CN103067062B - Base station antenna selecting method based on interference alignment in multi-cell system - Google Patents

Abstract

The invention discloses a base station antenna selecting method based on interference alignment in a multi-cell system, and mainly solves the problems that an existing multi-antenna interference alignment technology is relatively high in system hardware cost, and the base station antenna selecting method based on interference alignment can only be used in a network comprising 3 cells. The method comprises the following implementation steps of: estimating state information of channels from various base stations to users in the resident cells; initiating base station antenna selection schemes of the various cells; along a system and capacity increase optimizing direction, under the condition that other base station antenna selection schemes are unchanged, determining the optimal base station selection schemes of the various cells in sequence by using partial iterative interference alignment; and performing multi-cell interference coordination by adopting traditional iterative interference alignment. The base station antenna selecting method is applicable to an LTE (Long Term Evolution) network oriented to commercial use with a plurality of cells, is low in computation complexity, can reduce the system hardware cost, and realizes compromise between system and capacity and computation complexity.

Description

Based on the antenna for base station system of selection of interference alignment in multi-cell system

Technical field

The invention belongs to wireless communication technology field, further relate to the antenna for base station system of selection based on interference alignment in a kind of multi-cell system, can be used for the LTE network of business.

Background technology

In order to improve the capacity of cellular system, be that the Next-Generation Wireless Communication Systems of representative requires that the full rate realizing neighbor cell is as far as possible multiplexing with LTE network, i.e. the spectrum reuse factor is 1.Therefore, relative to noise and decline, presence of intercell interference becomes the principal element of influential system performance.How effectively carrying out interference coordination between multiple cell, the interference particularly eliminating Cell Edge User causes to be paid close attention to widely.Interference alignment is a kind of interference coordination technique proposed recent years, interference signal from other base stations can be snapped to a sub spaces in Received signal strength space by it at receiver user, useful signal from respective base station is projected to another interference-free subspace, thus carry out effective AF panel, improve power system capacity.

The scholars such as K.Gomadam are at article " Approaching the Capacity of Wireless Networksthrough Distributed Interference Alignment " (Proceedings of the IEEE GlobalTelecommunications Conference, Miami, USA, propose a kind of distributed iterative interference alignment schemes (being called conventional iterative interference alignment) 2008:1-5), be applicable to the interference coordination of multiple cell multiple-input and multiple-output (MIMO) network.The pre-coding matrix of the method first random initializtion base station, then reveals the AF panel matrix of minimum principle Iterative Design user and the pre-coding matrix of base station until convergence according to interference.The weak point that the method exists is: although the antenna number of user is generally less in mimo system, but a fairly large number of antenna of base station possible configuration, the method does not consider the increase along with antenna for base station quantity, the radio frequency link comprising the elements such as digital-to-analogue/analog to digital converter, low-converter, low noise amplifier also will increase gradually, thus increases the problem of system hardware cost.

The people such as J.G.Klotz are at article " Antenna Selection Criteria for Interference Alignment " (Proceedings of IEEE International Symposium on Personal, Indoor and Mobile RadioCommunications, Istanbul, Turkey, the antenna for base station system of selection for 3 cell systems adopting non-iterative interference alignment is proposed 2010:527-531), to reduce system hardware cost, obtain system and volumetric properties lifting simultaneously.The main performing step of the method is: the first, all skies line options situation that all alternative antenna sub-combinations utilizing poor searching method to travel through 3 base stations become; The second, for each day line options situation, carry out non-iterative interference alignment, and the useful and interference signal vector space chordal distance of computing system and capacity or each user with; 3rd, according to maximization system and capacity criterion or maximize the antenna selecting plan that chordal distance criterion selects the most each base station optimum, the antenna chosen is connected on corresponding radio frequency link; 4th, realize non-iterative interference alignment.The weak point that the method exists is: in view of non-iterative interference alignment can only complete the interference coordination of 3 communities, this method is only applicable to the system comprising 3 communities, therefore uses and has limitation, can not expand to the system of multiple cell.

Summary of the invention

The object of the invention is to overcome above-mentioned the deficiencies in the prior art, antenna for base station system of selection based on interference alignment in a kind of multi-cell system is proposed, reduce system hardware cost, obtain antenna selection gain, simultaneously by range of application from 3 cell extension to multiple community.

For achieving the above object, the thinking of the inventive method is:

For carrying out the fixing and not more feature of the multiplexing number of cells of full rate in LTE system, the inventive method adopts iteration interference alignment to carry out the interference coordination of multiple cell system; Owing to determining that the antenna for base station selection scheme of each community optimum is a combinatorial optimization problem, solve this combinatorial optimization problem to need repeatedly to search for the various skies line options situation be combined into by each cell base station, therefore, in each day line options situation search procedure, adopt the conventional iterative interference alignment comprising successive ignition that the computation complexity of method can be caused higher; Following two modes of the inventive method employing reduce the computation complexity in the number of times and each search procedure searching for various skies line options situation respectively: first, along the optimal anchor direction that system and capacity increase, when keeping other antenna for base station selection schemes constant, successively the antenna for base station selection scheme of each community is optimized, thus reduction solves in combinatorial optimization problem the number of times searching for various skies line options situation; The second, when optimizing the antenna for base station selection scheme of each community, adopting the part iteration interference alignment only carrying out several times iteration, then computing system and capacity, instead of completing conventional iterative interference alignment, thus reducing the computation complexity of each search procedure; According to the antenna for base station selection scheme of each community optimum, the inventive method finally adopts conventional iterative to disturb alignment to realize interference coordination between multiple cell.

