Clustering method under asynchronous extensible cellular-free large-scale MIMO partial coherent transmission

Abstract

The invention provides a clustering method under asynchronous extensible non-cellular large-scale MIMO partial coherent transmission, belongs to the technical field of communication, solves the problem of performance loss of a non-cellular large-scale MIMO system caused by asynchronous receiving, and specifically comprises the following steps: performing channel estimation on the non-cellular large-scale MIMO system under asynchronous receiving, and performing channel estimation on the non-cellular large-scale MIMO system under asynchronous receiving; obtaining linear minimum mean square error channel estimation of the non-cellular large-scale MIMO system in an uplink training stage under asynchronous receiving; based on linear minimum mean square error channel estimation, obtaining a spectrum efficiency expression of asynchronous cellular-free large-scale MIMO downlink partial coherent transmission in a local minimum mean square error precoding mode and a maximum ratio precoding mode, and completing spectrum efficiency performance analysis of downlink partial coherent transmission; an extensible clustering algorithm based on a large-scale fading coefficient and a distance is designed, APs providing services for each user are grouped into a series of coherent clusters, and downlink partial coherent transmission is carried out in combination with a partial minimum mean square error precoding mode and a maximum ratio precoding mode.

Description

Technical Field

[0001] The invention relates to a clustering method under asynchronous scalable non-cellular large-scale MIMO partially coherent transmission, and belongs to the technical field of communication. Background Art

[0002] Global wireless communication technology has experienced several rapid development stages, with a new generation of versions being launched approximately every ten years. Whether it is the early analog signal transmission and digital communication, or the later widespread application of 3G, 4G and 5G networks, each generation of wireless technology has been striving for higher speed, higher reliability and lower latency. However, with the explosive growth in the number of users and the emergence of various new applications such as artificial intelligence, the traditional cellular network architecture faces a series of challenges such as difficult handovers, insufficient signal coverage and low energy efficiency. A promising solution to the above problems is to use a cell-free massive MIMO architecture. In a cell-free massive MIMO system, multiple access points (APs) are distributed over a wide geographical area to coherently provide services to multiple users. Through a user-centric approach, the cell-free massive MIMO system can provide users with seamless and uniform coverage, reduce user handover delay and handover failure probability. In addition, the cell-free massive MIMO system has the characteristics of macro diversity gain and low path loss, and has higher spectral efficiency and energy efficiency than the cellular massive MIMO system.

[0003] Although the non-cellular massive MIMO system has the above advantages, it faces the problem of asynchronous signal reception in actual deployment due to its distributed architecture. The main reason for asynchronous reception is that the geographically distributed APs will inevitably lead to signal arrival time differences with users, such as Figure 1 As shown in Figure 2, asynchronous reception caused by transmission delay differences prevents APs from obtaining accurate channel state information and causes severe multi-user interference, which reduces the achievable spectrum efficiency of users and becomes the main obstacle to the practical deployment of non-cellular massive MIMO systems.

[0004] With coherent transmission, all APs in a cell-free massive MIMO system can transmit the same data stream to each user, achieving high beam gain and thus high data rates. However, due to the existence of asynchronous reception, ensuring coherent transmission in a cell-free massive MIMO system is very challenging. Therefore, a solution is needed to operate a cell-free massive MIMO system in the presence of asynchronous reception to unleash its advantages.

[0005] The emergence of partially coherent transmission becomes a feasible solution, allowing APs in cell-free massive MIMO systems to perform coherent and incoherent transmissions simultaneously. Specifically, when using partially coherent transmission, APs need to be grouped into multiple coherent clusters. Within each coherent cluster, the phases between APs are the same, sending the same data stream to all users. APs from different coherent clusters send different data streams to users in an incoherent manner, such as Figure 2 Partially coherent transmission can alleviate the problems caused by asynchronous reception, provide higher data rates, and make cell-free massive MIMO communication systems more practical and attractive.

[0006] Early cell-free massive MIMO assumed that all APs provide services for all users, which would result in huge fronthaul signaling and unacceptable computational complexity. To achieve high performance while ensuring manageable fronthaul signaling and computational complexity, user-centric cell-free massive MIMO was proposed, where each user is served by a specific AP cluster (i.e., a subset of APs with optimal channel conditions). Correspondingly, each AP only serves some of the users. The benefit of the user-centric approach is that only the APs in the cluster must receive downlink data for the user and send the corresponding uplink data estimates to the CPU, reducing fronthaul signaling. In addition, each AP only needs to process signals related to a subset of users, reducing computational complexity.

