CN109728891B - Cluster-based User Pairing Method in MU-MIMO System - Google Patents

Abstract

The invention discloses a clustering-based user pairing method in a MU-MIMO system, which mainly solves the problems of low spectrum utilization and high solution complexity in the prior art. The technical solution is: 1. Use the split method to generate a complete set of resource block splitting; 2. Determine the number and center of classes in each resource block split set; 3. Initialize user groups and resources in each class Block group; 4. Continuously perform user removal operations and user join operations in each class until all users are classified into classes, and the user pairing process is completed. On the basis of considering the system bit error rate constraint, the invention adopts the minimum mean square error-sort-based continuous interference cancellation MMSE-OSIC technology at the receiving end, which can not only efficiently perform dynamic multi-user pairing, but also meet the system communication requirements. In the case of quality requirements, maximize the spectrum utilization of the communication system. Can be used to pair mobile phone users in MU‑MIMO systems.

Description

Cluster-based User Pairing Method in MU-MIMO System

technical field

The invention belongs to the field of communication technology, and in particular relates to a clustering-based user pairing method, which can be used in a multi-user-multiple-input multiple-output-single-carrier-frequency division multiple access MU-MIMO-SC-FDMA system.

Background technique

In today's 4G mobile communication system and the upcoming 5G mobile communication system, the multi-user-multiple-input multiple-output MU-MIMO technology occupies a very important position. MU-MIMO technology combines multiple users with only one antenna into one user pairing at the transmitting end through reasonable user pairing criteria, and the user pairing is regarded as a physical terminal with multiple antennas. The multiple antennas installed on the base station side of the terminal and the receiving end together form a MU-MIMO array, thereby significantly improving the spectral efficiency of the system. In the MU-MIMO system, the most critical research problem is to choose an appropriate user pairing method. To measure the pros and cons of a user pairing method, it mainly depends on the spectral utilization ratio of the system.

At present, the classical user pairing algorithms include: random pairing criterion, orthogonal user pairing algorithm, determinant pairing criterion, and orthogonal defect degree pairing criterion. Although these basic user pairing algorithms can ensure that the interference between users is as small as possible to a certain extent, they do not significantly improve the spectrum utilization of the system.

Xiaofeng Lu et al. published the paper "Dynamic User Grouping and JointResource Allocation With Multi-Cell Cooperation for Uplink Virtual MIMOSystems" (IEEE Transactions on Wireless Communications, vol.16, no.6, pp.3854-3869, Jun 2017. ), an iterative Hungarian method is proposed for the user pairing problem in MU-MIMO systems. The specific steps of the method are: firstly grouping users to generate a complete set of user groups; then grouping resource blocks to generate a complete set of resource block groups; finally using the iterative Hungarian algorithm to obtain the optimal combination of user grouping and resource block grouping the combination of situations and system throughput. The disadvantage of this method is that the complexity is high and the spectral efficiency achieved is also low.

SUMMARY OF THE INVENTION

The purpose of the present invention is to propose a clustering-based user pairing method in a MU-MIMO system to reduce the complexity and improve the spectrum utilization rate of the system in view of the above-mentioned shortcomings of the prior art.

To achieve the above purpose, the technical solution includes the following:

(1) Generate a complete set of resource block splitting:

Use the splitting method to split all resource blocks in the system to generate a complete set of resource block splitting, the complete set of resource block splitting includes multiple resource block splitting sets, and each resource block splitting set contains multiple resources block group;

(2) Generate the number of clusters and initialize the resource block groups and users in each class:

The number of resource block groups in each resource block split set is defined as the number of classes K, and the resource block groups in the K classes are defined as T ₁ ,...,T _m ,...,T _K , and each resource block group is defined as T 1 ,...,T m ,...,T K The resource block groups in the block splitting set are placed in T ₁ ,…,T _m ,…,TK in turn, and the user groups in the _K classes are defined as Ω ₁ ,…,Ω _m ,…,Ω _K in turn, and the total user group is defined as Ω 1 ,…,Ω m ,…,Ω K . The number is L, and K users are sequentially selected from L users and put into Ω ₁ ,…,Ω _m ,…,Ω _K ;

