CN108173585B - Multi-user hybrid linear nonlinear precoding method - Google Patents

Abstract

The invention relates to a multi-user mixed linear nonlinear precoding method, which solves the technical problems of large interference among user groups and low transmission efficiency in the groups, and comprises the following steps: step 1, each user terminal estimates a downlink channel matrix of the user terminal, calculates a correlation matrix, and feeds back the correlation matrix of the downlink channel to a base station for a long time; step 2, the base station defines standard correlation and standard orthogonality according to the received correlation matrix sent by each user side, the user sides with the correlation larger than the standard correlation are collected into a user group, and the user sides with the orthogonality larger than the standard orthogonality are not in the same user group; step 3, performing channel dimensionality reduction between different user groups through block diagonalized first-layer linear precoding; and 4, scheduling a plurality of users in the same user group to perform second-layer nonlinear precoding, so as to realize the technical scheme of multi-user downlink transmission, better solve the problem and be used in a large-scale MIMO system.

Description

Multi-user hybrid linear nonlinear precoding method

Technical Field

The invention relates to the technical field of communication, in particular to a multi-user hybrid linear nonlinear precoding method.

Background

The Multiple-Input Multiple-Output (MIMO) technology can make full use of spatial multipath transmission, provide diversity gain, multiplexing gain, array gain, and the like, greatly improve the reliability and effectiveness of the system, and become a key technology for future wireless communication. In recent years, the research of MIMO technology has been extended from a point-to-point single-user MIMO system to a point-to-multipoint multi-user MIMO system. In a multi-user MIMO system, because a plurality of user information are transmitted in the same time-frequency resource block, the frequency spectrum efficiency is improved, and the co-channel interference among users is inevitably introduced. Since the multi-user MIMO system is limited by the conditions of receiving end cost, power consumption, geographical location dispersion, etc., it is difficult to apply some complex signal detection techniques to the receiving end.

The prior art generally adopts precoding at a transmitting end to eliminate co-channel interference. The linear block diagonalization precoding algorithm enables a precoding matrix of each user to be positioned on a null space of channel matrixes of other users, so that signal interference among the users is eliminated, the multi-user MIMO system is further decomposed into a plurality of parallel single-user MIMO channels, and the number of the transmission users is limited. In addition, the multi-user transmission method adopting user grouping can further reduce the complexity of the system, but the current grouping method has the technical problems of large interference among groups and poor viscosity in the groups, so that the number of users simultaneously transmitted is far less than that of base station antennas, and the advantages of the MIMO system cannot be exerted. Therefore, it is necessary to provide a multi-user hybrid linear nonlinear precoding method with small inter-group interference and high intra-group communication efficiency.

Disclosure of Invention

The technical problems to be solved by the invention are that the interference among all user groups is large and the transmission efficiency in the group is low in the prior art. The method has the characteristics of high orthogonality of users among groups, small interference, high transmission efficiency after dimension reduction, high correlation of users in the groups and convenience in transmission.

In order to solve the technical problems, the technical scheme is as follows:

a multi-user mixed linear nonlinear precoding method is used for a large-scale MIMO system, a base station in the large-scale MIMO system is a large-scale linear antenna array, and the multi-user mixed linear nonlinear precoding method comprises the following steps:

step 1, each user terminal estimates a downlink channel matrix of the user terminal, calculates a correlation matrix, and feeds back the correlation matrix of the downlink channel to a base station for a long time;

step 2, the base station defines standard correlation and standard orthogonality according to the received correlation matrix sent by each user side, and carries out user grouping, wherein the user sides with the correlation larger than the standard correlation are gathered into a user group, and the user sides with the orthogonality larger than the standard orthogonality are not in the same user group;

step 3, performing channel dimensionality reduction between different user groups through block diagonalized first-layer linear precoding;

and 4, scheduling a plurality of users in the same user group to perform second-layer nonlinear precoding, so as to realize multi-user downlink transmission.

In the above scheme, for optimization, further, the downlink channel matrix of user k in step 1 is

The correlation matrix for user k is

Where user k has only one antenna, N_tThe number of antenna elements of the large-scale linear antenna array; h is_lFor the channel response between the first antenna of the base station and the antenna of user K, l is 1,2, K, N_tK is the number of users, K is more than or equal to 1 and less than or equal to K, (. DEG)^HIndicating that the matrix is conjugate transformed.

