Reference signal multiple access (RSMA)-based cellular-removal large-scale multiple input multiple output (MIMO) wireless forward transmission network beam forming and rate segmentation method

Abstract

The invention discloses a honeycomb-removing large-scale MIMO wireless forward transmission network beam forming and rate segmentation method based on RSMA, and the method comprises the steps: constructing a CF-mMIMO wireless forward transmission network model based on RSMA, and the CF-mMIMO wireless forward transmission network model comprises an antenna CPU, M single antennas and K single antenna users; on the basis of a penalty dual decomposition method, a maximum minimum forward transmission rate optimization problem is decomposed into three sub-problems, all the sub-problems are solved alternately, and the minimum forward transmission rate in the maximum K users is deduced. According to the method, in the CF-mMIMO wireless forward transmission network based on RSMA, the minimum forward transmission rate can be maximized by designing the beam forming and rate segmentation ratio, and the fairness of users is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a de-cellularized massive MIMO wireless fronthaul network beamforming and rate splitting method based on RSMA. Background Art

[0002] 1. Existing Decellularized Massive MIMO Wireless Fronthaul Network: The Cell Free Massive Multiple Input Multiple Output (CF-mMIMO) system introduces a "user-centric" network concept, allowing users to be served by a large number of access points (APs) around them, solving the problem of serious inter-cellular interference in traditional cellular networks and becoming one of the key technologies in the sixth generation (6G) mobile communications. In CF-mMIMO, the central processing unit (CPU) is connected to all APs through the fronthaul network and sends data to the APs. As the number of APs increases, the deployment and operation costs of traditional wired fronthaul networks become huge. In terms of multiple access, common ones are Space Division Multiple Access (SDMA), Time Division Multiple Access (TDMA), and Non-Orthogonal Multiple Access (NOMA).

[0003] The existing decellularized massive MIMO wireless fronthaul network has the following shortcomings: In traditional decellularized massive MIMO fronthaul networks, many studies have not considered the deployment and operation and maintenance difficulties of wired fronthaul networks. In addition, in CF-mMIMO systems that consider wireless fronthaul, no research has focused on adapting to dynamically changing interference levels. RSMA is a multiple access scheme that is very flexible to interference and can perform well under both high and low interference levels. Therefore, it is of great significance to study CF-mMIMO wireless fronthaul networks based on RSMA.

[0004] 2. Joint beamforming and rate splitting algorithm for CF-mMIMO wireless fronthaul network based on RSMA: In the RSMA-based CF-mMIMO wireless fronthaul network, how to design beamforming and rate split ratio to maximize the minimum fronthaul rate is a key issue. Since the objective function and constraints of the optimization problem to be solved contain non-convex expressions such as Signal-to-Interference-plus-Noise Ratio (SINR), it is difficult to obtain the optimal solution for the above problem, and the complexity of the solution is high. The main beamforming algorithms at present are semi-definite relaxation (SDR) beamforming, which transforms the original non-convex optimization problem into a semi-definite programming (SDP) problem by introducing auxiliary variables, and then can be solved using traditional convex optimization solvers. In addition, there are studies on the joint optimization of beamforming and rate splitting. By deriving the expression of rate and weighted mean square error (WMSE), the algorithm replaces the complex rate expression in the original problem with WMSE, and then solves the weighted minimum mean square error (WMMSE) problem. The alternating optimization method is used for each decision variable in the problem to obtain a suboptimal solution.

[0005] The existing beamforming and rate splitting algorithms in CF-mMIMO wireless fronthaul networks have the following deficiencies: In order to solve the non-convex maximization minimum forward rate optimization problem, there are currently two main methods: the first is to obtain a convex problem by temporarily ignoring or transforming the non-convex part of the original problem, and then use the traditional convex optimization solver to solve it; the second is to derive an approximate problem that is easier to solve than the original problem, and obtain a suboptimal solution by solving the approximate problem. However, the first algorithm, such as the SDR method, is too complex and computationally expensive, and is not suitable for CF-mMIMO systems with large-scale antennas; the penalty method and augmented Lagrangian method often rely on a good problem structure when solving, and have great limitations for solving complex problems. The second algorithm, such as the WMMSE method, can often only obtain a suboptimal solution and cannot guarantee high performance. Therefore, the existing algorithms have a lot of room for improvement in complexity and performance, and it is necessary to design a joint optimization algorithm with low complexity and excellent performance. Summary of the invention

[0006] The main purpose of the present invention is to propose a de-cellular massive MIMO wireless fronthaul network beamforming and rate splitting method based on RSMA, aiming to maximize the minimum fronthaul rate and ensure user fairness by designing beamforming and rate splitting ratio.