[0007] However, when the number of users is large, the scalability of the non-cellular massive MIMO system must be considered. Scalability means that system resources (such as signal processing, fronthaul signaling, and total power) are independent of the number of users, that is, more users can be added to the system without increasing existing system resources. Applying a scalable clustering scheme will simplify the network management and resource allocation of the non-cellular massive MIMO system, making its deployment more flexible and providing a foundation for it to effectively cope with the growing communication needs and complex application scenarios in the future. Summary of the invention

[0008] The present invention solves the problem that asynchronous reception causes performance loss to a non-cellular massive MIMO system, and further proposes a clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission, which specifically includes:

[0009] Step 1: Perform channel estimation on the non-cellular massive MIMO system under asynchronous reception to obtain a linear minimum mean square error channel estimation in the uplink training phase of the non-cellular massive MIMO system under asynchronous reception;

[0010] Step 2: Based on the linear minimum mean square error channel estimation, the spectrum efficiency expression of the downlink partially coherent transmission under the local minimum mean square error precoding and maximum ratio precoding is obtained, and the spectrum efficiency performance analysis of the downlink partially coherent transmission is completed;

[0011] Step 3: Design a scalable clustering algorithm based on large-scale fading coefficients and distances, group the APs that provide services to each user into a series of coherent clusters, and perform partial coherent downlink transmission in combination with partial minimum mean square error precoding and maximum ratio precoding.

[0012] Preferably, step 1 specifically includes:

[0013] Step 1.1: Define the quantized time offset between user k and AP q as in, is the floor function, T s is the sampling period of OFDM symbol, t qk (d qk ) is the propagation time / distance from user k to APq, The user closest to AP q Propagation time / distance to AP q;

[0014] Step 1.2: Calculate the quantized phase shift between AP q and user k on the nth subcarrier based on the quantized time offset between user k and AP q. Where N is the number of subcarriers in the OFDM symbol;

[0015] Step 1.3: In the uplink training phase of the non-cellular massive MIMO system, the user sends a set of length t to the serving AP. p The pilot sequence The linear minimum mean square error channel estimate is obtained, where the linear minimum mean square error channel estimate includes the channel vector h qk and its covariance matrix R qk The linear minimum mean square error estimate of ;

[0016] Channel vector h qk and its covariance matrix R qk The linear minimum mean square error estimate of is expressed as:

[0017]

[0018] In formula (1), P k is the pilot power transmitted by user k, is the pilot signal received by AP q, The expression is:

[0019]

[0020] In formula (2), is the noise variance.

[0021] Preferably, step 2 specifically includes:

[0022] Step 2.1: Set the coherent cluster l in the partially coherent downlink transmission of the non-cellular massive MIMO system k The set of APs serving user k;

[0023] Step 2.2: Take the diagonal matrix Describing coherent clusters k The communication connection between AP q and user k is obtained, and the expression of spectrum efficiency under asynchronous cell-free massive MIMO downlink partially coherent transmission is obtained;

[0024] Step 2.3: Using maximum ratio precoding, obtain a closed expression for the spectrum efficiency of downlink partially coherent transmission under maximum ratio precoding;

[0025] Coherent Cluster The expression of the set of APs serving user k in is:

[0026]

[0027] In formula (3), c q,k ∈{0,1} is used to determine the coherent cluster l k Whether APq in communicates with user k, is a coherent cluster l k The AP set in

[0028] Coherent Cluster k The expression of the communication connection between AP q and user k in is:

[0029]

[0030] The spectrum efficiency expression of asynchronous cell-free massive MIMO downlink partially coherent transmission is:

[0031]

[0032] In formula (5), t is the total number of symbols in each coherent block, N OFDM is the total length of OFDM symbol, is the quantized phase shift on the nth subcarrier between APq and user k in the downlink, w qk is the precoding vector;

[0033] The closed expression of the spectrum efficiency of downlink partially coherent transmission under maximum ratio precoding is:

[0034]

[0035] Preferably, in step 3, the step of designing a scalable clustering algorithm based on large-scale fading coefficient and distance, and grouping APs providing services to each user into a series of coherent clusters comprises:

[0036] Step 3.1: Connect users to a subset of corresponding APs based on large-scale fading coefficients, so that users and APs establish optimal connections;

[0037] Step 3.2: The connected APs are grouped according to the distance, and the APs with the same quantized phase shift are divided into a group of coherent clusters.

