A method for ultra-massive MIMO dynamic radio frequency allocation

Abstract

This invention discloses a dynamic radio frequency allocation method for ultra-large-scale MIMO (UML), relating to the field of wireless communication physical layer technology. First, the method proposes a novel concept of "water-filled driven effective degrees of freedom," which incorporates system transmit power and accurately characterizes the maximum number of sub-channels that can be effectively activated under a given power constraint. Second, a hybrid precoding optimization algorithm is designed to solve for the optimal precoding matrix under a given number of radio frequency links. Finally, based on the water-filled driven effective degrees of freedom evaluation and the hybrid precoding optimization algorithm, a dynamic radio frequency allocation scheme is constructed, dynamically activating and deactivating radio frequency links according to real-time channel conditions to adapt to channel changes. By introducing water-filled driven effective degrees of freedom and proposing a complete dynamic radio frequency allocation and precoding optimization process, this invention effectively improves the energy efficiency of UML systems, providing an innovative solution for promoting the development of green communication in UML.

Description

Technical Field

[0001] This invention relates to the field of wireless communication physical layer technology, and in particular to a dynamic radio frequency allocation method for ultra-large-scale MIMO. Background Technology

[0002] Ultra-large-scale MIMO (Multi-Size MIMO) technology is one of the core supporting technologies for future wireless communication systems. By deploying unprecedentedly large-scale antenna arrays, it can greatly improve system capacity and reliability. However, the order-of-magnitude increase in antenna size also expands the system's operating region from the far field to the near field, fundamentally changing channel characteristics. Among these changes, the emergence of the spherical wavefront effect breaks the limitations of the traditional planar wavefront model, bringing new degrees of freedom to the system. How to effectively utilize this near-field characteristic to improve system performance has become a key research issue.

[0003] In near-field ultra-large-scale MIMO systems, effective degrees of freedom (DFS) are a core metric for measuring channel capacity, reflecting the number of orthogonal subchannels that can be effectively utilized. However, the existing concept of DFS has a fundamental limitation: it only characterizes the inherent number of subchannels with large singular values, without considering the strict transmit power constraints in real-world systems. In real communication systems, to maximize spectral efficiency, not all potentially high-quality subchannels need to be activated; the actual number of activated subchannels must be dynamically determined based on the total power budget using power allocation algorithms such as water-filling. Therefore, traditional DFS metrics are difficult to directly guide RF chain resource allocation under power-constrained conditions, easily leading to system design deviations from optimal energy efficiency.

[0004] On the other hand, the introduction of very large-scale antenna arrays has dramatically highlighted the problem of system energy consumption. All-digital precoding architectures are difficult to implement practically due to their high hardware complexity and power consumption. While partially connected hybrid precoding architectures significantly reduce hardware overhead, their performance largely depends on the match between the number of radio frequency chains and the actual number of data streams the channel can support. In near-field scenarios, the effective degrees of freedom of the channel dynamically change with system parameters such as distance and frequency. Using a fixed number of radio frequency chains will lead to two types of resource waste: when the number of radio frequency chains exceeds the actual requirement, excessive energy consumption occurs; when the number of radio frequency chains is insufficient, the channel capacity cannot be fully utilized.

[0005] Therefore, there is an urgent need for a new metric that can accurately reflect the actual number of activatable sub-channels under power constraints, as well as a dynamic radio frequency chain allocation mechanism based on this metric. By adjusting the number of activated radio frequency chains in real time to precisely match the actual support capacity of the current channel, it is possible to maximize system energy efficiency while ensuring spectral efficiency. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a dynamic radio frequency allocation method for ultra-large-scale MIMO. By introducing water-filled driving effective degrees of freedom, a complete dynamic radio frequency allocation and precoding optimization process is proposed, which effectively improves the energy efficiency of ultra-large-scale MIMO system and provides guidance for the design of ultra-large-scale MIMO system.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0008] A dynamic radio frequency allocation method for ultra-large-scale MIMO proposed according to the present invention includes:

[0009] Step S1: For ultra-large-scale MIMO systems, define the effective degrees of freedom for channel water-filling drive;

[0010] Step S2: Design a dynamic hybrid precoding architecture;

[0011] Step S3: Under the dynamic hybrid precoding architecture, given the constraint on the number of RF chain activations, perform alternating iterative optimization of analog precoding, digital precoding, and power allocation in the dynamic hybrid precoding architecture;

[0012] Step S4: Based on the channel's water-filled driving effective degrees of freedom, design the outer loop of the dynamic radio frequency allocation method. The outer loop of the dynamic radio frequency allocation method refers to achieving the optimal match between the number of active radio frequency chains and channel performance through closed-loop iteration.