[0007] To achieve the above object, the present invention provides a decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method based on RSMA, the method comprising the following steps: Step S10, constructing a CF-mMIMO wireless fronthaul network model based on RSMA, wherein the CF-mMIMO wireless fronthaul network model includes a Antenna CPU, M single antennas and K single antenna users, the forward transmission rate of user k is expressed as ,in, is the public portion rate divided by user k, is the maximum private message transmission rate of user k; Step S20, based on the penalty dual decomposition method, decompose the maximization minimum forward transmission rate optimization problem into three sub-problems, alternately solve each sub-problem, and derive the minimum forward transmission rate among the maximized K users.

[0008] A further technical solution of the present invention is that step S10 comprises: make Represents the index set of all APs, A collection of indexes representing all users, Represented as user The set of AP indexes of the service; Forward message for each user , RSMA divides it into two parts: the public part and private part ; The public part from which all user messages are split Combined into a super public message , the private part is independent into corresponding private messages; therefore, the data stream emitted by the CPU is , each data stream is mapped to Transmitted on the antenna, the CPU's transmission signal It can be expressed as: (1); in Indicates public message The beamforming vector of Indicates private message The beamforming vector of the user Service AP , the signal it receives from the CPU is: (2); in Indicates CPU and AP The channel vector between For noise, is the noise variance; AP After receiving the signal transmitted by the CPU, it immediately decodes the public message from it , when decoding, the private message part is regarded as interference and not decoded, AP decoding The signal to interference plus noise ratio is: (3); Define a Used to indicate the upper limit of the public message transmission rate of all APs. The public message transmission rate of all APs shall not exceed , so that all APs can decode the public message regardless of channel capacity , which is defined as follows: (4); Since the public message contains each user's message, it is necessary to determine the message segmentation ratio of each user and determine how many messages to segment into the public message; define For users The rate of the common part of the split, then the following inequality holds: (5); Once AP Complete public message The decoding will decode the private message through serial interference cancellation , A.P. decoding The SINR can be expressed as: (6); Similarly, to ensure that all APs serving a user can decode its private messages, the following maximum private message transmission rate is defined: (7); in Represents the set of APs serving the user; The forwarding rate of user k can be expressed as: (8).

[0009] A further technical solution of the present invention is that step S20 comprises: The optimization problem of maximizing the minimum forwarding rate is expressed as: (9); in , ; is the normalized CPU transmit power constraint; the optimization problem of maximizing the minimum forward transmission rate is non-convex and has complex coupling constraints. The problem is rewritten as follows: (10); in , , , , ; for dimensional diagonal matrix, The rest of the elements are 0; , , , are all non-negative auxiliary variables; definition Mapping of AP index to user index; is the CPU transmission power constraint. Since the transmission power constraint must be equal when the optimal solution is obtained, the original inequality constraint is converted into an equality constraint. and Moved to the objective function, we get the following augmented Lagrangian problem: (11); in It is to represent a conveniently defined set of decision variables; (12); (13); is the penalty coefficient, and is the dual variable; Decompose P3 into three sub-problems and solve them separately. Each sub-problem is optimized separately. , and ; First, solve for Sub-questions: (14); Will Expand to get: (15); in: (16); For complex vectors, we have the following theorem: (17); (18); Inequality (17) can be proved by the Cauchy-Schwarz inequality, and inequality (18) can be proved by using the properties of concave functions; when When , we can deduce from inequality (17): (19); in: (20); According to inequality (18), it can be deduced that: (twenty one); Similarly, when When , we can get: (twenty two); in: (31); For convenience of representation, the following matrix is ​​defined: (32); (33); So far, it has been deduced Local tight upper bound function of : (34); in For Irrelevant constants; According to the above steps, we can also deduce Local tight upper bound function of : (35); Then, P4 can be transformed into the following eigenvalue solution problem: (36); Obviously, the optimal solution of P5 is the matrix The eigenvector corresponding to the minimum eigenvalue of ; about The sub-problem can be expressed as: (37); This is a linear programming problem that can be solved using the interior point method; about The sub-problem can be expressed as: (38); in and They are and is a positive definite quadratic form, so the objective function is a convex function; however and It is non-convex and difficult to solve directly; and can be split into two convex subsets, namely and Therefore, we can use the branch and bound method to decompose P7 into two sub-problems, each of which is solved in a convex subspace, and finally compare the solutions of the two sub-problems to obtain the optimal solution of the original problem.