[0038] Preferably, step 3.1 specifically includes:

[0039] When new user k * When entering the network, measure the large-scale fading coefficients of all APs in the AP set And connect to M0￡Q APs with the largest large-scale fading coefficient, and assign at least one AP serving each user, where Q is the total number of APs;

[0040] The expression of AP set is:

[0041]

[0042] In formula (7), For collection The descending set of r Define a new set consisting of the first r elements of the set L.

[0043] Preferably, allocating at least one AP serving each user specifically includes:

[0044] Set each AP to only serve users, where the collection For the set of users served by AP q, define k ** for The index of the user with the smallest medium-to-large scale fading coefficient, if Then the collection Delete user k ** ,like Then the collection Delete AP q, for i∈{k * ,k **}have Indicates that user i has only one serving AP, then user i maintains the connection with AP q and discards the set Users in .

[0045] Preferably, step 3.2 specifically includes:

[0046] Step 3.2.1: Assemble APs The APs in the

[0047] Step 3.2.2: Gather at AP Find the distance between user k * The nearest AP and the farthest AP, where the distance to user k * The index of the nearest AP is Distance user k * The index of the farthest AP is

[0048] Step 3.2.3: Set up user k * The sampling distance D is based on the index And the sampling distance D to obtain user k * The number of coherent clusters;

[0049] Step 3.2.4: According to user k * The number of coherent clusters and sampling distance D, set a set of reference distances For collections AP q in , if it satisfies the condition Then AP q is grouped into coherent clusters In Output clustering results And according to the clustering results Constructing the Matrix

[0050] User k * The expression for the number of coherent clusters is:

[0051]

[0052] In formula (8), D = T s ×c is the sampling distance.

[0053] The beneficial effects of the present invention are:

[0054] 1. The spectral efficiency achieved by the partially coherent transmission method proposed in the present invention is always higher than the spectral efficiency under asynchronous reception, but lower than the spectral efficiency under synchronous reception. Compared with asynchronous reception, partially coherent transmission improves the user's spectral efficiency by performing coherent transmission within a coherent cluster, and greatly alleviates the performance loss caused by asynchronous reception to the non-cellular massive MIMO system.

[0055] 2. The scalable clustering algorithm proposed in the present invention has a higher total rate than the non-scalable clustering algorithm, and the clustering algorithm proposed in the present invention is scalable, ensuring manageable fronthaul signaling and computational complexity while achieving high performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A flow chart of a clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission provided by the present invention;

[0057] Figure 2 A schematic diagram of a comparison of user downlink SE performance provided by the present invention;

[0058] Figure 3 This is a schematic diagram comparing the clustering algorithm proposed in the present invention with the non-scalable clustering algorithm. DETAILED DESCRIPTION

[0059] Specific implementation method 1: Combination Figure 1 To illustrate this embodiment, Figure 1 As shown, the clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission described in this embodiment includes the following steps:

[0060] S1: Derivation of SE expression for downlink partially coherent transmission in asynchronous cell-free massive MIMO system;

[0061] S101: performing channel estimation for a non-cellular massive MIMO system under asynchronous reception;

[0062] Define the quantized time offset between user k and AP q as in, is the floor function, T s is the sampling period of OFDM symbol, t qk (d qk ) is the propagation time / distance from user k to APq, The user closest to AP q Propagation time / distance to AP q;

[0063] S102: Calculate the quantized phase shift between AP q and user k on the nth subcarrier based on the quantized time offset between user k and AP q Where N is the number of subcarriers in the OFDM symbol;

[0064] S103: In the uplink training phase, all users send a set of length t p The pilot sequence Calculate the channel vector h qk and its covariance matrix Rqk The linear minimum mean square error estimate of ;

[0065] Channel vector h qk and its covariance matrix R qk The linear minimum mean square error estimate of is expressed as:

[0066]

[0067] In formula (1), P k is the pilot power transmitted by user k, is the pilot signal received by AP q, The expression is:

[0068]

