthe number of contained subgroup;

(9) base station calculates kth _sindividual subgroup allow the maximum number of user of scheduling and to kth _sindividual subgroup contained user's set carry out packet scheduling, thus obtain final user's set to be serviced wherein M is the number of transmit antennas of base station, and D is kth _sindividual subgroup indices vector the number of contained nonzero value, for kth _sindividual subgroup the number of contained user,

(10) base station first utilizes the code book Ω={ w prestored ₁... w _p..., w _nand the code word sequence number indexi that receives _jrecover single-link channel direction: again according to this single-link channel direction gather with user reconstruct down channel matrix obtain final pre-coding matrix T ', wherein for the single-link channel direction of base station j to user i, $i=1,...,U_1\&Element;{\stackrel{\&OverBar;}{U}}^{\left(k_s\right)}.$

2. the method for multi-user pre-coding in Synergistic multi-point system according to claim 1, wherein described in step (9) to kth _sindividual subgroup contained user's set carry out packet scheduling, carry out as follows:

(9a) base station is by kth _sindividual subgroup interior contained number of users with the maximum number of user allowing to dispatch compare, if then return step (8); If then calculate kth _sindividual subgroup contained user's set $U^{\left(k_s\right)}=\{{u_1}^{\left(k_s\right)},...,{u_m}^{\left(k_s\right)},...,{u_{U_{k_s}}}^{\left(k_s\right)}\}$ Down channel quality instruction:

Wherein, for user's set m element, r is element number contained by set Φ, Σ () is summation operation;

(9b) sort to the size that this down channel quality indicates in base station, namely then selective channel quality instruction maximum before individual user is as final user's set to be serviced ${\stackrel{\&OverBar;}{U}}^{\left(k_s\right)}=\{{u_{m_1}}^{\left(k_s\right)},...,{u_{m_n}}^{\left(k_s\right)},...,{u_{m_{A_{k_{\text{s}}}}}}^{\left(k_s\right)}\}.$

3. the method for multi-user pre-coding in Synergistic multi-point system according to claim 1, wherein described in step (10) according to this single-link channel direction gather with user reconstruct down channel matrix obtain final pre-coding matrix T ', carry out as follows:

(10a) base station is according to single-link channel direction gather with user reconstruct down channel matrix:

Wherein, $i\&Element;{\stackrel{\&OverBar;}{U}}^{\left(k_s\right)},U_1=\underset{s=1}{\overset{Q}{\mathrm{\&Sigma;}}}{U}_{k_s}$ For final total number of users to be serviced;

(10b) base station is by down channel matrix obtain pre-coding matrix: wherein t _rfor the column vector of pre-coding matrix, r=1 ..., U ₁, () ^-1for inversion operation;

(10c) base station is to column vector t _rnew column vector is obtained as normalized: thus obtain final pre-coding matrix: $T^{\&prime;}=[{t_1}^{\&prime;},...,{t_r}^{\&prime;},...,{t_{U_1}}^{\&prime;}].$