[0010] The RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method of the present invention has the following beneficial effects: In the RSMA-based CF-mMIMO wireless fronthaul network, the present invention maximizes the minimum fronthaul rate and ensures user fairness by designing beamforming and rate splitting ratios. The present invention first proposes a joint beamforming and rate splitting optimization problem. Based on the penalty dual decomposition method, the original complex non-convex problem is decomposed into three sub-problems, and each sub-problem is solved alternately, and finally convergence to a suboptimal solution is guaranteed. Compared with the WMMSE method, the proposed algorithm has better performance and lower complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0012] Figure 1 It is a flow chart of a preferred embodiment of the RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method of the present invention; Figure 2 It is a system diagram of the CF-mMIMO wireless fronthaul network model based on RSMA; Figure 3 This is a schematic diagram comparing the convergence speed of the algorithms; Figure 4 This is a schematic diagram comparing the performance of different multiple access schemes.

[0013] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] The present invention is aimed at a rate-splitting division multiple access (RSMA) CF-mMIMO wireless fronthaul network, takes maximizing the minimum fronthaul rate as the optimization goal, and proposes a low-complexity joint beamforming and rate division algorithm.

[0016] Specifically, the present invention proposes a RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method, such as Figure 1 As shown, a preferred embodiment of the RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method of the present invention includes the following steps: Step S10, constructing a CF-mMIMO wireless fronthaul network model based on RSMA, wherein the CF-mMIMO wireless fronthaul network model includes a Antenna CPU, M single antennas and K single antenna users, the forward transmission rate of user k is expressed as ,in, is the public portion rate divided by user k, is the maximum private message transmission rate of user k.

[0017] In order to reduce the deployment and operation and maintenance costs of the fronthaul network, the present invention considers using millimeter wave communication between the CPU and the AP. The system schematic diagram of the CF-mMIMO wireless fronthaul network model is as follows: Figure 2 shown.

[0018] In this embodiment, the step S10 specifically includes: make Represents the index set of all APs, A collection of indexes representing all users, Represented as user The set of AP indexes of the service; Forward message for each user , RSMA divides it into two parts: the common part and private part ; The public part from which all user messages are split Combined into a super public message , the private part is independent into corresponding private messages; therefore, the data stream emitted by the CPU is , each data stream is mapped to Transmitted on the antenna, the CPU's transmission signal It can be expressed as: (1); in Indicates public message The beamforming vector of Indicates private message The beamforming vector of the user Service AP , the signal it receives from the CPU is: (2); in Indicates CPU and AP The channel vector between For noise, is the noise variance; AP After receiving the signal transmitted by the CPU, it immediately decodes the public message from it , when decoding, the private message part is regarded as interference and not decoded, AP decoding The signal to interference plus noise ratio (SINR) is: (3); Define a Used to indicate the upper limit of the public message transmission rate of all APs. The public message transmission rate of all APs shall not exceed , so that all APs can decode the public message regardless of channel capacity , which is defined as follows: (4); Since the public message contains each user's message, it is necessary to determine the message segmentation ratio of each user and determine how many messages to segment into the public message; define For users The rate of the common part of the split, then the following inequality holds: (5); Once AP Complete public message The decoding will decode the private message through serial interference cancellation , A.P. decoding The SINR can be expressed as: (6); Similarly, to ensure that users All APs in the service can decode its private messages and define the following maximum private message transmission rate: (7); in Represented as user The set of APs to be served; The forwarding rate of user k can be expressed as: (8).

[0019] Step S20, based on the penalty dual decomposition method, decompose the maximization minimum forward transmission rate optimization problem into three sub-problems, alternately solve each sub-problem, and derive the minimum forward transmission rate among the maximized K users.