[0069] In formula (2), is the noise variance:

[0070] S104: In the downlink partially coherent transmission, this embodiment uses dynamic cooperative clusters to reduce the computational complexity and meet the scalability requirements of the system. k The set of APs serving user k is:

[0071]

[0072] In formula (3), c q,k ∈{0,1} is used to determine the coherent cluster l k Whether APq in communicates with user k, is a coherent cluster l k The AP set in

[0073] S105: For convenience of representation, this embodiment uses a set of diagonal matrices To describe the coherent cluster l k The communication connection between AP q and user k in is as follows:

[0074]

[0075] S106: For asynchronous non-cellular massive MIMO, the downlink achievable SE expression of user k under partially coherent transmission is:

[0076]

[0077] In formula (5), t is the total number of symbols in each coherent block, N OFDM is the total length of OFDM symbol,

[0078] is the quantized phase shift on the nth subcarrier between APq and user k in the downlink, w qk is the precoding vector;

[0079] S107: From the achievable rate expression in formula (5), it can be seen that the useful signal in the numerator contains the sum of the squares of the contributions of different coherent clusters, which is different from the synchronous reception case (the sum is within the square) and the asynchronous case (the sum contains different quantized phase shifts). Using partially coherent transmission, APs with the same quantized phase shift are grouped into the same coherent cluster, thereby eliminating the impact of the quantized phase shift caused by asynchronous reception on the useful signal. In addition, the multi-user interference in the denominator of formula (5) also changes accordingly, but the impact of the quantized phase shift still exists. Therefore, partially coherent transmission can effectively reduce the impact of asynchronous reception on non-cellular large-scale MIMO systems. If maximum ratio precoding is used, a closed expression can be obtained, as shown in formula (6):

[0080]

[0081] S2: Design a scalable clustering algorithm based on large-scale fading coefficients and distances, group the APs that provide services to each user into a series of coherent clusters, and perform partial coherent transmission of the downlink by combining partial minimum mean square error precoding and maximum ratio precoding;

[0082] In order to enable the non-cellular massive MIMO system to perform partially coherent transmission under asynchronous reception while meeting the scalability of the system, this embodiment proposes a clustering scheme based on large-scale fading coefficients and distance criteria. The clustering scheme proposed in this embodiment is divided into the following two stages: In the first stage, users are connected to a subset of APs according to large-scale fading coefficients, with the aim of establishing the best connection between users and APs (in terms of large-scale fading). In the second stage, the connected APs are further grouped according to distance, where APs with the same quantized phase shift are divided into a group of coherent clusters to improve the performance of partially coherent transmission; the first stage can solve the scalability problem because the matching decision allows the AP to provide services to the best user in its vicinity; the second stage considers the corresponding coherent cluster of each user to improve the achievable rate of the user under asynchronous reception.

[0083] S201: Connecting the user to a subset of corresponding APs according to the large-scale fading coefficients, so that the user and the AP establish an optimal connection (in terms of large-scale fading);

[0084] S20101: When new user k * When entering the network, measure the large-scale fading coefficients of all APs in the AP set And connected to M0￡Q (M0 is usually close to the total number of APs Q) APs with the largest large-scale fading coefficient, where Q is the total number of APs;

[0085] The expression of AP set is:

[0086]

[0087] In formula (7), For collection The descending set of r Define a new set consisting of the first r elements of the set L.

[0088] S20102: To ensure the scalability of the system, it is necessary to consider the limitation of AP processing capacity and ensure that each user has at least one AP serving it. More precisely, this implementation assumes that each AP can only serve users, thereby effectively avoiding strong pilot pollution, where the set The set of users served by AP q, therefore, user k * Need to communicate with the AP q Users compete, and k is defined ** express The index of the user with the smallest medium- and large-scale fading coefficient, if Then the collection Delete user k ** Otherwise, the set Delete AP q. In addition, if for i∈{k * ,k **}have This means that user i has only one serving AP, so user i remains connected to AP q, and the set The user in (user k ** The ones with the smallest large-scale fading coefficient outside the range will be discarded.