[0013] As a further optimization of the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, the outer loop of the dynamic radio frequency allocation method is designed to include:

[0014] The first step is to dynamically adjust the number of active radio frequency links based on the effective degrees of freedom driven by the water-filling of the current equivalent channel.

[0015] The second step is to use the alternating optimization method in step S3 to solve for the optimal analog precoding, digital precoding and power allocation corresponding to the adjusted number of RF chain activations.

[0016] The third step is to update the equivalent channel based on the optimal analog precoding and recalculate the water-filled drive effective degrees of freedom of the updated equivalent channel. If the water-filled drive effective degrees of freedom are not equal to the current number of active RF links, return to the first step and repeat the iteration. If the water-filled drive effective degrees of freedom are equal to the current number of active RF links, output the final optimal number of active RF links, as well as the optimal analog precoding, digital precoding and power allocation corresponding to the optimal number of active RF links.

[0017] As a further optimization scheme of the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, the equivalent channel refers to the superposition of the real channel, analog precoding, and digital precoding.

[0018] As a further optimization scheme for the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, the gradient ascent algorithm is used to optimize the analog precoding, and the eigenvalue decomposition and water-filling algorithm principles are used to jointly optimize the digital precoding and power allocation.

[0019] As a further optimization of the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, step 1 includes:

[0020] Step S101: Consider a single-user near-field ultra-large-scale MIMO communication system, wherein the transmitter is configured with a... A uniform planar array of antennas, directed towards a receiver including... A uniform planar array of antennas transmits signals; the received signal vector at the receiver... It is given by the following formula:

[0021] , ;

[0022] in, , Representing the The received signal from each receiving antenna, superscript Indicates transpose. , express A complex matrix of dimension 1; The transmitted signal vector, Representing the The transmitted signal from each transmitting antenna, ,vector This represents additive white Gaussian noise, whose elements have a mean of 0 and a variance of . complex Gaussian distribution ; The channel matrix based on the Green's function is expressed as follows:

[0023] , ;

[0024] in, Representing the The transmitting antenna to the first The channel coefficient between the receiving antennas is defined as follows:

[0025] , ;

[0026] in, and They represent the first in the transmission array. The first antenna and receiver array The spatial location of each antenna k is the wave number. Wavelength;

[0027] Step S102: Based on the ultra-large-scale MIMO communication system, the effective degrees of freedom for water-filled drive are defined as follows: a) for the channel matrix a) Perform singular value decomposition to obtain its singular values; b) Given the total transmit power, use the water-filling algorithm based on the singular values ​​to obtain the optimal power allocation scheme; c) Count the number of streams with non-zero allocated power, and this number of streams is the effective degree of freedom of water-filling drive.

[0028] As a further optimization of the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, step S2 includes:

[0029] Step S201: Considering that the ultra-large-scale MIMO communication system adopts a dynamic partial connection hybrid precoding scheme, based on step S101, the signal vector is transmitted. Represented as

[0030] , ;

[0031] in, Indicates length is Unit data stream, power allocation matrix , Representative assigned to the first The power of each data stream satisfies the constraint. , , This represents the upper limit of the system's total transmit power. Indicates the number of active RF links, let... ,matrix It is a digital precoding matrix. To simulate the precoding matrix;

[0032] Step S202 Employing a partially connected architecture, each antenna is connected to a radio frequency (RF) link via a single-pole multi-throw (SPMW) switch. The SPMWS enables antenna-RF link mapping, allowing real-time adjustment of the active RF link. Assuming the active RF link is... One radio frequency link, and will Each antenna is evenly distributed among the active RF links, and the antenna indices are consecutive; at this time, Represented as:

[0033] , ;

[0034] No. non-zero column vectors , , represented as:

[0035] ;

[0036] in, For mold taking operation, Indicates the first RF link to the first The gain coefficient of each transmitting antenna, , and They respectively represent connections to the first The start and end antenna indices of each radio frequency link. This is the power normalization factor.

[0037] ;

[0038] in, This indicates the number of antennas allocated to each radio frequency link. , Indicates the number of remaining antennas, such that the first An additional antenna is allocated to each radio frequency link.

[0039] As a further optimization scheme of the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, step S3 specifically includes the following steps:

[0040] Step S301, according to the formula and formula The received signal at the receiving end is represented as:

[0041] ;

[0042] The corresponding spectral efficiency is:

[0043] , ;

[0044] in, Represents the system's spectral efficiency. represent 1D identity matrix Represents noise power, superscript It is the conjugate transpose;

[0045] At a given number of RF chain activations The following method employs an alternating optimization scheme to alternately optimize the analog precoding matrix. Digital precoding matrix and power allocation matrix Each iteration of the alternating optimization scheme consists of two parts: first, the optimization of the analog precoding matrix; and second, the joint optimization of the digital precoding matrix design and the power allocation matrix, until the convergence condition is met.