[0020] The step S20 specifically includes the following steps: First, for the joint beamforming and rate splitting problem of CF-mMIMO wireless fronthaul based on RSMA: The optimization problem of maximizing the minimum forwarding rate is expressed as: (9); in , ; is the normalized CPU transmit power constraint; the optimization problem of maximizing the minimum forward transmission rate is non-convex and has complex coupling constraints. The problem is rewritten as follows: (10); in , , , , ; for dimensional diagonal matrix, The rest of the elements are 0; , , , are all non-negative auxiliary variables; definition Mapping of AP index to user index; is the CPU transmission power constraint. Since the transmission power constraint must be equal when the optimal solution is obtained, the original inequality constraint is converted into an equality constraint. and Moved to the objective function, we get the following augmented Lagrangian problem (AL): (11); in It is to represent a conveniently defined set of decision variables; in, is the penalty coefficient, and is the dual variable.

[0021] Secondly, for the low-complexity joint beamforming and rate splitting algorithm: Decompose P3 into three sub-problems and solve them separately. Each sub-problem is optimized separately. , and ; First, solve for Sub-questions: (14); Will Expand to get: (15); in: (16); For complex vectors, we have the following theorem: (17); (18); Inequality (17) can be proved by the Cauchy-Schwarz inequality, and inequality (18) can be proved by using the properties of concave functions; when When , we can deduce from inequality (17): (19); in: (20); According to inequality (18), it can be deduced that: (twenty one); Similarly, when When , we can get: (twenty two); in: (31); For convenience of representation, the following matrix is ​​defined: (32); So far, it has been deduced Local tight upper bound function of : (34); in For Irrelevant constants; According to the above steps, we can also deduce Local tight upper bound function of : (35); Then, P4 can be transformed into the following eigenvalue solution problem: (36); Obviously, the optimal solution of P5 is the matrix The eigenvector corresponding to the minimum eigenvalue of ; about The sub-problem can be expressed as: (37); This is a linear programming problem that can be solved using the interior point method.

[0022] about The sub-problem can be expressed as: (38); in and They are and is a positive definite quadratic form, so the objective function is a convex function; however and It is non-convex and difficult to solve directly; and can be split into two convex subsets, namely and Therefore, we can use the branch and bound method to decompose P7 into two sub-problems, each of which is solved in a convex subspace, and finally compare the solutions of the two sub-problems to obtain the optimal solution of the original problem.

[0023] The present invention summarizes the joint beamforming and rate splitting algorithm based on the penalty dual decomposition method in Algorithm 1 shown in Table 1, where: (43) (44) Table 1

[0024] In order to better demonstrate the effectiveness of the present invention, the present invention conducted a simulation experiment, and verified the performance of the RSMA-based de-cellular massive MIMO wireless fronthaul network beamforming and rate splitting method proposed in the present invention through simulation.

[0025] We assume that there are 100 APs evenly distributed in a 100m*100m site, each with 1 antenna and 4 users. The channel model takes into account the clustered millimeter wave channel model, assuming that the number of CPU antennas is 64 and the maximum transmission power is 15dbm.

[0026] Figure 3 It is verified that the proposed joint beamforming and rate splitting algorithm based on the penalty dual decomposition method can converge, and its convergence speed is faster than that of the algorithm based on WMMSE, and its performance is also better than that of the WMMSE algorithm.

[0027] Figure 4 It shows that RSMA has a more obvious performance advantage in overload conditions with insufficient number of antennas, too large group size and too many users. TDMA is least affected by the above three indicators, while SDMA is most affected. In non-overload conditions, there is no significant difference in the performance of RSMA and SDMA in the CF-mMIMO wireless fronthaul network. This shows that the wireless fronthaul network based on RSMA is more adaptable to the dynamic changes of interference levels.

[0028] In terms of algorithm complexity, the complexity of the algorithm we proposed is , the algorithm complexity based on WMMSE is ,in and are the total number of iterations of the algorithm. In the CF-mMIMO network, , so our proposed algorithm has lower complexity.

[0029] The RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method of the present invention has the following beneficial effects: In the RSMA-based CF-mMIMO wireless fronthaul network, the present invention maximizes the minimum fronthaul rate and ensures user fairness by designing beamforming and rate splitting ratios. The present invention first proposes a joint beamforming and rate splitting optimization problem. Based on the penalty dual decomposition method, the original complex non-convex problem is decomposed into three sub-problems, and each sub-problem is solved alternately, and finally convergence to a suboptimal solution is guaranteed. Compared with the WMMSE method, the proposed algorithm has better performance and lower complexity.