[0089] S202: Grouping connected APs according to distances, and dividing APs with the same quantized phase shift into a group of coherent clusters;

[0090] The goal of the second phase is to collect The APs in the The clustering algorithm is shown in Table 1. Since the quantized phase shift caused by asynchronous reception is related to the quantized time offset (which is determined by the distance), the second stage of the proposed clustering scheme relies on the distance to select the AP. The specific steps are as follows:

[0091] S20201: User k * First in the collection Find the nearest and farthest APs, whose indexes are and Then, determine user k * The number of coherent clusters:

[0092]

[0093] In formula (8), D = T s ×c is the sampling distance;

[0094] S20202: According to user k * The number of coherent clusters and sampling distance D, set a set of reference distances For collections AP q in , if the condition is satisfied Then AP q is grouped into coherent clusters In Output clustering results And according to the clustering results Constructing the Matrix

[0095]

[0096]

[0097] Specific implementation method 2: Combination Figure 2 and Figure 3 This embodiment is described. To verify the user downlink SE performance of the clustering method under the asynchronous scalable non-cellular massive MIMO partially coherent transmission proposed in the present invention, this embodiment is simulated and designed. The simulation parameters are shown in Table 2. The simulation steps include:

[0098] This embodiment generates APs and users with random positions in a simulation area of ​​1km′1km, and tests the performance of a non-cellular massive MIMO system based on the proposed expression of the achievable rate of partially coherent transmission, and compares it with the SE performance obtained during synchronous reception and asynchronous reception. In addition, two scalable precoding methods are considered, namely local minimum mean square error (LP-MMSE) precoding and maximum ratio (MR) precoding, which are expressed as:

[0099]

[0100] In formula (9), C qi =R qi -B qi is the estimation error matrix, Represents the uplink noise variance. The power allocation is a scalable equal power allocation scheme.

[0101] Table 2

[0102]

[0103]

[0104] Figure 2 The cumulative distribution function (CDF) of the achievable rate for synchronous reception, partially coherent transmission, and asynchronous reception transmission scenarios using LP-MMSE and MR precoding is shown, as Figure 2 As shown in the figure, no matter whether LP-MMSE or MR precoding is used, the spectrum efficiency achieved by the proposed partially coherent transmission method is always higher than the spectrum efficiency under asynchronous reception, but lower than the spectrum efficiency under synchronous reception. Compared with asynchronous reception, partially coherent transmission improves the spectrum efficiency of users by coherent transmission within the coherent cluster, but due to the incoherent transmission between coherent clusters, this performance improvement is not enough to achieve the ideal synchronous reception performance. Although there is still a certain gap between the spectrum efficiency of partially coherent transmission and synchronous reception, it greatly alleviates the performance loss caused by asynchronous reception to non-cellular massive MIMO systems.

[0105] Figure 3 The performance of the proposed scalable clustering algorithm and the non-scalable clustering algorithm under partially coherent transmission are compared when using LP-MMSE and MR precoding. Figure 3 As shown, the proposed scalable clustering algorithm has a higher total rate than the non-scalable clustering algorithm, which shows the effectiveness of the algorithm proposed in this embodiment. In addition, the clustering algorithm proposed in this embodiment is scalable, ensuring manageable fronthaul signaling and computational complexity while achieving high performance.

[0106] The above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent replacement and improvement made to the above embodiments without departing from the content of the technical solution of the present invention, based on the technical essence of the present invention, within the spirit and principles of the present invention, still fall within the protection scope of the technical solution of the present invention.

Claims

1. A clustering method for asynchronous scalable non-cellular massive MIMO partially coherent transmission, characterized in that: The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission includes the following steps: Step 1: Perform channel estimation on the non-cellular massive MIMO system under asynchronous reception to obtain a linear minimum mean square error channel estimation in the uplink training phase of the non-cellular massive MIMO system under asynchronous reception; Step 2: Based on the linear minimum mean square error channel estimation, the spectrum efficiency expression of the downlink partially coherent transmission under the local minimum mean square error precoding and maximum ratio precoding is obtained, and the spectrum efficiency performance analysis of the downlink partially coherent transmission is completed; Step 3: Design a scalable clustering algorithm based on large-scale fading coefficients and distances, group the APs that provide services to each user into a series of coherent clusters, and perform partial coherent downlink transmission in combination with partial minimum mean square error precoding and maximum ratio precoding.