Concrete steps of the present invention are as follows:

(1) channel condition information of each base station to this community user is estimated

Base station k sends pilot frequency information to this community user k, and user k estimates the channel condition information of k each antenna in base station to self, and feeds back to base station k, wherein k=1 ..., B, B are community number; The channel condition information that base station k feeds back according to this community user, determines channel matrix wherein N and M represents the antenna number of base station k and user k respectively, represent complex field;

(2) each cell-site antenna selection scheme of initialization

The antenna selecting plan ω of initialization base station k _kfor having the matrix of maximum F-norm corresponding alternative antenna selects vector, ω _kcomputing formula as follows:

${\mathrm{\ω}}_k=\arg \underset{l\∈\{1,...,C_N^{N_f}\}}{\max }{|\left|H_{\mathrm{kk}}^{{\mathrm{\φ}}_l}\right||}_F^2,$

Wherein, k=1 ..., B, represent when antenna selecting plan is φ _ltime according to channel matrix H _kkthe base station k determined to the channel matrix of this community user, φ _lrepresent that l alternative antenna of base station selects vector, for alternative antenna selects the index variables of vector, N _ffor the rf chain way of base station, represent and choose N from N number of number _fthe number of combinations of number, || || _fthe F-norm of representing matrix; The antenna selecting plan of each base station is combined into B cell-site antenna and selects set omega={ ω ₁, ω ₂..., ω _b;

(3) along the optimal anchor direction that system and capacity increase, when keeping other antenna for base station selection schemes constant, part iteration is utilized to disturb alignment to determine the antenna for base station selection scheme of each community optimum successively:

3.1) initialization alternative antenna selects the index variables l of vector to be 1, the system that initialization is simultaneously maximum and capacity C _maxbe 0;

3.2) by the antenna for base station selection scheme ω of community 1 ₁be updated to alternative antenna and select vectorial φ _l, keep the antenna selecting plan of other base stations constant, selected by B cell-site antenna the 1st element in set omega to be updated to φ simultaneously _l;

3.3) according to above-mentioned antenna selecting plan, carry out part iteration interference alignment, design base station pre-coding matrix and user's AF panel matrix of each community, be implemented as follows:

3.3a) the pre-coding matrix of random initializtion base station k wherein k=1 ..., B, d represent that base station sends the dimension of data flow;

3.3b) initialization iterations indicator variable c is 1;

3.3c) determine the interference covariance matrix of the user k ' when the c time iteration wherein k '=1 ..., B;

3.3d) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the AF panel matrix of design user k ' when the c time iteration wherein k '=1 ..., B, υ _dd the characteristic value characteristic of correspondence vector that { } representing matrix is minimum;

3.3e) according to above-mentioned AF panel matrix determine the interference covariance matrix of the base station k when the c time iteration wherein k '=1 ..., B, k=1 ..., B;

3.3f) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the pre-coding matrix of design base station k when the c time iteration wherein k=1 ..., B;

3.3g) value of iterations indicator variable c adds 1, if c≤N _ite, then return step 3.3c, proceed iteration; Otherwise, jump to step 3.4, obtain N _itethe pre-coding matrix of base station k after secondary iteration with the AF panel matrix of user k ' wherein k=1 ..., B, k '=1 ..., B, N _itefor the iterations of part iteration interference alignment;

3.4) equivalent channel of base station k ' to user k ' is determined and the interference covariance matrix of user k ' wherein represent at antenna selecting plan ω _{k '}in situation, base station k ' is to the channel matrix of user k ', represent N _itethe pre-coding matrix of base station k ' after secondary iteration, k '=1 ..., B;

3.5) according to the equivalent channel that part iteration interference alignment obtains aF panel matrix and interference covariance matrix computing system and capacity wherein for the capacity of user k ', wherein k '=1 ..., B;

3.6) if C ^Ω> C _max, then C is set _max=C ^Ω, and the alternative antenna of order optimum selects the index variables l of vector ^*=l, namely optimum alternative antenna selects vector otherwise, C _maxremain unchanged;

3.7) alternative antenna selects the value of the index variables l of vector to add 1, if then return step 3.2, continue other alternative antenna of traversal and select vector; Otherwise, skip to step 3.8;

3.8) the antenna for base station selection scheme of community 1 optimum is determined

3.9) according to the method described in step 3.1 to 3.8, the antenna for base station selection scheme of community 2 to community B optimum is determined successively ${\mathrm{\ω}}_2^{*},{\mathrm{\ω}}_3^{*},...,{\mathrm{\ω}}_B^{*};$

3.10) the antenna for base station selection scheme of above-mentioned each community optimum is combined into B optimum cell-site antenna and selects set ${\mathrm{\Ω}}^{*}=\{{\mathrm{\ω}}_1^{*},{\mathrm{\ω}}_2^{*},...,{\mathrm{\ω}}_B^{*}\};$

(4) conventional iterative interference alignment is adopted to carry out multiple cell interference coordination:

4.1) set omega is selected according to B cell-site antenna of above-mentioned optimum ^*, carrying out iterations is N _iconventional iterative interference alignment, design base station pre-coding matrix and user's AF panel matrix of each community;

4.2) according to pre-coding matrix and the AF panel matrix of above-mentioned conventional iterative interference alignment design, each cell base station and user transmit and receive data respectively, carry out multiple cell interference coordination, eliminate presence of intercell interference.

Tool of the present invention has the following advantages:

1. the present invention disturbs alignment to carry out multiple cell interference coordination owing to utilizing iteration, and the antenna for base station realizing each community optimum is selected, overcome the existing antenna for base station system of selection based on interference alignment and can only comprise the shortcoming used in 3 subzone networks, by range of application from 3 cell extension to multiple community, be applicable to the LTE network of the business with multiple community;

2. the optimal anchor direction of the present invention owing to increasing along system and capacity, when keeping the antenna for base station selection scheme of other communities constant, determine the antenna for base station selection scheme of each community optimum successively, decrease the number of times of search various skies line options situation, and the part iteration adopting iterations less in the search procedure of each day line options situation interference alignment, thus significantly reduce computation complexity;

3. the present invention is owing to selecting antenna for base station, decreases the rf chain way of base station, thus reduces system hardware cost, obtains antenna selection gain simultaneously;

4. the present invention adjustment member iteration can disturb the iterations alignd, thus reaches the compromise of system and capacity and computation complexity.

