User clustering algorithm based on cluster head selection and channel characteristics

Abstract

The invention provides a user clustering algorithm based on cluster head selection and channel characteristics, which is suitable for a non-orthogonal multiple access (Non-Orthogonal Multiple Access, NOMA) system of a large-scale multiple input multiple output (MIMO) without a cellular, and is characterized in that the user clustering algorithm based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics based on the cluster head selection and the channel characteristics. The method comprises the following three main steps of: firstly, preferentially selecting a user with an optimal access point channel condition from users as a cluster head user; secondly, a clustering evaluation index is constructed based on the channel gain difference and the channel correlation, and the clustering evaluation index is used for dynamically quantifying the matching degree between users; and finally, according to the evaluation indexes, gradually distributing the remaining users to the cluster where the corresponding cluster head user is located, and forming a final clustering scheme. According to the method, good balance between clustering performance and complexity is achieved, compared with a traditional statistical theory clustering algorithm, the calculation complexity is reduced by 47.5%, meanwhile, clustering fairness, anti-interference capacity and spectrum efficiency are remarkably improved, and the method is suitable for dynamic clustering and resource optimization distribution of a high-density user access scene.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication technology, and its main application scenario is a non-cellular large-scale MIMO-NOMA system. Its main application is to solve the problem of balancing user clustering performance and computational complexity in a non-cellular large-scale MIMO-NOMA system. Specifically, it involves a user clustering algorithm based on cluster head selection and channel characteristics, which constructs clustering evaluation indicators through cluster head user selection, channel gain difference and channel correlation, optimizes user clustering schemes, significantly reduces clustering complexity, and improves clustering fairness and anti-interference capabilities. It is suitable for dynamic clustering and resource allocation scenarios for large-scale user access. Background Art

[0002] In the non-cellular massive multiple input multiple output (MIMO) non-orthogonal multiple access (NOMA) system, power domain NOMA superimposes the data signals of multiple users in the power domain by utilizing the differences in user channel conditions, so that multiple users can share the same time-frequency resources, thereby significantly improving the spectrum efficiency of the system. At the receiving end, the superimposed signals are gradually decoded through the successive interference cancellation (SIC) technology to effectively reduce the intra-cluster interference. However, the decoding performance of SIC is highly dependent on the differences in user channel conditions. If the user channel conditions after clustering are too similar, SIC decoding will face serious interference problems, resulting in a significant decrease in system performance. Therefore, reasonable user clustering is crucial to fully utilize the system gain of NOMA.

[0003] Existing user clustering methods have significant deficiencies in balancing performance and complexity. Random clustering usually only obtains suboptimal solutions because it does not fully utilize the differences in user channel conditions, and cannot fully utilize the spectrum utilization potential of the system; although exhaustive search can achieve optimal clustering, its computational complexity increases exponentially with the increase in the number of users and clusters, making it difficult to apply in large-scale user scenarios. In addition, traditional clustering methods mainly rely on large-scale fading coefficients or the distance between users as the basis for clustering, ignoring two important channel characteristics: effective channel gain difference and channel correlation. Effective channel gain difference is an important factor affecting intra-cluster interference, while channel correlation directly determines the strength of inter-cluster interference and the severity of pilot pollution. In non-cellular systems, users communicate with multiple access points (APs) at the same time, and the dual effects of intra-group interference and inter-group interference significantly amplify this problem. However, existing clustering methods have not yet been able to comprehensively consider channel gain difference and channel correlation, resulting in poor interference management effects and limiting further improvements in system performance.