2. The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission according to claim 1, characterized in that: Step 1 specifically includes: Step 1.1: Define the quantized time offset between user k and AP q as in, is the floor function, T s is the sampling period of OFDM symbol, t qk (d qk ) is the propagation time / distance from user k to APq, The user closest to AP q Propagation time / distance to AP q; Step 1.2: Calculate the quantized phase shift between AP q and user k on the nth subcarrier based on the quantized time offset between user k and AP q. Where N is the number of subcarriers in the OFDM symbol; Step 1.3: In the uplink training phase of the non-cellular massive MIMO system, the user sends a set of length t to the serving AP. p The pilot sequence The linear minimum mean square error channel estimate is obtained, where the linear minimum mean square error channel estimate includes the channel vector h qk and its covariance matrix R qk The linear minimum mean square error estimate of ; Channel vector h qk and its covariance matrix R qk The linear minimum mean square error estimate of is expressed as: In formula (1), P k is the pilot power transmitted by user k, is the pilot signal received by AP q, The expression is: In formula (2), is the noise variance.

3. The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission according to claim 1, characterized in that: Step 2 specifically includes: Step 2.1: Set the coherent cluster l in the partially coherent downlink transmission of the non-cellular massive MIMO system k The set of APs serving user k; Step 2.2: Take the diagonal matrix Describing coherent clusters k The communication connection between APq and user k is obtained, and the expression of spectrum efficiency under asynchronous non-cellular massive MIMO downlink partially coherent transmission is obtained; Step 2.3: Using maximum ratio precoding, obtain a closed expression for the spectrum efficiency of downlink partially coherent transmission under maximum ratio precoding; Coherent Cluster k The expression of the set of APs serving user k in is: In formula (3), c q,k ∈{0,1} is used to determine the coherent cluster l k Whether APq in communicates with user k, is a coherent cluster l k The AP set in Coherent Cluster k The expression of the communication connection between AP q and user k in is: The spectrum efficiency expression of asynchronous cell-free massive MIMO downlink partially coherent transmission is: In formula (5), t is the total number of symbols in each coherent block, N OFDM is the total length of OFDM symbol, is the quantized phase shift on the nth subcarrier between APq and user k in the downlink, w qk is the precoding vector; The closed expression of the spectrum efficiency of downlink partially coherent transmission under maximum ratio precoding is:

4. The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission according to claim 1, characterized in that: In step 3, the steps of designing a scalable clustering algorithm based on large-scale fading coefficients and distances to group APs serving each user into a series of coherent clusters include: Step 3.1: Connect users to a subset of corresponding APs based on large-scale fading coefficients, so that users and APs establish optimal connections; Step 3.2: The connected APs are grouped according to the distance, and the APs with the same quantized phase shift are divided into a group of coherent clusters.

5. The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission according to claim 4, characterized in that: Step 3.1 specifically includes: When new user k * When entering the network, measure the large-scale fading coefficients of all APs in the AP set And connect to M0￡Q APs with the largest large-scale fading coefficient, and assign at least one AP serving each user, where Q is the total number of APs; The expression of AP set is: In formula (7), For collection A descending set of r Define a new set consisting of the first r elements of the set L.

6. The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission according to claim 5, characterized in that: Allocating at least one AP serving each user specifically includes: Set each AP to only serve users, where the collection For the set of users served by AP q, define k ** for The index of the user with the smallest medium-to-large scale fading coefficient, if Then the collection Delete user k ** ,like Then the collection Delete AP q, for i∈{k * ,k ** }have Indicates that user i has only one serving AP, then user i maintains the connection with AP q and discards the set Users in .

7. The clustering method under asynchronous scalable non-cellular massive MIMO partially coherent transmission according to claim 4, characterized in that: Step 3.2 specifically includes: Step 3.2.1: Assemble APs The APs in the Step 3.2.2: Gather at AP Find the distance between user k * The nearest AP and the farthest AP, where the distance to user k * The index of the nearest AP is Distance user k * The index of the farthest AP is Step 3.2.3: Set up user k * The sampling distance D is based on the index And the sampling distance D to obtain user k * The number of coherent clusters; Step 3.2.4: According to user k * The number of coherent clusters and sampling distance D, set a set of reference distances For collections AP q in , if it satisfies the condition Then AP q is grouped into coherent clusters In Output clustering results And according to the clustering results Constructing the Matrix User k * The expression for the number of coherent clusters is: In formula (8), D = T s ×c is the sampling distance.

Images