Accompanying drawing explanation

Fig. 1 is the B cell system illustraton of model that the present invention is suitable for;

Fig. 2 is general flow chart of the present invention;

Fig. 3 is the sub-process figure of the antenna for base station selection scheme determining each community optimum in the present invention;

Fig. 4 is the sub-process figure of part iteration interference alignment in the present invention;

Fig. 5 is the simulation comparison figure that the system of the present invention and random antenna for base station system of selection and poor search base station antenna selecting method and capacity change with signal to noise ratio.

Embodiment

Below in conjunction with accompanying drawing, further detailed description is done to the present invention.

With reference in Fig. 1, B cell system, a user is served in the base station of each community, and antenna number and the rf chain way of base station are respectively N and N _f, antenna number and the rf chain way of user are M.The sequence number of community represents with k, and k=1 ..., B, then base station k and user k represents base station and the user of community k respectively.Each community is had to the system of multiple user, such as time-multiplexed mode can be adopted to be reduced to the system that each time slot only has a user.Suppose N > N _f, each base station needs to choose N from all N number of antennas _findividual antenna, and be connected on corresponding radio frequency link, therefore the alternative antenna set Φ of base station comprises individual alternative antenna selects vector, can be expressed as wherein represent and choose N from N number of number _fthe number of combinations of number, φ _lrepresent that l alternative antenna selects vector, it is the long column vector of N, and its element value is 0 or 1, and l is the index variables that alternative antenna selects vector.In column vector, to be the element of i be sequence number that 1 to represent i-th antenna of base station selected, otherwise not selected. represent the channel matrix of base station k to user k ', wherein ω _krepresent the antenna selecting plan that base station k chooses from alternative antenna selection set, represent complex field.

With reference to Fig. 2, performing step of the present invention comprises as follows:

Step 1, estimates the channel condition information of each base station to this community user.

1a) base station k sends to user k secondary pilot frequency information, chooses some antennas each time and they is connected to radio frequency link transmission pilot frequency information, wherein from all N number of antennas, choosing N the 1st time _findividual antenna, the 2nd time from remaining N-N _fn is chosen in individual antenna _findividual antenna, the 3rd time from remaining N-2N _fn is chosen in individual antenna _findividual antenna, until the secondaryly choose last remaining antenna, thus guarantee that all antennas all can once be connected to radio frequency link at certain and send pilot frequency information, wherein k=1 ..., B, representative rounds up to scalar x;

1b) user k estimates the channel condition information of base station k respective antenna to self each antenna by the pilot frequency information received at every turn, and by corresponding channel matrix feedback to base station k, wherein k=1 ..., B;

1c) base station k obtains its all antenna to the channel coefficients of each antenna of user according to the channel matrix of each feedback, and channel coefficients is formed channel matrix wherein k=1 ..., B.

Step 2, each cell-site antenna selection scheme of initialization.

2a) according to channel matrix H _kkvectorial φ is selected with alternative antenna _l, determine to work as φ _las base station k during antenna selecting plan to the channel matrix of this community user

Wherein, k=1 ..., B, diag{ φ _lrepresent that the element on a leading diagonal is column vector φ _lthe diagonal matrix of corresponding element, function represent the submatrix be made up of the non-zero row of matrix;

2b) the antenna selecting plan ω of initialization base station k _kfor the matrix of maximum F-norm corresponding alternative antenna selects vector, and computing formula is as follows:

${\mathrm{\ω}}_k=\arg \underset{l\∈\{1,...,C_N^{N_f}\}}{\max }{|\left|H_{\mathrm{kk}}^{{\mathrm{\φ}}_l}\right||}_F^2,$

Wherein, k=1 ..., B, || || _fthe F-norm of representing matrix;

2c) antenna selecting plan of above-mentioned each base station is combined into B cell-site antenna and selects set omega={ ω ₁, ω ₂..., ω _b.

Step 3, along the optimal anchor direction that system and capacity increase, when keeping other antenna for base station selection schemes constant, utilizes part iteration to disturb alignment to determine the antenna for base station selection scheme of each community optimum successively.

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

3.1) initialization alternative antenna selects the index variables l of vector to be 1, the system that initialization is simultaneously maximum and capacity C _maxbe 0;

3.2) by the antenna for base station selection scheme ω of community 1 ₁be updated to alternative antenna and select vectorial φ _l, keep the antenna selecting plan of other base stations constant, selected by B cell-site antenna the 1st element in set omega to be updated to φ simultaneously _l;

3.3) according to above-mentioned antenna selecting plan, carry out part iteration interference alignment, design base station pre-coding matrix and user's AF panel matrix of each community:

With reference to Fig. 4, being implemented as follows of this step:

3.3a) the pre-coding matrix of random initializtion base station k and meet wherein k=1 ..., B, () ^hthe conjugate transpose of representing matrix, I _nrepresent the unit matrix of n × n dimension, d represents that base station sends the dimension of data flow;

3.3b) initialization iterations indicator variable c is 1;

3.3c) according to scholars such as K.Gomadam at article " Approaching the Capacity of WirelessNetworks through Distributed Interference Alignment " (Proceedings of the IEEE GlobalTelecommunications Conference, Miami, USA, training method 2008:1-5), determines the interference covariance matrix of the user k ' when the c time iteration

$Q_{k^{\′},c}^{\mathrm{\Ω}}=\underset{k=1,k\≠k^{\′}}{\overset{B}{\mathrm{\Σ}}}\frac{P_k}{d}{H}_{k^{\′}k}^{{\mathrm{\ω}}_k}{V}_{k,c-1}^{\mathrm{\Ω}}{\left(V_{k,c-1}^{\mathrm{\Ω}}\right)}^H{\left(H_{k^{\′}k}^{{\mathrm{\ω}}_k}\right)}^H,$