Further, step 2 further comprises:

step A, defining the number of user groups as G, optionally selecting G users from K users, and taking the G users to send related matrixes as an initial group center matrix V₁,V₂,...,V_G；

Step B, calculating a correlation matrix R of the user k_kGroup center matrix V from the g-th group_gThe distance between the users k and the g group of users is measured, and the correlation is as follows:

wherein R is_k＝U_kΛ_kU_kFor eigenvalue decomposition, | | | | | non-calculation_FIs F norm, G is more than or equal to 1 and less than or equal to G;

and C, repeating the step B, adding the user k into the user group with the maximum correlation, and adding all the users into the user group:

d, updating the group center matrix of each group according to the divided user group result; according to the average correlation matrix R of users in the g group_gUpdating the central matrix of the g-th group with the feature vectors and the feature values:

step E, executing step B until the grouping results of the two times are consistent;

wherein, K_gIs the number of users in the g-th group.

Further, the first layer linear dimension reduction precoding method includes:

step A, dividing the G user groups in step 3 into 2 subsets, including:

(i) SVD is carried out on the group center matrix of the g user group, and right singular vectors corresponding to non-zero singular values are taken to form a main singular vector set of the g user group

(ii) Calculate the arbitrary g₁Group and g₂Chordal distance between groups:

wherein g is₁,g₂1,2, G, and G₁≠g₂；

(iii) The sum of two groups of minimum chord distances is maximized by averagely dividing the G groups into two subsets₁And₂：

s.t.|₁|＝|₂|＝G/2,₁∩₂＝φ；

and step B, carrying out approximate block diagonalization precoding on the two subsets.

Further, the approximate block diagonalization precoding includes:

(i) definition of

Set of interference singular vectors of g-th group, pair xi_gPerforming SVD with left singular matrix of

Wherein

Is a singular vector set corresponding to a zero singular value;

(ii) definition of

To pair

Performing EVD decomposition with a feature matrix of

Wherein

Is that

The characteristic vector set corresponding to the non-zero characteristic value;

(iii) the approximate block diagonalized precoding matrix of the g-th group is

Further, the nonlinear precoding method in the second layer group is as follows:

step A, calculating the signal-to-interference ratio of a g group of users k:

selecting users according to the signal-to-interference ratio in the g-th group, arranging the users in the group according to the descending order of the signal-to-interference ratio, and selecting the front L_gThe individual users are used as service users of the g group;

wherein L is_g＝min(K_g,b_g)，K_gIs the number of users of the g-th group, b_gIs a dimension reduction matrix B_gThe number of columns;

step B, within the g-th group, for the top L selected in step 5a_gCarrying out THP precoding on each user;

step C, calculating the signals finally received by the users in the group g as:

wherein P is the total transmitting power of the base station, L is the number of scheduled users,

the group g is subjected to THP precoding to obtain a transmission vector, pi_gIs a weighting matrix obtained in THP precoding, n_gThe noise vector of the user is served for the g-th group.

Further, the THP precoding includes:

(i) and (3) calculating the equivalent channel matrix of the selected user k in the group g:

H_g,k＝H_g,kB_g

feeding back the equivalent channel matrix after dimension reduction to a base station end;

(ii) the equivalent channel matrix of the users in the g-th group is

To pair

Carrying out QR decomposition:

then there is a weighting matrix

Wherein r is₁₁,K,

Is an element on the main diagonal of the matrix R, the feedback matrix phi_g＝Π_gR_g(:,1:L_g)^HFeed forward matrix F_g＝Q_g；

(iii) Original data signal

The sending signals obtained after the modulus taking and the continuous interference elimination are as follows:

wherein,

(iv) the transmission vector obtained by THP precoding is:

the invention has the beneficial effects that:

the method has the advantages that the multi-user mixed linear nonlinear precoding is adopted, so that the orthogonality of users among groups and the correlation of users in the groups are improved while the users are ensured to access at any time, and the high-efficiency transmission is realized while the interference is reduced;

compared with the existing situation that the number of the antennas is more than the number of the users, the number of the users is consistent with the number of the antennas, and the cost is low.

Drawings

The invention is further illustrated with reference to the following figures and examples.

Fig. 1 is a schematic diagram of a hybrid linear nonlinear precoding method in embodiment 1.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Example 1

The embodiment provides a multi-user hybrid linear nonlinear precoding method, which is used in a large-scale MIMO system, where a base station in the large-scale MIMO system is a large-scale linear antenna array, as shown in fig. 1, the multi-user hybrid linear nonlinear precoding method includes:

step 1, each user terminal estimates a downlink channel matrix of the user terminal, calculates a correlation matrix, and feeds back the correlation matrix of the downlink channel to a base station for a long time;

the downlink channel matrix of user k is

The correlation matrix for user k is

Where user k has only one antenna, N_tThe number of antenna elements of the large-scale linear antenna array; h is_lFor the channel response between the first antenna of the base station and the antenna of user K, l is 1,2, K, N_tK is the number of users, K is more than or equal to 1 and less than or equal to K, (. DEG)^HRepresenting conjugate transformation of the matrix;

step 2, the base station defines standard correlation and standard orthogonality according to the received correlation matrix sent by each user side, the user sides with the correlation larger than the standard correlation are collected into a user group, and the user sides with the orthogonality larger than the standard orthogonality are not in the same user group;

step A, defining the number of user groups as G, optionally selecting G users from K users, and taking the G users to send related matrixes as an initial group center matrix V₁,V₂,...,V_G；