(3) Calculate the center Z ₁ ,...,Z _m ,...,Z _K of each class according to the minimum mean square error-sort-based continuous interference cancellation MMSE-OSIC detection method and the bit error rate-constrained adaptive modulation method;

(4) Execute the user removal operation:

(4a) set the number of users to be removed from a class to be n each time;

(4b) select a class from the unselected classes, judge whether the number of users in this class is greater than n, if so, execute (4c), otherwise, execute (4e);

计算该类中每个用户到该聚类中心的距离为

然后将到中心距离最小的用户移出该类，并根据(3)更新每个类的中心Z₁,…,Z_m,…,Z_K；(4c) Define the transmission efficiency of the resource block group T _m occupied by user u in the user group Ω _m in the mth class as

Calculate the distance of each user in this class to the cluster center as

Then move the user with the smallest distance to the center out of the class, and update the center Z ₁ ,…,Z _m ,…,Z _K of each class according to (3);

(4d) Repeat (4c) until n users are removed from the class;

(4e) Repeat (4b)-(4d) until all K classes are selected;

(5) Execute user add operation:

(5a) Set the number of users added to the class as λ each time;

并将该用户加到与其距离最小的聚类中心对应的类中；(5b) Select a user from the users to be added, and calculate the distance from the user to all cluster centers as

And add the user to the class corresponding to the cluster center with the smallest distance;

(5c) Repeat (5b) until λ users are added to the class;

(5d) update the centers Z ₁ ,...,Z _m ,...,Z _K of each class according to (3);

(6) Repeat (4) and (5) until all L users are assigned to the class, completing the user pairing process.

Compared with the prior art, the present invention has the following advantages:

First, because the present invention adopts the method of dynamically performing multi-cell user pairing and resource allocation according to the channel state under the condition of a given system bit error rate threshold, it overcomes the inability of the prior art to guarantee the system communication quality during the resource allocation process. Therefore, the present invention can ensure that the bit error rate of the system is below the threshold value while maximizing the frequency utilization rate of the system, thereby improving the communication quality of the system;

Second, since the present invention adopts a user clustering algorithm based on the minimum mean square error-sorted serial interference elimination MMSE-OSIC technology, the large problem of joint user pairing and resource allocation is decomposed into several small-scale sub-problems It can be solved in parallel, which greatly reduces the search space for solving the problem, overcomes the problems that the existing technology cannot make full use of the frequency spectrum and has high solving complexity, so that the present invention improves the spectral efficiency while reducing the complexity.

Description of drawings

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

Fig. 2 is the simulation result of the complexity of the method of the present invention and the existing method;

FIG. 3 is the simulation result of the spectral efficiency of the method of the present invention and the existing method.

Detailed ways

The present invention will be described in further detail below with reference to the accompanying drawings.

1, the implementation steps of the present invention are as follows:

Step 1: Generate a complete set of resource block splitting.

Set the number of resource blocks in the multi-user-multiple-input multiple-output-single-carrier-frequency division multiple access MU-MIMO-SC-FDMA system to n, and use the splitting method to split the resource blocks to generate resource block splitting Complete set, the specific steps are as follows:

(1a) Represent all consecutive resource blocks in the MU-MIMO-SC-FDMA system as a sequence of {B ₁ ,B ₂ ,...,B _i ,B _i+1 ,...,B _n }, where B _i represents the first i resource blocks, i is from 1 to n, n represents the total number of resource blocks;

(1b) Use n-1 horizontal lines to separate the resource block sequence to obtain the separated resource block sequence {B ₁ ,_,B ₂ ,_,…,B _i ,_,B _i+1 ,…,_ ,B _n };

(1c) Randomly insert a number 0 or a number 1 on the n-1 horizontal lines, that is, convert the decimal numbers 1 to ^2n-1 into a binary sequence of length n-1, and insert ^2n-1 kinds of binary sequences in turn n-1 horizontal lines, 2 ^n-1 resource block split sets are obtained;

(1d) In each resource block split set, determine whether the number inserted on the horizontal line between B _i and B _i+1 is 0 or 1: if the number is 0, use B _i and B _i+1 to form a resource block group;

If the number is 1, use B _i and B _i+1 to form two different resource block groups;

(1e) 2 ^n-1 resource block split sets are formed into a set, and a complete set of resource block split is generated.