Wherein, k '=1 ..., B, P _krepresent the transmitting power of base station k, represent the pre-coding matrix of the base station k when the c-1 time iteration;

3.3d) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the AF panel matrix of design user k ' when the c time iteration wherein k '=1 ..., B, υ _dd the characteristic value characteristic of correspondence vector that { } representing matrix is minimum;

3.3e) according to scholars such as K.Gomadam at article " Approaching the Capacity of WirelessNetworks through Distributed Interference Alignment " (Proceedings of the IEEE GlobalTelecommunications Conference, Miami, USA, training method 2008:1-5), determines the interference covariance matrix of the base station k when the c time iteration

${\stackrel{\‾}{Q}}_{k,c}^{\mathrm{\Ω}}=\underset{k^{\′}=1,k^{\′}\≠k}{\overset{B}{\mathrm{\Σ}}}\frac{P_{k^{\′}}}{d}{\left(H_{k^{\′}k}^{{\mathrm{\ω}}_k}\right)}^H{U}_{k^{\′},c}^{\mathrm{\Ω}}{\left(U_{k^{\′},c}^{\mathrm{\Ω}}\right)}^H{H}_{k^{\′}k}^{{\mathrm{\ω}}_k},$

Wherein, k=1 ..., B, P _{k '}represent the transmitting power of user k ', value is equal with the transmitting power of this cell base station;

3.3f) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the pre-coding matrix of design base station k when the c time iteration wherein k=1 ..., B;

3.3g) value of iterations indicator variable c adds 1, if c≤N _ite, then return step 3.3c, proceed iteration; Otherwise, jump to step 3.4, obtain N _itethe pre-coding matrix of base station k after secondary iteration with the AF panel matrix of user k ' wherein k=1 ..., B, k '=1 ..., B, N _itefor the iterations of part iteration interference alignment, and meet N _ite≤ N _i, N _irepresent and be preferably 30 by the iterations empirical value that tradition interference alignment convergence needs;

3.4) according to scholars such as K.Gomadam at article " Approaching the Capacity of WirelessNetworks through Distributed Interference Alignment " (Proceedings of the IEEE GlobalTelecommunications Conference, Miami, USA, training method 2008:1-5), determines the equivalent channel of base station k ' to user k ' and the interference covariance matrix of user k '

$Q_{k^{\′}}^{\mathrm{\Ω}}=\underset{k=1,k\≠k^{\′}}{\overset{B}{\mathrm{\Σ}}}\frac{P_k}{d}{H}_{k^{\′}k}^{{\mathrm{\ω}}_k}{V}_{k,N_{\mathrm{ite}}}^{\mathrm{\Ω}}{\left(V_{k,N_{\mathrm{ite}}}^{\mathrm{\Ω}}\right)}^H{\left(H_{k^{\′}k}^{{\mathrm{\ω}}_k}\right)}^H,$

Wherein, k '=1 ..., B, wherein represent at antenna selecting plan ω _{k '}in situation, base station k ' is to the channel matrix of user k ';

3.5) according to the equivalent channel that part iteration interference alignment obtains aF panel matrix and interference covariance matrix computing system and capacity C ^Ω:

$C^{\mathrm{\Ω}}=\underset{k^{\′}=1}{\overset{B}{\mathrm{\Σ}}}{C}_{k^{\′}}^{\mathrm{\Ω}}=\underset{k^{\′}=1}{\overset{B}{\mathrm{\Σ}}}{\log }_2|I_d+\frac{P_{k^{\′}}{\left(U_{k^{\′},N_{\mathrm{ite}}}^{\mathrm{\Ω}}\right)}^H{H}_{k^{\′}{k}^{\′}}^{{\mathrm{\ω}}_{k^{\′}}}{V}_{k^{\′},N_{\mathrm{ite}}}^{\mathrm{\Ω}}{\left(V_{k^{\′},N_{\mathrm{ite}}}^{\mathrm{\Ω}}\right)}^H{\left(H_{k^{\′}{k}^{\′}}^{{\mathrm{\ω}}_{k^{\′}}}\right)}^{\text{H}}{U}_{k^{\′}{,N}_{\mathrm{ite}}}^{\mathrm{\Ω}}/d}{{\left(U_{k^{\′}{,N}_{\mathrm{ite}}}^{\mathrm{\Ω}}\right)}^H({\mathrm{\σ}}^2{I}_M+Q_{k^{\′}}^{\mathrm{\Ω}})U_{k^{\′},N_{\mathrm{ite}}}^{\mathrm{\Ω}}}|,$

Wherein, k '=1 ..., B, σ ²the variance of white Gaussian noise in channel, || the determinant of representing matrix, bodge is bit/s/Hz;

3.6) if C ^Ω> C _max, then C is set _max=C ^Ω, and the alternative antenna of order optimum selects the index variables l of vector ^*=l, namely optimum alternative antenna selects vector otherwise, C _maxremain unchanged;

3.7) alternative antenna selects the value of the index variables l of vector to add 1, if then return step 3.2, other alternative antenna continued in traversal base alternative antenna set Φ select vector; Otherwise, skip to step 3.8;

3.8) the antenna for base station selection scheme of community 1 optimum is determined

3.9) according to the method described in step 3.1 to 3.8, the antenna for base station selection scheme of community 2 to community B optimum is determined successively ${\mathrm{\ω}}_2^{*},{\mathrm{\ω}}_3^{*},...,{\mathrm{\ω}}_B^{*};$

3.10) the antenna for base station selection scheme of above-mentioned each community optimum is combined into B optimum cell-site antenna and selects set ${\mathrm{\Ω}}^{*}=\{{\mathrm{\ω}}_1^{*},{\mathrm{\ω}}_2^{*},...,{\mathrm{\ω}}_B^{*}\}.$

Step 4, adopts conventional iterative interference alignment to carry out multiple cell interference coordination.