[0046] Step S302, Analog precoding matrix optimization: Fix the digital precoding matrix from the previous iteration or initial setting. With power allocation matrix Unchanged; subsequently, to maximize system spectral efficiency To achieve the objective, the gradient ascent algorithm is used to optimize the simulated precoding matrix; the mathematical problem of simulated precoding matrix optimization is described as follows:

[0047] ;

[0048] in, For system spectral efficiency, the constraints require... Specific elements in the equation satisfy the formula The constant modulus constraint given in the document;

[0049] Gradient Ascent Algorithm Initialization: Input Channel Matrix Initial analog precoding matrix Digital precoding matrix Power allocation matrix and noise power Set temporary variables Used to store the latest analog precoding matrix during the iteration process, where The operation means assigning the value of the variable on the right to the value on the left. The initial simulation precoding matrix is ​​generated randomly;

[0050] Begin iterative loops of the gradient ascent algorithm until the convergence condition is met; at the start of each iteration, Assigned to the analog precoding matrix ,Right now ;

[0051] Calculate spectral efficiency about Euclidean gradient :

[0052] , ;

[0053] in, Representative calculation about Euclidean gradient, auxiliary matrix Defined as ;

[0054] Iterate through and process each non-zero element in the analog precoding matrix. Each non-zero element that needs optimization , corresponding to The Line number Column elements;

[0055] Calculate the Riemann gradient, and project the calculated Euclidean gradient onto the tangent space of the constant modulus constrained manifold to obtain the corresponding Riemann gradient:

[0056] ; ;

[0057] in, Representative calculation about The Riemann gradient, Representing the first of the matrix Line number Column elements, Represents the operation of taking the real part, superscript To obtain conjugate;

[0058] Update element values ​​along the Riemann gradient direction with a preset step size parameter. Update the current element value:

[0059] , ;

[0060] in, The element values ​​are updated along the Riemann gradient direction. The preset step size parameter controls the magnitude of each update;

[0061] The manifold shrinking operation reprojects the updated element values ​​back onto the constant modulus constrained manifold, ensuring that the modulus constraint is satisfied.

[0062] ; ;

[0063] in, Represents the updated element value. Represents the phase taking a complex number. The base is the natural number;

[0064] Then, after updating all elements in the current gradient ascent algorithm iteration, we obtain (14) ;

[0065] Check the convergence condition of the gradient ascent algorithm and calculate... Corresponding spectral efficiency value And compared with the spectral efficiency value of the previous iteration Comparison, if the system's spectral efficiency The gain change is less than a preset minimum positive threshold. If the preset maximum number of iterations is reached, the gradient ascent algorithm is considered to have converged, and the iteration loop of the gradient ascent algorithm is exited. The value is assigned to the analog precoding matrix As output, i.e. Otherwise, repeat the gradient ascent algorithm iteration for the next round.

[0066] Step S303, Joint optimization of digital precoding matrix design and power allocation matrix: Update the analog precoding matrix in step S302. After that, fix The current latest value is used; subsequently, the digital precoding matrix is ​​generated using the principles of eigenvalue decomposition and water-filling algorithms. and power allocation matrix Joint optimization design;

[0067] Received channel matrix Noise power and the system's total transmit power constraint Analog precoding matrix ;

[0068] Perform eigenvalue decomposition on the equivalent channel matrix. Perform eigenvalue decomposition:

[0069] , ;

[0070] in, It is a unitary matrix, and its column vectors are eigenvectors; For a diagonal matrix, its diagonal elements The eigenvalues ​​are sorted in descending order; then... Assign to ,Right now This yields the digital precoding matrix output.

[0071] Optimize the power allocation matrix based on the water injection algorithm principle; initialize parameters and set the iteration counter. ;

[0072] Calculate the water injection level, for the first The next iteration calculates the current water injection level. :

[0073] ; ;

[0074] For the first The number of sub-channels considered in each iteration, where the first iteration considers all Sub-channels, i.e. ; It is the first eigenvalues ​​arranged in descending order;

[0075] Simultaneously, based on the current water injection level, the test power value for each sub-channel is calculated. :

[0076] ; ;

[0077] Check if the minimum test power satisfies the positive constraint. ,like Then set Remove the first active sub-channel from the currently active sub-channel set. Sub-channels, update , Return to formula (16) for the next iteration; if If so, then perform the following operations;

[0078] remember The optimal power allocation scheme is as follows:

[0079] , ;

[0080] in The final water level for the iteration. The final number of activated sub-channels, i.e. The number of sub-channels; at this point, the power allocation matrix is ​​set as follows: ; for The values ​​of the elements on the diagonal;

[0081] Step S304, completing steps S302 and S303 constitutes a complete iteration of the alternating optimization algorithm; the updated digital precoding matrix from step S303 is then... and power allocation matrix As the initial value for the next iteration, steps S302 and S303 are repeated; this alternating optimization process continues until the system's spectral efficiency is achieved. The gain change is less than a preset minimum positive threshold. If the preset maximum number of iterations is reached, the alternating optimization algorithm will declare convergence and terminate the iteration.