[0004] At the same time, with the continuous expansion of the system user scale and the complexity of non-cellular communication scenarios, how to effectively reduce the complexity of clustering calculations and improve clustering performance has become an important challenge in practical applications. On the one hand, overly simple clustering strategies cannot meet the needs of interference suppression; on the other hand, clustering methods based on complex channel matrix operations or global traversal have too high computational overhead and are difficult to adapt to dynamic and high-density user access scenarios. Therefore, designing an efficient user clustering algorithm that can not only balance clustering performance and complexity in large-scale user scenarios, but also make full use of channel gain differences and channel correlations to accurately manage interference has become the key to solving current problems. Summary of the invention

[0005] The present invention is inspired by the user clustering algorithm based on statistical theory. By optimizing the user clustering process and improving the channel characteristic modeling method, an efficient user clustering algorithm combining cluster head selection and channel characteristics is proposed. The proposed algorithm preferentially selects the user with the best channel condition with the access point as the cluster head user, and constructs a clustering evaluation index by combining the channel gain difference and channel correlation to achieve a dynamic and efficient user clustering process. Compared with the traditional statistical theory clustering algorithm, the algorithm of the present invention avoids the computational complexity problem caused by global traversal, and is comparable to the statistical theory algorithm in terms of user clustering performance, and is significantly better than the general random clustering scheme.

[0006] (I) Basic ideas and operations of the innovative points proposed by the present invention

[0007] The following specifically introduces the idea of ​​the user clustering algorithm based on cluster head selection and channel characteristics proposed by the present invention. The present invention divides the design of the user clustering algorithm into three main steps.

[0008] Step 1: Using the large-scale fading coefficient, the user with the best connection to the AP is selected as the cluster head user.

[0009] Step 2: Use large-scale fading coefficients to characterize channel gain differences and channel correlations to construct a comprehensive clustering evaluation index to quantify the matching degree between other users and the cluster head user.

[0010] Step 3: Use the clustering evaluation index to sequentially assign the users to be assigned to the clusters where the cluster head users that match them are located. After the optimization is completed, fix the assigned cluster members and continue to update the clustering schemes of the remaining users until all users are assigned and the final clustering matrix is ​​generated.

[0011] 1) System Model

[0012] Assume that the large-scale loss coefficient matrix is ​​β = [β1, β2, ... ,β M ] means that each column β m represents the large-scale channel gain between M APs and users, that is, β m =[β 1m ,β 2m , ... ,β (KN)m ] T First, the user's combined channel gain is constructed by calculating the covariance matrix and decomposing the eigenvalues, and the channel gain difference is further quantified. The standardized covariance matrix S of the matrix β is (S ij ) M×M ,in

[0013]

[0014] Next, the eigenvalues ​​of the matrix S are λ=[λ1,λ2,...,λ M ], eigenvector q=(q ij ) M×M , and the normalized feature weights The combined channel gain of KN users is obtained as H = [H1, H2, ..., H K ]. Specifically expressed as Thus, any two users u i and u j The channel gain difference is

[0015] In order to further quantify the channel correlation, user u is defined based on the Pearson correlation coefficient theory. i and u j The channel correlation is:

[0016]

[0017] Among them, Cov(i,j) is the user u i and u j The covariance of channel correlation is defined as:

[0018]

[0019] The channel gain difference matrix is ​​obtained and channel correlation matrix After that, the two are further normalized into matrices and Define the user clustering evaluation index as When i≠j, The larger the user u i and u j The smaller the signal interference and pilot pollution, the clearer the user clustering is.

[0020] In order to realize large-scale user clustering based on channel characteristics, this paper proposes a clustering algorithm. The core idea of ​​the algorithm is to select the user with the best connection characteristics as the cluster head user and use the channel gain difference index Construct user clusters to reduce interference while ensuring efficient spectrum reuse. The complete user clustering algorithm based on cluster head selection and channel characteristics is summarized in the following algorithm flow.

[0021] (II) Advantages of the present invention

[0022] The user clustering algorithm based on cluster head selection and channel characteristics proposed in the present invention has the following advantages:

[0023] 1. The present invention reduces the global traversal operation in the user clustering process by cluster head selection, avoiding the exponential complexity growth required by the exhaustive search method. Compared with the traditional clustering algorithm based on statistical theory, the computational complexity of the present invention is reduced by 47.5%, significantly improving the actual application efficiency of the algorithm in large-scale user scenarios.