Step 2, generate the number of clusters and initialize the resource block group and user group in each class.

(2a) Select a resource block split set from the resource block split complete set, and define the number of resource block groups in the resource block split set as the number K of classes;

(2b) Defining resource block groups in K classes as T ₁ ,...,T _m ,...,T _K , m is from 1 to K, and put the resource block groups in the resource block splitting set into T ₁ ,... ,T _m ,…,T _K in;

(2c) Define the user groups in the K classes as Ω ₁ ,…,Ω _m ,…,Ω _K in turn, define the total number of users as L, and select K users from the L users into Ω ₁ ,… ,Ω _m ,…,Ω _K.

Step 3, calculate the center of each class:

According to the existing minimum mean square error-ranking-based continuous interference cancellation MMSE-OSIC detection method and bit error rate constrained adaptive modulation method, the center Z ₁ ,…,Z _m ,…,Z _K of each class is calculated, specifically Proceed as follows:

(3a) Calculate the signal-to-interference-noise ratio of user u occupying the e-th subcarrier in the m-th user group Ω _m as:

是用户组Ω_m中用户u在第e个子载波上的检测矢量，

是用户组Ω_m和第e个子载波之间信道矩阵的第u列对应的矢量，σ²是高斯噪声方差；where e is from 1 to |T _m |, |T _m | is the number of subcarriers in resource block group T _m , _u is from 1 to |Ω _m |, |Ω _m | number,

is the detection vector of user u on the e-th subcarrier in the user group Ω _m ,

is the vector corresponding to the u-th column of the channel matrix between the user group Ω _m and the e-th subcarrier, and σ ² is the Gaussian noise variance;

(3b) Calculate the transmission efficiency of user u occupying the e-th subcarrier in the m-th user group Ω _m as:

是向下取舍操作，BER^tar是信号传输的BER约束的上限；in,

is the round-down operation, and BER ^tar is the upper limit of the BER constraint for signal transmission;

(3c) Calculate the transmission efficiency of the resource block group T _m occupied by user u in the mth user group Ω _m as:

(3d) Calculate the transmission efficiency of the resource block group T _m occupied by the mth user group Ω _m as:

(3e) Calculate the center of the mth class as:

Z _m =R _m .

Step 4, perform the user move out operation:

(4a) set the number of users to be removed from a class to be n each time;

(4b) select a class from the unselected classes, judge whether the number of users in this class is greater than n, if so, execute (4c), otherwise, execute (4e);

计算该类中每个用户到该聚类中心的距离为

然后将到中心距离最小的用户移出该类，并根据(3)更新每个类的中心Z₁,…,Z_m,…,Z_K；(4c) Define the transmission efficiency of the resource block group T _m occupied by user u in the user group Ω _m in the mth class as

Calculate the distance of each user in this class to the cluster center as

Then move the user with the smallest distance to the center out of the class, and update the center Z ₁ ,…,Z _m ,…,Z _K of each class according to (3);

(4d) Repeat (4c) until n users are removed from the class;

(4e) Repeat (4b)-(4d) until all K classes are selected.

Step 5, perform user add operation:

(5a) Set the number of users added to the class as λ each time;

具体步骤如下：(5b) Select a user from the users to be added, and calculate the distance from the user to all cluster centers as

Specific steps are as follows:

(5b1) Select a user from the users to be added and add it to the mth class to obtain the user group as

中用户d占用第e个子载波的传输效率为：(5b2) Calculate the mth user group

The transmission efficiency of user d occupying the e-th subcarrier is:

是向下取舍操作，d是从1到

是用户组

中用户的数目，SINR(d)_e,m+是用户组

中用户d占用第e个子载波的信干噪比，BER^tar是信号传输的BER约束的上限；in,

is the round-down operation, and d is from 1 to

is a user group

The number of users in SINR(d) _e,m+ is the user group

where user d occupies the signal-to-interference and noise ratio of the e-th subcarrier, and BER ^tar is the upper limit of the BER constraint for signal transmission;