4.1) set omega is selected according to B cell-site antenna of above-mentioned optimum ^*, carrying out iterations is N _iconventional iterative interference alignment, design base station pre-coding matrix and user's AF panel matrix of each community:

4.1a) the pre-coding matrix of random initializtion base station k and meet wherein k=1 ..., B, () ^hthe conjugate transpose of representing matrix, I _nrepresent the unit matrix of n × n dimension, d represents that base station sends the dimension of data flow;

4.1b) initialization iterations indicator variable m is 1;

4.1c) according to scholars such as K.Gomadam at article " Approaching the Capacity of WirelessNetworks through Distributed Interference Alignment " (Proceedings of the IEEE GlobalTelecommunications Conference, Miami, USA, training method 2008:1-5), determines the interference covariance matrix of the user k ' when the m time iteration

$Q_{k^{\′},m}^{{\mathrm{\Ω}}^{*}}=\underset{k=1,k\≠k^{\′}}{\overset{B}{\mathrm{\Σ}}}\frac{P_k}{d}{H}_{k^{\′}k}^{{\mathrm{\ω}}_k^{*}}{V}_{k,m-1}^{{\mathrm{\Ω}}^{*}}{\left(V_{k,m-1}^{{\mathrm{\Ω}}^{*}}\right)}^H{\left(H_{k^{\′}k}^{{\mathrm{\ω}}_k^{*}}\right)}^H,$

Wherein, k '=1 ..., B, P _krepresent the transmitting power of base station k, represent the pre-coding matrix of the base station k when the m-1 time iteration, represent at antenna selecting plan in situation, base station k is to the channel matrix of user k ';

4.1d) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the AF panel matrix of design user k ' when the m time iteration wherein k '=1 ..., B, υ _dd the characteristic value characteristic of correspondence vector that { } representing matrix is minimum;

4.1e) according to scholars such as K.Gomadam at article " Approaching the Capacity of WirelessNetworks through Distributed Interference Alignment " (Proceedings of the IEEE GlobalTelecommunications Conference, Miami, USA, training method 2008:1-5), determines the interference covariance matrix of the base station k when the m time iteration

${\stackrel{\‾}{Q}}_{k,m}^{{\mathrm{\Ω}}^{*}}=\underset{k^{\′}=1,k^{\′}\≠k}{\overset{B}{\mathrm{\Σ}}}\frac{P_{k^{\′}}}{d}{\left(H_{k^{\′}k}^{{\mathrm{\ω}}_k^{*}}\right)}^H{U}_{k^{\′},m}^{{\mathrm{\Ω}}^{*}}{\left(U_{k^{\′},m}^{{\mathrm{\Ω}}^{*}}\right)}^H{H}_{k^{\′},k}^{{\mathrm{\ω}}_k^{*}},$

Wherein, k=1 ..., B, P _kthe transmitting power of ' expression user k ', its value is equal with the transmitting power of this cell base station;

4.1f) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the pre-coding matrix of design base station k when the m time iteration wherein k=1 ..., B;

4.1g) value of iterations variable m adds 1, if m≤N _i, then return step 4.1c, proceed iteration; Otherwise iterative process terminates, obtain the pre-coding matrix of base station k with the AF panel matrix of user k ' wherein k=1 ..., B, k '=1 ..., B, jumps to step 4.2;

4.2) according to pre-coding matrix and the AF panel matrix of above-mentioned conventional iterative interference alignment design, each cell base station and user transmit and receive data respectively, carry out multiple cell interference coordination, eliminate presence of intercell interference.

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

1) computation complexity analysis

Before analysis computation complexity of the present invention, first provide the computation complexity of poor search base station antenna selecting method.Poor search base station antenna selecting method is the most direct method of one, namely travel through the various skies line options situation selecting Vector Groups to synthesize by all alternative antenna of all base stations, then selected the antenna for base station selection scheme making system and the maximum a kind of each community of capacity by conventional iterative interference alignment.Owing to there being B base station in system, and each base station has individual alternative antenna selects vector, and therefore the method is searched for the number of times of various skies line options situation and is in each search procedure, carrying out iterations is N _iconventional iterative interference alignment, disturb the computation complexity of alignment to be in each search procedure:

${\mathrm{BN}}_I[O\left(N_f^3\right)+O\left(M^3\right)]+{\mathrm{BN}}_I\left[2(B-1)(O\left(N_f^{2}M\right)+O\left(N_f{M}^2\right))\right].$

Therefore, the computation complexity that poor search base station antenna selecting method is total is:

${\mathrm{BN}}_I{\left(C_N^{N_f}\right)}^B[O\left(N_f^3\right)+O\left(M^3\right)+2(B-1(O\left(N_f^{2}M\right)+O\left(N_f{M}^2\right))].$

The optimal anchor direction that the present invention increases along system and capacity, when keeping other antenna for base station selection schemes constant, be not optimized the antenna selecting plan of each base station successively, not limit institute likely situation, therefore the present invention searches for the number of times of various skies line options situation and is in each search procedure, the present invention adopts part iteration to disturb alignment, reduces the iterations of interference alignment, disturbs the computation complexity of alignment to be in each search procedure:

In step 2 matrix F-norm computation complexity with disturb compared with the computation complexity that aligns very little in each search procedure, therefore this part complexity can be ignored.Consider that step 4 carries out the computation complexity of conventional iterative interference alignment, the total computation complexity of the present invention can be expressed as:

$(B^2{N}_{\mathrm{ite}}{C}_N^{N_f}+{\mathrm{BN}}_I)[O\left(N_f^3\right)+O\left(M^3\right)+2(B-1)(O\left(N_f^{2}M\right)+O\left(N_f{M}^2\right))],$

Further abbreviation is:

${\mathrm{BN}}_I(1+\mathrm{\θ}{\mathrm{BC}}_N^{N_f})[O\left(N_f^3\right)+O\left(M^3\right)+2(B-1)(O\left(N_f^{2}M\right)+O\left(N_f{M}^2\right))],$

Wherein, θ=N _ite/ N _ifor part iteration factor, its span is 0 < θ≤1.Due at B > 1 and condition under be less than especially at B or time larger much smaller than therefore the present invention is lower relative to poor search base station antenna selecting method computation complexity.