[0024] 2. The present invention combines channel gain difference and channel correlation to construct clustering evaluation indicators, which can not only effectively reduce intra-cluster interference, but also suppress inter-cluster interference and pilot pollution by reducing inter-cluster channel correlation. By dynamically optimizing the clustering scheme, clustering fairness and system performance are significantly better than the random clustering method.

[0025] 3. The present invention fully considers the user channel characteristics in cluster head selection and evaluation index construction, maximizes the difference in channel conditions of users within the cluster, optimizes the decoding performance of continuous interference elimination, and thus significantly improves the system's anti-interference ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention provides the following drawings for illustration:

[0027] Attached Figure 1 Flow chart of the user clustering algorithm based on cluster head selection and channel characteristics DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the objectives, technical solutions and advantages of the present invention, a complete description is given below in conjunction with specific examples and drawings.

[0029] Attached Figure 1 The flowchart of the user clustering algorithm based on cluster head selection and channel characteristics is as follows:

[0030] Step 1: Initialize the unclustered user set Clustering result set

[0031] Step 2: Each AP selects the user with the largest large-scale loss, counts the number of times the user is selected, and takes the top N users with the highest frequency as the cluster head candidate set Update the unclustered user set

[0032] Step 3: In the process, cluster head user u is selected from the Nth cluster head to the first cluster head. n .

[0033] Step 4: For cluster head user u n , calculate the evaluation index of the unclustered users

[0034] Step 5: According to Sort in descending order, select the first K-1 users, and n Composed of current cluster G n .

[0035] Step 6: G n Add G and update the unclustered user set

[0036] Step 7: Repeat steps 3-6 until

[0037] Step 8: Return: Clustering set G = {G1, G2, ... ,G N}.

Claims

1. A user clustering algorithm based on cluster head selection and channel characteristics in a non-cellular massive MIMO-NOMA system, characterized in that: The algorithm realizes the user clustering process through cluster head selection and channel characteristic optimization. The clustering process includes the following steps: the first step is to select the user with the best channel condition with the access point as the cluster head user based on the large-scale fading coefficient to reduce the computational complexity of global traversal; the second step is to construct a clustering evaluation index based on the channel gain difference and channel correlation to quantify the matching degree between the unassigned users and the cluster head users; the third step is to dynamically assign users to the clusters where the corresponding cluster head users are located until the clustering process of all users is completed. Through the above steps, the algorithm effectively reduces the computational complexity while ensuring the channel quality of users in the cluster.

2. The user clustering algorithm according to claim 1, characterized in that: The selection of cluster head users not only reduces the complexity of traversing all user pairs during the calculation process, but also ensures a high channel gain within each user cluster by giving priority to selecting users with the best channel conditions with the access point as cluster head users, thereby improving the communication quality within the cluster and optimizing system performance.

3. The user clustering algorithm according to claim 1, characterized in that: The clustering evaluation index includes channel gain difference and channel correlation, wherein the channel gain difference is used to measure the intra-cluster interference intensity, and the channel correlation is used to measure the inter-cluster interference degree, and the two are combined to form a clustering evaluation standard after normalization processing.

4. The user clustering algorithm according to claim 1, characterized in that: By dynamically optimizing the clustering scheme, the user with the highest matching degree is preferentially assigned to the cluster where the cluster head user matching it is located, and the clustering scheme is updated until all users are assigned, thus forming the final clustering result.

5. The user clustering algorithm according to claim 1, characterized in that: The algorithm is suitable for dynamic high-density user access scenarios. Through reasonable intra-cluster and inter-cluster interference management, it achieves a good balance between clustering performance and computational complexity, providing an efficient solution for resource optimization allocation in non-cellular massive MIMO-NOMA systems.

Images