Step B, calculating a correlation matrix R of the user k^kGroup center matrix V from the g-th group^gThe distance between the users k and the g group of users is measured, and the correlation is as follows:

wherein R is^k＝U_kΛ_kU_kFor eigenvalue decomposition, | | | | | non-calculation_FIs F norm, G is more than or equal to 1 and less than or equal to G;

and C, repeating the step B, adding the user k into the user group with the maximum correlation, and adding all the users into the user group:

step D, according to the divided user groupsUpdating the group center matrix of each group as a result; according to the average correlation matrix R of users in the g group^gUpdating the central matrix of the g-th group with the feature vectors and the feature values:

step E, executing step B until the grouping results of the two times are consistent;

wherein, K^gIs the number of users in the g-th group;

step 3, performing signal dimension reduction between different user groups through block diagonalized first-layer linear precoding;

the first layer linear dimension reduction precoding method comprises the following steps:

step A, dividing the G user groups in step 3 into 2 subsets, including:

(i) SVD is carried out on the group center matrix of the g user group, and right singular vectors corresponding to non-zero singular values are taken to form a main singular vector set of the g user group

(ii) Calculate the arbitrary g¹Group and g²Chordal distance between groups:

wherein g is¹,g₂1,2, G, and G₁≠g₂；

(iii) The sum of two groups of minimum chord distances is maximized by averagely dividing the G groups into two subsets₁And₂：

s.t.|₁|＝|₂|＝G/2,₁∩₂＝φ；

and step B, carrying out approximate block diagonalization precoding on the two subsets, wherein the approximate block diagonalization precoding comprises the following steps:

(i) definition of

Set of interference singular vectors of g-th group, pair xi_gPerforming SVD with left singular matrix of

Wherein

Is a singular vector set corresponding to a zero singular value;

(ii) definition of

To pair

Performing EVD decomposition with a feature matrix of

Wherein

Is that

The characteristic vector set corresponding to the non-zero characteristic value;

(iii) the approximate block diagonalized precoding matrix of the g-th group is

Step 4, scheduling a plurality of users in the same user group to perform second layer nonlinear precoding, and realizing multi-user downlink transmission, wherein the nonlinear precoding method in the second layer group comprises the following steps:

step A, calculating the signal-to-interference ratio of a g group of users k:

selecting users according to the signal-to-interference ratio in the g-th group, arranging the users in the group according to the descending order of the signal-to-interference ratio, and selecting the front L_gThe individual users are used as service users of the g group;

wherein L is_g＝min(K_g,b_g)，K_gIs the number of users of the g-th group, b_gIs a dimension reduction matrix B_gThe number of columns;

step B, within the g-th group, for the top L selected in step 5a_gCarrying out THP precoding on each user; the THP precoding includes:

(i) and (3) calculating the equivalent channel matrix of the selected user k in the group g:

H_g,k＝H_g,kB_g

feeding back the equivalent channel matrix after dimension reduction to a base station end;

(ii) the equivalent channel matrix of the users in the g-th group is

To pair

Carrying out QR decomposition:

then there is a weighting matrix

Wherein r is₁₁,K,

Is an element on the main diagonal of the matrix R, the feedback matrix phi_g＝Π_gR_g(:,1:L_g)^HFeed forward matrix F_g＝Q_g；

(iii) Original data signal

The sending signals obtained after the modulus taking and the continuous interference elimination are as follows:

wherein,

(iv) the transmission vector obtained by THP precoding is:

step C, calculating the signals finally received by the users in the group g as:

wherein P is the total transmitting power of the base station, L is the number of scheduled users,

n_gthe noise vector of the user is served for the g-th group.

Although the illustrative embodiments of the present invention have been described above to enable those skilled in the art to understand the present invention, the present invention is not limited to the scope of the embodiments, and it is apparent to those skilled in the art that all the inventive concepts using the present invention are protected as long as they can be changed within the spirit and scope of the present invention as defined and defined by the appended claims.