中用户d占用资源块组T_m的传输效率为：(5b3) Calculate the mth user group

The transmission efficiency of the resource block group T _m occupied by user d is:

是用户组

中用户d占用第e个子载波的传输效率；in,

is a user group

The transmission efficiency of user d occupying the e-th subcarrier;

占用资源块组T_m的传输效率为：(5b4) Calculate the mth user group

The transmission efficiency of the occupied resource block group T _m is:

(5b5) Calculate the distance from the selected user to the mth cluster center as:

where Z _m is the center of the mth class.

(5c) adding the user to the class corresponding to the cluster center with the smallest distance from it;

(5d) Repeat (5b)-(5c) until λ users are added to the class;

(5e) Update the centers Z ₁ ,…,Z _m ,…,Z _K of each class according to (3).

Step 6, complete the user pairing process:

Repeat (4) and (5) until all L users are assigned to the class, completing the user pairing process.

The effect of the present invention can be further explained by simulation

1. Simulation conditions:

The simulation of the present invention is carried out in the wireless communication scenario of a single base station, the number of resource blocks is 6, the threshold value of the system bit error rate is 10 ^-5 , and the detection mode of the signal receiver is set as the minimum average value in the simulation experiment of the present invention. Square Error-Sequential Interference Cancellation MMSE-OSIC detection based on ordering and assumes that the channel matrix is invariant within a single slot. The performance of the existing user pairing technology and the method of the present invention in terms of complexity and system spectral efficiency are compared.

2. Simulation content and result analysis

Simulation 1: According to the above simulation conditions, the complexity of the method of the present invention and the existing method is simulated, and the result is shown in FIG. 2 . It can be seen from FIG. 2 that when the number of users is 10, 20, 30, and 40, the method of the present invention has lower computational complexity.

Simulation 2: According to the above simulation conditions, the spectrum efficiency of the method of the present invention and the existing method is simulated, and the result is shown in FIG. 3 .

It can be seen from FIG. 3 that the spectral efficiency of the method of the present invention and the existing method are both rising with the continuous increase of the signal-to-noise ratio, but the slope of the spectral efficiency curve of the method of the present invention is significantly larger than the slope of the curve of the existing method, and Regardless of the level of signal-to-noise ratio, the performance curve of the method of the present invention is always above the existing method. Therefore, the proposed user pairing method can fully utilize the spectrum resources in the system and improve the spectrum efficiency.

Claims (3)

1. A user pairing method based on clustering in an MU-MIMO system is characterized by comprising the following steps:

(1) generating a complete split set of resource blocks:

splitting all resource blocks in the system by using a splitting method to generate a complete splitting set of the resource blocks, wherein the method is realized as follows:

(1a) all consecutive resource blocks in a MU-MIMO-SC-FDMA system are denoted as { B }₁,B₂,···,B_i,B_i+1,···,B_nSequence of which B is_iRepresents the ith resource block, and n represents the total number of resource blocks;

(1b) the resource block sequence is separated by n-1 transverse lines to obtain a separated resource block sequence { B₁,_,B₂,_,···,B_i,_,B_i+1,···,_,B_n}；

(1c) Randomly inserting digit 0 or digit 1 on n-1 horizontal lines, i.e. decimal numbers 1-2^n-1Are all converted into binary sequences of length n-1, 2^n-1The binary sequence is inserted into n-1 horizontal lines in sequence to obtain 2^n-1Splitting and collecting the seed resource blocks;

(1d) in each resource block split set, judge B_iAnd B_i+1The number inserted on the horizontal line between them is 0 or 1, and if it is 0, B_iAnd B_i+1Form a resource block group, if the number is 1, B_iAnd B_i+1Forming two different resource block groups;

(1e) will 2^n-1The seed resource block split sets form a set, and a complete resource block split set is generated;

the resource block complete splitting set comprises a plurality of resource block splitting sets, and each resource block splitting set comprises a plurality of resource block groups;