[0030] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method, characterized in that: The method comprises the following steps: Step S10, constructing a CF-mMIMO wireless fronthaul network model based on RSMA, wherein the CF-mMIMO wireless fronthaul network model includes a Antenna CPU, M single antennas and K single antenna users, the forward transmission rate of user k is expressed as ,in, is the public portion rate divided by user k, is the maximum private message transmission rate of user k; Step S20, based on the penalty dual decomposition method, decompose the maximization minimum forward transmission rate optimization problem into three sub-problems, alternately solve each sub-problem, and derive the minimum forward transmission rate among the maximized K users.

2. The RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method according to claim 1, characterized in that: The step S10 comprises: make Represents the index set of all APs, A collection of indexes representing all users, Represented as user The set of AP indexes of the service; Forward message for each user , RSMA divides it into two parts: the public part and private part ; The public part from which all user messages are split Combined into a super public message , the private part is independent into corresponding private messages; therefore, the data stream emitted by the CPU is , each data stream is mapped to Transmitted on the antenna, CPU's transmission signal It can be expressed as: (1); in Indicates public message The beamforming vector of Indicates private message The beamforming vector of the user Service AP , the signal it receives from the CPU is: (2)； in Indicates CPU and AP The channel vector between For noise, is the noise variance; AP After receiving the signal transmitted by the CPU, it immediately decodes the public message from it , when decoding, the private message part is regarded as interference and not decoded, AP decoding The signal to interference plus noise ratio is: (3)； Define a Used to indicate the upper limit of the public message transmission rate of all APs. The public message transmission rate of all APs shall not exceed , so that all APs can decode the public message regardless of channel capacity , which is defined as follows: (4)； Since the public message contains each user's message, it is necessary to determine the message segmentation ratio of each user and determine how many messages to segment into the public message; define For users The rate of the common part of the split, then the following inequality holds: (5)； Once AP Complete public message The decoding will decode the private message through serial interference cancellation , A.P. decoding The SINR can be expressed as: (6)； Similarly, to ensure that all APs serving a user can decode its private messages, the following maximum private message transmission rate is defined: (7)； in Represents the set of APs serving the user; The forwarding rate of user k can be expressed as: (8)。 3. The RSMA-based decellularized massive MIMO wireless fronthaul network beamforming and rate splitting method according to claim 2, characterized in that: The step S20 comprises: The optimization problem of maximizing the minimum forwarding rate is expressed as: (9)； in , ; is the normalized CPU transmit power constraint; the optimization problem of maximizing the minimum forward transmission rate is non-convex and has complex coupling constraints. The problem is rewritten as follows: (10)； in , , , , ; for dimensional diagonal matrix, The rest of the elements are 0; , , , are all non-negative auxiliary variables; definition Mapping of AP index to user index; is the CPU transmission power constraint. Since the transmission power constraint must be equal when the optimal solution is obtained, the original inequality constraint is converted into an equality constraint. and Moved to the objective function, we get the following augmented Lagrangian problem: (11)； in It is to represent a conveniently defined set of decision variables; (12)； (13)； is the penalty coefficient, and is the dual variable; Decompose P3 into three sub-problems and solve them separately. Each sub-problem is optimized separately. , and ; First, solve for Sub-questions: (14)； Will Expand to get: (15)； in: (16)； For complex vectors, we have the following theorem: (17)； (18)； Inequality (17) can be proved by the Cauchy-Schwarz inequality, and inequality (18) can be proved by using the properties of concave functions; when When , we can deduce from inequality (17): (19)； in: (20)； According to inequality (18), it can be deduced that: (21)； Similarly, when When , we can get: (22)； in: (31)； For convenience of representation, the following matrix is ​​defined: (32)； (33)； So far, it has been deduced Local tight upper bound function of : (34)； in For Irrelevant constants; According to the above steps, we can also deduce Local tight upper bound function of : (35)； Then, P4 can be transformed into the following eigenvalue solution problem: (36)； Obviously, the optimal solution of P5 is the matrix The eigenvector corresponding to the minimum eigenvalue of ; about The sub-problem can be expressed as: (37)； This is a linear programming problem that can be solved using the interior point method; about The sub-problem can be expressed as: (38)； in and They are and is a positive definite quadratic form, so the objective function is a convex function; however and It is non-convex and difficult to solve directly; and can be split into two convex subsets, namely and Therefore, we can use the branch and bound method to decompose P7 into two sub-problems, each of which is solved in a convex subspace, and finally compare the solutions of the two sub-problems to obtain the optimal solution of the original problem.

Images