[0082] Finally, the alternating optimization algorithm outputs the converged simulated precoding matrix. Digital precoding matrix and power allocation matrix Together they constitute the given A hybrid precoder that maximizes the system's spectral efficiency.

[0083] As a further optimization of the ultra-large-scale MIMO dynamic radio frequency allocation method described in this invention, step S4 includes:

[0084] Step S401: Based on the effective degrees of freedom of water-filled drive defined in step S1 and the alternating optimization algorithm proposed in step S3, design the outer loop of the dynamic radio frequency allocation scheme.

[0085] Step S402, system initialization, according to Calculate its effective degrees of freedom driven by water filling. The effective degree of freedom value driven by the water filling is used as the initial value for the number of activated radio frequency chains:

[0086] ; ;

[0087] Simultaneously set , as well as ;

[0088] Step S403: Begin the adaptive adjustment loop. In each iteration, first execute the alternating optimization algorithm to update the corresponding analog precoding matrix. Digital precoding matrix and power allocation matrix Then utilize the current analog precoding matrix and digital precoding matrix , and the channel matrix Multiply to construct the equivalent channel matrix

[0089] , ;

[0090] This equivalent channel matrix reflects the end-to-end channel characteristics after processing by the current precoder; then, the water-filled effective degrees of freedom of the equivalent channel matrix are calculated. According to the definition of effective degrees of freedom for water filling, we have

[0091] , ;

[0092] Step S404: Determine if the RF chain configuration is optimal by comparing the current configuration. With equivalent channel effective degrees of freedom ,like Adjust the current RF chain configuration to At the same time, return to step S403;

[0093] like This indicates that the current number of RF chains matches the number of data streams supported by the channel, and the configuration has reached its optimal state. The loop exits, and the current value is returned as the final optimization result: the optimal number of RF chains. Optimal analog precoding matrix Optimal digital precoding matrix Optimal power allocation matrix .

[0094] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0095] This invention proposes a dynamic radio frequency allocation method for ultra-large-scale MIMO. Compared with the fixed number of active radio frequencies used in existing literature, this method can adjust the radio frequency units in real time according to changes in the channel's spatial multiplexing capability, thereby significantly improving the spectral efficiency and energy efficiency of the communication system. Furthermore, this invention defines the effective degree of freedom for channel water-filling as a standard for measuring channel spatial multiplexing capability. Compared to traditional effective degrees of freedom, this standard can more accurately characterize the channel's spatial multiplexing capability, thus making the dynamic adjustment of radio frequency units more precise and effective. Therefore, this invention has important guiding significance for the design of future ultra-large-scale MIMO systems. Attached Figure Description

[0096] Figure 1 A flowchart illustrating a dynamic radio frequency allocation scheme for ultra-large-scale MIMO based on water-filled driven effective degrees of freedom, according to an embodiment of the present invention.

[0097] Figure 2 This is a schematic diagram of an ultra-large-scale MIMO system according to an embodiment of the present invention;

[0098] Figure 3 This is a diagram of the dynamic hybrid precoding architecture according to an embodiment of the present invention;

[0099] Figure 4 This invention presents a comparison of the optimal number of radio frequencies obtained based on water-filled driving effective degrees of freedom and the optimal number of radio frequencies obtained based on conventional effective degrees of freedom in terms of spectral efficiency and energy efficiency.

[0100] Figure 5This is a comparison of the dynamic radio frequency allocation scheme based on water-filled driving effective degrees of freedom and the fixed radio frequency scheme in terms of spectral efficiency and energy efficiency in the embodiments of the present invention; wherein, (a) is a graph of spectral efficiency as a function of distance, and (b) is a graph of energy efficiency as a function of distance. Detailed Implementation

[0102] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0103] Figure 1 This is a flowchart of a dynamic radio frequency allocation scheme for ultra-large-scale MIMO based on water-filled driving effective degrees of freedom, according to an embodiment of the present invention.

[0104] like Figure 1 As shown, this ultra-large-scale MIMO dynamic radio frequency allocation scheme based on water-filled driving effective degrees of freedom includes the following steps:

[0105] Step S1, for Figure 2 The ultra-large-scale MIMO system shown defines the water-filled drive effective degrees of freedom of the channel.