(2) generating cluster number and initializing resource block group and user in each cluster:

defining the number of resource block groups in each resource block splitting set as the number K of classes, and defining the resource block groups in the K classes as T in turn₁,···,T_m,···,T_KSequentially putting each resource block group in the T₁,···,T_m,···,T_KIn (3), sequentially defining the user groups in K classes as omega₁,···,Ω_m,···,Ω_KDefining the total number of users as L, selecting K users from the L users in sequence and putting the K users into omega₁,···,Ω_m,···,Ω_KPerforming the following steps;

(3) calculating the center Z of each class according to the minimum mean square error-ordering-based continuous interference cancellation MMSE-OSIC detection method and the bit error rate constraint adaptive modulation method₁,···,Z_m,···,Z_K；

(4) And executing user removal operation:

(4a) setting the number of users moved out of a class each time as eta;

(4b) selecting a class from the unselected classes, judging whether the number of users in the class is greater than eta, if so, executing (4c), otherwise, executing (4 e);

(4c) the user group omega in the mth class_mResource block group T occupied by user u_mIs defined as a transmission efficiency of

Calculating the distance from each user in the cluster to the cluster center as

Then, the user with the minimum distance to the center is moved out of the class, and the center Z of each class is updated according to (3)₁,···,Z_m,···,Z_K；

(4d) Repeating (4c) until η users are moved out of the class;

(4e) repeating (4b) - (4d) until all K classes are selected;

(5) and executing user adding operation:

(5a) setting the number of users added to the class each time as lambda;

(5b) selecting one user from the users to be added, and calculating the distance from the user to all the cluster centers to obtain

Adding the user to the class corresponding to the clustering center with the minimum distance;

(5c) repeating (5b) until λ users are all added to the class;

(5d) updating the center Z of each class according to (3)₁,···,Z_m,···,Z_K；

(6) And (5) repeating the steps (4) and (5) until all L users are allocated to the class, and finishing the user pairing process.

2. The method of claim 1, wherein the step (3) calculates the center of each class according to the minimum mean square error-ordering-based successive interference cancellation (MMSE) -OSIC detection method and the bit error rate constrained adaptive modulation method, and the method is implemented as follows:

(3a) computing user group Ω_mThe signal-to-interference-and-noise ratio of the sub-carrier e occupied by the user u is as follows:

wherein,

is a user group omega_mThe detected vector of user u on the e-th subcarrier,

is a user group omega_mAnd the vector corresponding to the u-th column of the channel matrix between the e-th subcarriers, | Ω_m| is the user group Ω_mNumber of users, σ²Is the gaussian noise variance;

(3b) computing user group Ω_mThe transmission efficiency of the user u occupying the e-th subcarrier is as follows:

wherein,

is a downward cut operation, BER^tarIs the upper bound of the BER constraint for signal transmission;

(3c) computing user group Ω_mResource block group T occupied by user u_mThe transmission efficiency of (a) is:

(3d) computing user group Ω_mOccupied resource block group T_mThe transmission efficiency of (a) is:

(3e) the center of the mth class is calculated as:

Z_m＝R_m。

3. the method of claim 1, wherein in step (5b), one user is selected from the users to be added, and the distances from the user to all cluster centers are calculated as follows:

(5b1) selecting one user from the users to be added, and adding the user into the mth category to obtain a user group of

(5b2) Computing user groups

The transmission efficiency of the user u occupying the e-th subcarrier is as follows:

wherein,

is a downward trade-off operation, SINR (u)_e,m+Is a group of users

Signal to interference plus noise ratio (BER) of the e-th subcarrier occupied by user u^tarIs the upper bound of the BER constraint for signal transmission;

(5b3) computing user groups

Resource block group T occupied by user u_mThe transmission efficiency of (a) is:

wherein,

is a group of users

The transmission efficiency of the e sub-carrier occupied by the user u;

(5b4) computing user groups

Occupied resource block group T_mThe transmission efficiency of (a) is:

(5b5) calculating the distance from the selected user to the mth clustering center as follows:

wherein Z is_mIs the center of the mth class.

Images