Claims (3)

1. A multi-user mixed linear nonlinear precoding method is used for a large-scale MIMO system, and a base station in the large-scale MIMO system is a large-scale linear antenna array, and is characterized in that: the multi-user hybrid linear nonlinear precoding method comprises the following steps:

step 1, each user terminal estimates a downlink channel matrix of the user terminal, calculates a correlation matrix, and feeds back the correlation matrix of the downlink channel to a base station for a long time;

step 2, the base station defines standard correlation and standard orthogonality according to the received correlation matrix sent by each user side, and carries out user grouping, wherein the user sides with the correlation larger than the standard correlation are gathered into a user group, and the user sides with the orthogonality larger than the standard orthogonality are not in the same user group;

step 3, performing channel dimensionality reduction between different user groups through block diagonalized first-layer linear precoding;

step 4, scheduling a plurality of users in the same user group to perform second-layer nonlinear precoding to realize multi-user downlink transmission;

the downlink channel matrix of user k in step 1 is

The correlation matrix for user k is

Where user k has only one antenna, N_tThe number of antenna elements of the large-scale linear antenna array; h is_lIs the channel response between the first antenna of the base station and the antenna of user k, i 1,2_tK is the number of users, K is more than or equal to 1 and less than or equal to K, (. DEG)^HRepresenting conjugate transformation of the matrix;

in step 2, the user grouping method includes:

step A, defining the number of user groups as G, optionally selecting G users from K users, and selecting G usersSending the correlation matrix as an initial set of central matrices V₁,V₂,...,V_G；

Step B, calculating a correlation matrix R of the user k_kGroup center matrix V from the g-th group_gThe distance between the users k and the g group of users is measured, and the correlation is as follows:

wherein R is_k＝U_kΛ_kU_kFor eigenvalue decomposition, | | | | | non-calculation_FIs F norm, G is more than or equal to 1 and less than or equal to G;

and C, repeating the step B, adding the user k into the user group with the maximum correlation, and adding all the users into the user group:

d, updating the group center matrix of each group according to the divided user group result; according to the average correlation matrix R of users in the g group_gUpdating the central matrix of the g-th group with the feature vectors and the feature values:

step E, executing step B until the grouping results of the two times are consistent;

wherein, K_gIs the number of users in the g-th group;

in step 3, the first layer linear dimension reduction precoding method includes:

step A, dividing the G user groups in step 3 into 2 subsets, including:

(i) SVD is carried out on the group center matrix of the g user group, and right singular vectors corresponding to non-zero singular values are taken to form a main singular vector set of the g user group

(ii) Calculate the arbitrary g₁Group and g₂Chordal distance between groups:

wherein g is₁,g₂1,2, G, and G₁≠g₂；

(iii) The sum of two groups of minimum chord distances is maximized by averagely dividing the G groups into two subsets₁And₂：

step B, carrying out approximate block diagonalization precoding on the two subsets;

approximating block diagonalization precoding includes:

(i) definition of

Set of interference singular vectors of g-th group, pair xi_gSingular value decomposition is carried out, and a left singular matrix is

Wherein

Is a singular vector set corresponding to a zero singular value;

(ii) definition of

To pair

Performing eigenvalue decomposition with an eigenvalue matrix of

Wherein

Is that

The characteristic vector set corresponding to the non-zero characteristic value;

(iii) the approximate block diagonalized precoding matrix of the g-th group is

2. The multi-user hybrid linear nonlinear precoding method of claim 1, wherein:

in step 4, the nonlinear precoding method in the second layer group is as follows:

step A, calculating the signal-to-interference ratio of a g group of users k:

selecting users according to the signal-to-interference ratio in the g-th group, arranging the users in the group according to the descending order of the signal-to-interference ratio, and selecting the front L_gThe individual users are used as service users of the g group;

wherein L is_g＝min(K_g,b_g)，K_gIs the number of users of the g-th group, b_gIs a dimension reduction matrix B_gThe number of columns;

step B, within the g group, for the first L selected in step A_gCarrying out THP precoding on each user;

step C, calculating the signals finally received by the users in the group g as:

wherein P is the total transmitting power of the base station, L is the number of scheduled users,

n_ga noise vector for serving users of the g-th group;

Π_gis a weighting matrix;

and the transmission vectors obtained by the group g through THP precoding.

3. The multi-user hybrid linear nonlinear precoding method of claim 2, wherein: the THP precoding includes:

(i) and (3) calculating the equivalent channel matrix of the selected user k in the group g:

H_g,k＝H_g,kB_g

feeding back the equivalent channel matrix after dimension reduction to a base station end;

(ii) the equivalent channel matrix of the users in the g-th group is

To pair

Carrying out QR decomposition:

then there is a weighting matrix

Wherein,

is an element on the main diagonal of the matrix R, the feedback matrix phi_g＝Π_gR_g(:,1:L_g)^HFeed forward matrix F_g＝Q_g；

(iii) Original data signal

The sending signals obtained after the modulus taking and the continuous interference elimination are as follows:

wherein,

m is the modulation order of the original data signal,

and

respectively representing the extraction of real and imaginary parts of the complex numbers in brackets;

[x_g]_lthe vector is obtained after the signal of the first user of the g group is subjected to modulus extraction and continuous interference elimination;

(iv) the transmission vector obtained by THP precoding is:

Images