[0106] In an embodiment of the present invention, step S1 specifically includes the following steps:

[0107] Step S101: Consider a single-user near-field ultra-large-scale MIMO communication system, wherein the transmitter is configured with a device containing... A uniform planar array of antennas, directed towards a receiver containing... A uniform planar array of antennas transmits signals. The received signal at its receiver... It is given by the following formula:

[0108] ,

[0109] in, , Representing the The received signal from each receiving antenna, superscript Indicates transpose. , express A complex matrix of dimension 1; The transmitted signal vector, Representing the The transmitted signal from each transmitting antenna, ,vector This represents additive white Gaussian noise, whose elements have a mean of 0 and a variance of . complex Gaussian distribution ; The channel matrix based on the Green's function is expressed as follows:

[0110] ,

[0111] Among them, elements Representing the The transmitting antenna to the first The channel coefficients between receiving days are defined as follows:

[0112] ,

[0113] in, and They represent the first in the transmission array. The first antenna and receiver array The spatial location of each antenna k is the wave number. Wavelength;

[0114] Step S102: Based on the ultra-large-scale MIMO communication system, the effective degrees of freedom for water-filling drive are defined as follows: a) for the channel matrix a) Perform singular value decomposition to obtain its singular values; b) Given the total transmit power, use the water-filling algorithm based on the singular values ​​to obtain the optimal power allocation scheme; c) Count the number of streams with non-zero allocated power, and this number of streams is the effective degree of freedom of water-filling drive.

[0115] Compared to the traditional definition of effective degrees of freedom, the water-filled driven effective degrees of freedom takes into account the system's transmit power, which can accurately characterize the maximum number of sub-channels that can be effectively activated under a given power constraint.

[0116] Step S2, design a dynamic hybrid precoding architecture, such as Figure 3 As shown.

[0117] In an embodiment of the present invention, step S2 specifically includes the following steps:

[0118] Step S201: Design a very large-scale MIMO communication system using a dynamic partial connection hybrid precoding scheme, such as... Figure 3 As shown, based on step S101, the signal vector is transmitted. Represented as

[0119] ,

[0120] in, Indicates length is Unit data stream, power allocation matrix , Representative assigned to the first The power of each data stream satisfies the constraint. , , This represents the upper limit of the system's total transmit power. Indicates the number of active RF links, let... ,matrix It is a digital precoding matrix. To simulate the precoding matrix;

[0121] Step S202: Employing a partially connected architecture, each antenna is connected to a radio frequency (RF) link via a single-pole multi-throw (SPMW) switch. The SPMWS enables antenna-RF link mapping, allowing real-time adjustment of the active RF link. Assuming the active RF link is... One radio frequency link, and will Each antenna is evenly distributed among the active RF links, and the antenna indices are consecutive; at this time, Represented as:

[0122] ,

[0123] No. non-zero column vectors , , represented as:

[0124] ,

[0125] in, For mold taking operation, Indicates the first RF link to the first The gain coefficient of each transmitting antenna, , and They respectively represent connections to the first The start and end antenna indices of each radio frequency link. These are power normalization factors; they are given by the following formula:

[0126]

[0127] in, This indicates the number of antennas allocated to each radio frequency link. , Indicates the number of remaining antennas, such that the first An additional antenna is allocated to each radio frequency link.

[0128] Step S3, for a given number of RF chains activated Design an alternating optimization scheme to optimize the analog precoding matrix. Digital precoding matrix and power allocation matrix In order to maximize the spectral efficiency of the system.

[0129] In an embodiment of the present invention, step S3 specifically includes the following steps:

[0130] Step S301: According to the formula and formula The received signal at the receiving end is represented as:

[0131] ,

[0132] The corresponding spectral efficiency is:

[0133] , ;

[0134] in, Represents the system's spectral efficiency. represent 1D identity matrix Represents noise power, superscript It is the conjugate transpose;

[0135] At a given number of RF chain activations The following method employs an alternating optimization scheme to alternately optimize the analog precoding matrix. Digital precoding matrix and power allocation matrix Each iteration of the alternating optimization scheme consists of two parts: first, the optimization of the analog precoding matrix; and second, the joint optimization of the digital precoding matrix design and the power allocation matrix, until the convergence condition is met.