2) emulation experiment

2.1) simulated conditions:

Adopt MATLAB simulation software, simulate 1000 times.The antenna number N=4 of community number B=4, base station, the rf chain way N of base station _f=3, the iterations of the antenna of user and rf chain way M=3, the dimension d=1 sending data flow, conventional iterative interference alignment is N _i=30.Supposing that the transmitting power of each cell base station is equal, is all P, and the variances sigma of Gaussian noise ²=1, then power P represents the signal to noise ratio snr of each community user in system.

2.2) content and result is emulated:

At the iterations N of part iteration interference alignment _ite=1,5,10, under 30 conditions, with signal to noise ratio snr situation of change, Monte-Carlo Simulation is carried out to the system of the present invention and random antenna for base station system of selection and poor search base station antenna selecting method and capacity, obtains corresponding simulation comparison figure, as shown in Figure 5, wherein abscissa represents signal to noise ratio snr, and ordinate represents system and capacity.

As can be seen from Figure 5, system and the volumetric properties of random antenna for base station system of selection are the poorest, because random device is random selecting antenna for base station and access respective radio-frequency link, are difficult to obtain antenna selection gain.Poor search base station antenna selecting method has traveled through the various skies line options situation selecting Vector Groups to synthesize by all alternative antenna of all base stations, and adopts the method for conventional iterative interference alignment, therefore obtains optimum system and capacity.When signal to noise ratio snr is increased to 30dB from 0dB, system of the present invention and capacity all a little less than poor search base station antenna selecting method apparently higher than the system of random antenna for base station system of selection and capacity.The present invention is at N _iteconcrete system and volumetric properties situation when getting different value: work as N _itewhen getting minimum value 1, poor-performing of the present invention, the optimal performance obtained apart from poor search base station antenna selecting method is comparatively far away, because N _itetoo little, the interference of part iteration is alignd, and once, but performance now is still better than the performance of random antenna for base station system of selection to only iteration; Work as N _itewhen getting maximum 30, the present invention obtains the optimal performance that can reach, and this optimal performance is close to the performance of poor search base station antenna selecting method.Work as N _itewhen getting 5 and 10 respectively, performance of the present invention is relative to N _iteperformance when getting 1 is obviously promoted.Particularly N _itewhen getting 10, the present invention can obtain almost with N _iteperformance same when getting 30, and iterations is relative to N _iteless when getting 30, show N _ite=10 compromises that can obtain system and capacity and computation complexity.

The analysis of simulation result and computation complexity shows, the present invention can reduce computation complexity effectively, obtain simultaneously close to poor search base station antenna selecting method and volumetric properties.In addition, the present invention, by the iterations of adjustment member iteration interference alignment, realizes the compromise of system and capacity and computation complexity.

Claims (3)

1. in multi-cell system based on interference alignment an antenna for base station system of selection, comprise the steps:

(1) channel condition information of each base station to this community user is estimated:

Base station k sends pilot frequency information to this community user k, and user k estimates the channel condition information of k each antenna in base station to self, and feeds back to base station k, wherein k=1 ..., B, B are community number; The channel condition information that base station k feeds back according to this community user, determines channel matrix wherein N and M represents the antenna number of base station k and user k respectively, represent complex field;

(2) each cell-site antenna selection scheme of initialization:

The antenna selecting plan ω of initialization base station k _kfor having the matrix of maximum F-norm corresponding alternative antenna selects vector, ω _kcomputing formula as follows:

${\mathrm{\&omega;}}_k=\arg \underset{l\&Element;\{1,...,C_N^{N_f}\}}{\max }{\left|\right|H_{\mathrm{kk}}^{\mathrm{\&phi;}}\left|\right|}_F^2,$

Wherein, represent when antenna selecting plan is φ _ltime according to channel matrix H _kkthe base station k determined to the channel matrix of this community user,

In formula, H _kkfor base station k is to the channel matrix of this community user, φ _lrepresent that l alternative antenna of base station selects vector, it is the column vector that N grows, and in this vector, element value is whether 1 or 0 not represent respective antenna selected, for alternative antenna selects vector index variable, N and N _frepresent antenna number and the rf chain way of base station respectively, and N>N _f, represent and choose N from N number of number _fthe number of combinations of number, Diag{ φ _lrepresent that the element on a leading diagonal is column vector φ _lthe diagonal matrix of corresponding element, function represent the submatrix be made up of the non-zero row of matrix, ‖ ‖ _fthe F-norm of representing matrix; The antenna selecting plan of each base station is combined into B cell-site antenna and selects set omega={ ω ₁, ω ₂..., ω _b;

(3) along the optimal anchor direction that system and capacity increase, when keeping other antenna for base station selection schemes constant, part iteration is utilized to disturb alignment to determine the antenna for base station selection scheme of each community optimum successively:

3.1) initialization alternative antenna selects the index variables l of vector to be 1, the system that initialization is simultaneously maximum and capacity C _maxbe 0;

3.2) by the antenna for base station selection scheme ω of community 1 ₁be updated to alternative antenna and select vectorial φ _l, keep the antenna selecting plan of other base stations constant, selected by B cell-site antenna the 1st element in set omega to be updated to φ simultaneously _l;

3.3) according to above-mentioned antenna selecting plan, carry out part iteration interference alignment, design base station pre-coding matrix and user's AF panel matrix of each community, be implemented as follows:

3.3a) the pre-coding matrix of random initializtion base station k this pre-coding matrix satisfy condition: wherein, Ω is that B cell-site antenna selects set, N _frepresent the rf chain way of base station, d represents that base station sends the dimension of data flow, () ^hthe conjugate transpose of representing matrix, I _nrepresent the unit matrix of n × n dimension, represent complex field;

3.3b) initialization iterations indicator variable c is 1;

3.3c) determine the interference covariance matrix of the user k ' when the c time iteration

$Q_{k\text{'},c}^{\mathrm{\&Omega;}}=\underset{k=1,k\&NotEqual;k\text{'}}{\overset{B}{\mathrm{\&Sigma;}}}\frac{P_k}{d}{H}_{k\text{'}k}^{{\mathrm{\&omega;}}_k}{V}_{k,c-1}^{\mathrm{\&Omega;}}{\left(V_{k,c-1}^{\mathrm{\&Omega;}}\right)}^H{\left(H_{k\text{'}k}^{{\mathrm{\&omega;}}_k}\right)}^H,$

Wherein, k '=1 ..., B, B are community number, P _krepresent the transmitting power of base station k, d represents that base station sends the dimension of data flow, represent the pre-coding matrix of the base station k when the c-1 time iteration, represent at antenna selecting plan ω _kin situation, base station k is to the channel matrix of user k ', () ^hthe conjugate transpose of representing matrix;

3.3d) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the AF panel matrix of design user k ' when the c time iteration wherein, υ _dd the characteristic value characteristic of correspondence vector that { } representing matrix is minimum;

3.3e) according to above-mentioned AF panel matrix determine the interference covariance matrix of the base station k when the c time iteration

${\stackrel{\&OverBar;}{Q}}_{k,c}^{\mathrm{\&Omega;}}=\underset{k\text{'}=1,k\text{'}\&NotEqual;k}{\overset{B}{\mathrm{\&Sigma;}}}\frac{P_{k\text{'}}}{d}{{\left(H_{k\text{'}k}^{{\mathrm{\&omega;}}_k}\right)}^{H}V}_{k\text{'},c}^{\mathrm{\&Omega;}}{\left(V_{k\text{'},c}^{\mathrm{\&Omega;}}\right)}^H{H}_{k\text{'},k}^{{\mathrm{\&omega;}}_k},$

Wherein, P _{k '}represent the transmitting power of user k ', value is equal with the transmitting power of this cell base station, and d represents that base station sends the dimension of data flow, represent the AF panel matrix of the user k ' when the c time iteration, represent at antenna selecting plan ω _kin situation, base station k is to the channel matrix of user k ', () ^hthe conjugate transpose of representing matrix;

3.3f) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the pre-coding matrix of design base station k when the c time iteration $V_{k,c}^{\mathrm{\&Omega;}}={\mathrm{\&upsi;}}_d\left\{{\stackrel{\&OverBar;}{Q}}_{k,c}^{\mathrm{\&Omega;}}\right\};$

3.3g) value of iterations indicator variable c adds 1, if c≤N _ite, then return step 3.3c, proceed iteration; Otherwise, jump to step 3.4, obtain N _itethe pre-coding matrix of base station k after secondary iteration with the AF panel matrix of user k ' n _itefor the iterations of part iteration interference alignment;

3.4) equivalent channel of base station k ' to user k ' is determined and the interference covariance matrix of user k ' wherein:

$Q_{k\text{'}}^{\mathrm{\&Omega;}}=\underset{k=1,k\&NotEqual;k\text{'}}{\overset{B}{\mathrm{\&Sigma;}}}\frac{P_k}{d}{H}_{k\text{'}k}^{{\mathrm{\&omega;}}_k}{V}_{k,N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}{\left(V_{k,N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}\right)}^H{\left(H_{k\text{'}k}^{{\mathrm{\&omega;}}_k}\right)}^H,$

In formula, P _krepresent the transmitting power of base station k, d represents that base station sends the dimension of data flow, represent at N _itethe pre-coding matrix of base station k after secondary iteration, represent at antenna selecting plan ω _kin situation, base station k is to the channel matrix of user k ', () ^hthe conjugate transpose of representing matrix;

3.5) according to the equivalent channel that part iteration interference alignment obtains aF panel matrix and interference covariance matrix computing system and capacity wherein for the capacity of user k ', computing formula as follows:

$C_{k\text{'}}^{\mathrm{\&Omega;}}={\log }_2|I_d+\frac{P_{k\text{'}}{\left(U_{k\text{'},N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}\right)}^H{H}_{k\text{'}k\text{'}}^{{\mathrm{\&omega;}}_{k\text{'}}}{V}_{k\text{'},N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}{\left(V_{k\text{'},N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}\right)}^H{\left(H_{k\text{'}k\text{'}}^{{\mathrm{\&omega;}}_{k\text{'}}}\right)}^H{U}_{k\text{'},N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}/d}{{\left(U_{k\text{'},N_{\mathrm{ite}}}^{\mathrm{\&Omega;}}\right)}^H({\mathrm{\&sigma;}}^2{I}_M+Q_{k\text{'}}^{\mathrm{\&Omega;}})U_{k\text{'}{,N}_{\mathrm{ite}}}^{\mathrm{\&Omega;}}}|,$

Wherein, P _{k '}represent the transmitting power of base station k ', d represents that base station sends the dimension of data flow, with represent N respectively _itethe pre-coding matrix of base station k ' and the AF panel matrix of user k ' after secondary iteration, represent at antenna selecting plan ω _{k '}in situation, base station k ' is to the channel matrix of user k ', represent the interference covariance matrix of user k ', () ^hthe conjugate transpose of representing matrix, σ ²the variance of white Gaussian noise in channel, I _nrepresent the unit matrix of n × n dimension, M is the antenna number of user, || the determinant of representing matrix, bodge is bit/s/Hz;