[0136] Step S302: Analog precoding matrix optimization: Fix the digital precoding matrix from the previous iteration or initial settings. With power allocation matrix Unchanged; subsequently, to maximize system spectral efficiency To achieve the objective, the gradient ascent algorithm is used to optimize the simulated precoding matrix; the mathematical problem of simulated precoding matrix optimization is described as follows:

[0137]

[0138] in, For system spectral efficiency, the constraints require... Specific elements in the equation satisfy the formula The constant modulus constraint given in the document;

[0139] Gradient Ascent Algorithm Initialization: Input Channel Matrix Initial analog precoding matrix Digital precoding matrix Power allocation matrix and noise power Set temporary variables Used to store the latest analog precoding matrix during the iteration process, where The operation means assigning the value of the variable on the right to the value on the left. The initial simulation precoding matrix is ​​generated randomly;

[0140] Begin iterative loops of the gradient ascent algorithm until the convergence condition is met; at the start of each iteration, Assigned to the analog precoding matrix ,Right now ;

[0141] Calculate spectral efficiency about Euclidean gradient :

[0142] ,

[0143] in, Representative calculation about Euclidean gradient, auxiliary matrix Defined as ;

[0144] Iterate through and process each non-zero element in the analog precoding matrix. Each non-zero element that needs optimization , corresponding to The Line number Column elements;

[0145] Calculate the Riemann gradient, and project the calculated Euclidean gradient onto the tangent space of the constant modulus constrained manifold to obtain the corresponding Riemann gradient:

[0146] ;

[0147] in, Representative calculation about The Riemann gradient, Representing the first of the matrix Line number Column elements, Represents the operation of taking the real part, superscript To obtain the conjugate; this operation ensures that the optimization direction proceeds along the tangent space of the manifold, satisfying the constraints;

[0148] Update element values ​​along the Riemann gradient direction with a preset step size parameter. Update the current element value:

[0149] ,

[0150] in, The element values ​​are updated along the Riemann gradient direction. The preset step size parameter controls the magnitude of each update; here, the step size... Take a set of values For each value, a simulated precoding matrix is ​​updated.

[0151] The manifold shrinking operation reprojects the updated element values ​​back onto the constant modulus constrained manifold, ensuring that the modulus constraint is satisfied.

[0152] ;

[0153] in, Represents the updated element value. Represents the phase taking a complex number. The base is the natural number; this operation preserves the phase information of the updated element while adjusting its modulus to a specified constant value.

[0154] Then, after updating all elements in the current gradient ascent algorithm iteration, according to get ;

[0155] Check the convergence condition of the gradient ascent algorithm and calculate... Corresponding spectral efficiency value And compared with the spectral efficiency value of the previous iteration Comparison, if the system's spectral efficiency The gain change is less than a preset minimum positive threshold. If the maximum number of iterations (20) is reached, the gradient ascent algorithm is considered to have converged, and the iteration loop of the gradient ascent algorithm is exited. The value is assigned to the analog precoding matrix As output, i.e. Otherwise, repeat the gradient ascent algorithm iteration for the next round.

[0156] Step S303: Joint optimization of digital precoding matrix design and power allocation matrix: Update the analog precoding matrix in step S302. After that, fix The current latest value is used; subsequently, the digital precoding matrix is ​​generated using the principles of eigenvalue decomposition and water-filling algorithms. and power allocation matrix Joint optimization design;

[0157] Received channel matrix Noise power and the system's total transmit power constraint Analog precoding matrix ;

[0158] Perform eigenvalue decomposition on the equivalent channel matrix. Perform eigenvalue decomposition:

[0159] ,

[0160] in, It is a unitary matrix, and its column vectors are eigenvectors; For a diagonal matrix, its diagonal elements The eigenvalues ​​are sorted in descending order; then... Assign to ,Right now This yields the digital precoding matrix output; this setting enables the diagonalization of the equivalent channel, decomposing the MIMO channel into multiple parallel independent sub-channels.

[0161] Optimize the power allocation matrix based on the water injection algorithm principle; initialize parameters and set the iteration counter. ;

[0162] Calculate the water injection level, for the first The next iteration calculates the current water injection level. :

[0163] ;

[0164] For the first The number of sub-channels considered in each iteration, where the first iteration considers all Sub-channels, i.e. ; It is the first eigenvalues ​​arranged in descending order;

[0165] Simultaneously, based on the current water injection level, the test power value for each sub-channel is calculated. :

[0166] ;

[0167] Check if the minimum test power satisfies the positive constraint. ,like Then set Remove the first active sub-channel from the currently active sub-channel set. Sub-channels, update , Return to formula Proceed to the next iteration; if If so, then perform the following operations;

[0168] remember The optimal power allocation scheme is as follows:

[0169] ,

[0170] in The final water level for the iteration. The final number of activated sub-channels, i.e. The number of sub-channels; at this point, the power allocation matrix is ​​set as follows: ; for The values ​​of the elements on the diagonal;

[0171] Step S304: Completing steps S302 and S303 constitutes one complete iteration of the alternating optimization algorithm; the updated digital precoding matrix from step S303 is then... and power allocation matrix As the initial value for the next iteration, steps S302 and S303 are repeated; this alternating optimization process continues until the system's spectral efficiency is achieved. The gain change is less than a preset minimum positive threshold. If the preset maximum number of iterations of 100 is reached, the alternating optimization algorithm will declare convergence and terminate the iteration.