3.6) if C ^Ω>C _max, then C is set _max=C ^Ω, and the alternative antenna of order optimum selects the index variables l of vector ^*=l, namely optimum alternative antenna selects vectorial φ _l*=φ _l; Otherwise, C _maxremain unchanged;

3.7) alternative antenna selects the value of the index variables l of vector to add 1, if then return step 3.2, continue other alternative antenna of traversal and select vector; Otherwise, skip to step 3.8;

3.8) the antenna for base station selection scheme of community 1 optimum is determined

3.9) according to the method described in step 3.1 to 3.8, the antenna for base station selection scheme of community 2 to community B optimum is determined successively ${\mathrm{\&omega;}}_2^{*},{\mathrm{\&omega;}}_3^{*},...,{\mathrm{\&omega;}}_B^{*};$

3.10) the antenna for base station selection scheme of above-mentioned each community optimum is combined into B optimum cell-site antenna and selects set ${{\mathrm{\&Omega;}}^{*}=\{\mathrm{\&omega;}}_2^{*},{\mathrm{\&omega;}}_3^{*},...,{\mathrm{\&omega;}}_B^{*}\};$

(4) conventional iterative interference alignment is adopted to carry out multiple cell interference coordination:

4.1) set omega is selected according to B cell-site antenna of above-mentioned optimum ^*, carrying out iterations is N _iconventional iterative interference alignment, design base station pre-coding matrix and user's AF panel matrix of each community:

4.1a) the pre-coding matrix of random initializtion base station k and meet wherein, N _frepresent the rf chain way of base station, d represents that base station sends the dimension of data flow, () ^hthe conjugate transpose of representing matrix, I _nrepresent the unit matrix of n × n dimension, represent complex field;

4.1b) initialization iterations variable m is 1;

4.1c) determine the interference covariance matrix of the user k ' when the m time iteration

$Q_{k\text{'},m}^{{\mathrm{\&Omega;}}^{*}}=\underset{k=1,k\&NotEqual;k\text{'}}{\overset{B}{\mathrm{\&Sigma;}}}\frac{P_k}{d}{H}_{k\text{'}k}^{{\mathrm{\&omega;}}_k^{*}}{V}_{k,m-1}^{{\mathrm{\&Omega;}}^{*}}{\left(V_{k,m-1}^{{\mathrm{\&Omega;}}^{*}}\right)}^H{\left(H_{k\text{'}k}^{{\mathrm{\&omega;}}_k^{*}}\right)}^H,$

Wherein, P _krepresent the transmitting power of base station k, represent the pre-coding matrix of the base station k when the m-1 time iteration, represent at antenna selecting plan in situation, base station k is to the channel matrix of user k ';

4.1d) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the AF panel matrix of design user k ' when the m time iteration wherein υ _dd the characteristic value characteristic of correspondence vector that { } representing matrix is minimum;

4.1e) according to above-mentioned AF panel matrix determine the interference covariance matrix of the base station k when the m time iteration

${\stackrel{\&OverBar;}{Q}}_{k,m}^{{\mathrm{\&Omega;}}^{*}}=\underset{k\text{'}=1,k\text{'}\&NotEqual;k}{\overset{B}{\mathrm{\&Sigma;}}}\frac{P_{k\text{'}}}{d}{{\left(H_{k\text{'}k}^{{\mathrm{\&omega;}}_k^{*}}\right)}^{H}V}_{k\text{'},m}^{{\mathrm{\&Omega;}}^{*}}{\left(V_{k\text{'},m}^{{\mathrm{\&Omega;}}^{*}}\right)}^H{H}_{k\text{'},k}^{{\mathrm{\&omega;}}_k^{*}},$

Wherein, P _{k '}represent the transmitting power of user k ', value is equal with the transmitting power of this cell base station;

4.1f) reveal minimum principle, according to above-mentioned interference covariance matrix according to interference the pre-coding matrix of design base station k when the m time iteration $V_{k,m}^{{\mathrm{\&Omega;}}^{*}}={\mathrm{\&upsi;}}_d\left\{{\stackrel{\&OverBar;}{Q}}_{k,m}^{{\mathrm{\&Omega;}}^{*}}\right\};$

4.1g) value of iterations variable m adds 1, if m≤N _i, wherein N _irepresent and be preferably the iterations empirical value that tradition interference alignment convergence needs 30, then return step 4.1c, proceed iteration; Otherwise iterative process terminates, obtain the pre-coding matrix of base station k with the AF panel matrix of user k '

4.2) according to pre-coding matrix and the AF panel matrix of above-mentioned conventional iterative interference alignment design, each cell base station and user transmit and receive data respectively, carry out multiple cell interference coordination, eliminate presence of intercell interference.

2. in multi-cell system according to claim 1 based on interference alignment antenna for base station system of selection, the base station k wherein described in step (1) to user k send pilot frequency information, be carry out in the following manner: base station k divides the secondary user k of giving sends pilot frequency information, chooses some antennas each time and they are connected to radio frequency link to send pilot frequency information, wherein from all N number of antennas, chooses N the 1st time _findividual antenna, the 2nd time from remaining N-N _fn is chosen in individual antenna _findividual antenna, the 3rd time from remaining N-2N _fn is chosen in individual antenna _findividual antenna, until the secondaryly choose last remaining antenna, thus guarantee that all antennas all can once be connected to radio frequency link at certain and send pilot frequency information, wherein k=1 ..., B, B are community number, N and N _frepresent antenna number and the rf chain way of base station, and N>N _f, representative rounds up to scalar x.

3. in multi-cell system according to claim 1 based on interference alignment antenna for base station system of selection, wherein in step (3.3g) part iteration interference alignment iterations N _itedemand fulfillment N _ite≤ N _i, wherein N _irepresent and be preferably 30 by the iterations empirical value that tradition interference alignment convergence needs.