[0172] Finally, the alternating optimization algorithm outputs the converged simulated precoding matrix. Digital precoding matrix and power allocation matrix Together they constitute the given A hybrid precoder that maximizes the system's spectral efficiency.

[0173] Step S4: Using the water-filling effect of the channel to drive the effective degrees of freedom, and based on the alternating optimization scheme, design the outer loop of the dynamic radio frequency allocation scheme.

[0174] In an embodiment of the present invention, step S4 specifically includes the following steps:

[0175] Step S401: Based on the effective degrees of freedom of water-filling drive defined in step S1 and the alternating optimization scheme proposed in step S3, design the outer loop of the dynamic radio frequency allocation scheme;

[0176] Step S402: System initialization, according to Calculate its effective degrees of freedom driven by water filling. The effective degree of freedom value driven by the water filling is used as the initial value for the number of activated radio frequency chains:

[0177] ;

[0178] Simultaneously set , as well as ;

[0179] Step S403: Start the adaptive adjustment loop. In each loop iteration, first execute the alternating optimization algorithm to update the corresponding analog precoding matrix. Digital precoding matrix and power allocation matrix Then utilize the current analog precoding matrix and digital precoding matrix , and the channel matrix Multiply to construct the equivalent channel matrix

[0180] ,

[0181] This equivalent channel matrix reflects the end-to-end channel characteristics after processing by the current precoder; then, the water-filled effective degrees of freedom of the equivalent channel matrix are calculated. According to the definition of effective degrees of freedom for water filling, we have

[0182] ,

[0183] in The parameters obtained in step S303;

[0184] Step S404: Determine if the RF chain configuration is optimal by comparing it with the current configuration. With equivalent channel effective degrees of freedom ,like Adjust the current RF chain configuration to At the same time, return to step S403;

[0185] like This indicates that the current number of RF chains matches the number of data streams supported by the channel, and the configuration has reached its optimal state. The loop exits, and the current value is returned as the final optimization result: the optimal number of RF chains. Optimal analog precoding matrix Optimal digital precoding matrix Optimal power allocation matrix .

[0186] Based on steps S1-S4 of the embodiments of the present invention, the simulation parameters for the number of transmitting and receiving antennas are set. ,wavelength Array antenna spacing Relative transmit power value Relative noise power value When the distance between the centers of the transmitting and receiving arrays is fixed as At that time, one can obtain Figure 4 The simulation results shown indicate that when the center distance between the transmitting and receiving arrays is... exist When the interval changes, we can obtain Figure 5 The simulation results are shown.

[0187] This invention proposes a dynamic radio frequency allocation scheme for ultra-large-scale MIMO based on water-filled driving effective degrees of freedom. First, it defines the effective degrees of freedom for water-filled driving, which takes system transmit power into account and accurately characterizes the maximum number of sub-channels that can be effectively activated under a given power constraint. Second, it designs a hybrid precoding optimization algorithm to solve for the optimal precoding matrix under a given number of radio frequency links. Finally, based on the evaluation of the effective degrees of freedom for water-filled driving and the hybrid precoding optimization algorithm, a dynamic radio frequency allocation scheme is constructed, dynamically activating and deactivating radio frequency links according to real-time channel conditions to adapt to channel changes. This invention, by introducing water-filled driving effective degrees of freedom and proposing a complete dynamic radio frequency allocation and precoding optimization process, effectively improves the energy efficiency of ultra-large-scale MIMO systems, providing an innovative solution for promoting the development of green communication in ultra-large-scale MIMO.

[0188] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0189] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0190] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0191] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for dynamic radio frequency allocation in ultra-large-scale MIMO, characterized in that, include: Step S1: For ultra-large-scale MIMO systems, define the effective degrees of freedom for channel water-filling drive; Step S2: Design a dynamic hybrid precoding architecture; Step S3: Under the dynamic hybrid precoding architecture, given the constraint on the number of RF chain activations, perform alternating iterative optimization of analog precoding, digital precoding, and power allocation in the dynamic hybrid precoding architecture; Step S4: Based on the channel's water-filled driving effective degrees of freedom, design the outer loop of the dynamic radio frequency allocation method. The outer loop of the dynamic radio frequency allocation method refers to achieving the optimal match between the number of active radio frequency chains and channel performance through closed-loop iteration.

2. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 1, characterized in that, The outer loop of the dynamic radio frequency allocation method includes: The first step is to dynamically adjust the number of active radio frequency links based on the effective degrees of freedom driven by the water-filling of the current equivalent channel. The second step is to use the alternating optimization method in step S3 to solve for the optimal analog precoding, digital precoding and power allocation corresponding to the adjusted number of RF chain activations. The third step is to update the equivalent channel based on the optimal analog precoding and recalculate the water-filled drive effective degrees of freedom of the updated equivalent channel. If the water-filled drive effective degrees of freedom are not equal to the current number of active RF links, return to the first step and repeat the iteration. If the water-filled drive effective degrees of freedom are equal to the current number of active RF links, output the final optimal number of active RF links, as well as the optimal analog precoding, digital precoding and power allocation corresponding to the optimal number of active RF links.

3. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 2, characterized in that, An equivalent channel refers to the superposition of a real channel, analog precoding, and digital precoding.

4. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 1, characterized in that, The gradient ascent algorithm is used to optimize the analog precoding, and the principles of eigenvalue decomposition and water injection algorithm are used to jointly optimize the digital precoding and power allocation.

5. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 1, characterized in that, Step 1 includes: Step S101: Consider a single-user near-field ultra-large-scale MIMO communication system, wherein the transmitter is configured with a... A uniform planar array of antennas, directed towards a receiver including... A uniform planar array of antennas transmits signals; the received signal vector at the receiver... It is given by the following formula: ， ； in, , Representing the The received signal from each receiving antenna, superscript Indicates transpose. , express A complex matrix of dimension 1; The transmitted signal vector, Representing the The transmitted signal from each transmitting antenna, ,vector This represents additive white Gaussian noise, whose elements have a mean of 0 and a variance of . complex Gaussian distribution ; The channel matrix based on the Green's function is expressed as follows: ， ； in, Representing the The transmitting antenna to the first The channel coefficient between the receiving antennas is defined as follows: ， ； in, and They represent the first in the transmission array. The first antenna and receiver array The spatial location of each antenna k is the wave number. Wavelength; Step S102: Based on the ultra-large-scale MIMO communication system, the effective degrees of freedom for water-filled drive are defined as follows: a) for the channel matrix a) Perform singular value decomposition to obtain its singular values; b) Given the total transmit power, use the water-filling algorithm based on the singular values ​​to obtain the optimal power allocation scheme; c) Count the number of streams with non-zero allocated power, and this number of streams is the effective degree of freedom of water-filling drive.

6. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 5, characterized in that, Step S2 includes: Step S201: Considering that the ultra-large-scale MIMO communication system adopts a dynamic partial connection hybrid precoding scheme, based on step S101, the signal vector is transmitted. Represented as ， ； in, Indicates length is Unit data stream, power allocation matrix , Representative assigned to the first The power of each data stream satisfies the constraint. , , This represents the upper limit of the system's total transmit power. Indicates the number of active RF links, let... ,matrix It is a digital precoding matrix. To simulate the precoding matrix; Step S202 Employing a partially connected architecture, each antenna is connected to a radio frequency (RF) link via a single-pole multiple-throw (SPMD) switch. The SPMD switch enables antenna-RF link mapping, allowing real-time adjustment of the active RF link. Assuming the active RF link is... One radio frequency link, and will Each antenna is evenly distributed among the active RF links, and the antenna indices are consecutive; at this time, Represented as: ， ； No. non-zero column vectors , , represented as: ； in, For mold taking operation, Indicates the first RF link to the first Gain coefficient of each transmitting antenna, , and They respectively represent connections to the first The start and end antenna indices of each radio frequency link. The power normalization factor, ； in, This indicates the number of antenna bases allocated to each radio frequency link. , Indicates the number of remaining antennas, such that the first An additional antenna is allocated to each radio frequency link.

7. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 6, characterized in that, Step S3 specifically includes the following steps: Step S301, according to the formula and formula The received signal at the receiving end is represented as: ； The corresponding spectral efficiency is: ， ； in, Represents the system's spectral efficiency. represent 1D identity matrix Represents noise power, superscript It is the conjugate transpose; At a given number of RF chain activations The following method employs an alternating optimization scheme to alternately optimize the analog precoding matrix. Digital precoding matrix and power allocation matrix Each iteration of the alternating optimization scheme consists of two parts: first, the optimization of the analog precoding matrix; and second, the joint optimization of the digital precoding matrix design and the power allocation matrix, until the convergence condition is met. Step S302, Analog precoding matrix optimization: Fix the digital precoding matrix from the previous iteration or initial setting. With power allocation matrix Unchanged; subsequently, to maximize system spectral efficiency To achieve the objective, the gradient ascent algorithm is used to optimize the simulated precoding matrix; the mathematical problem of simulated precoding matrix optimization is described as follows: ； in, For system spectral efficiency, the constraints require... Specific elements in the equation satisfy the formula The constant modulus constraint given in the document; Gradient Ascent Algorithm Initialization: Input Channel Matrix Initial analog precoding matrix Digital precoding matrix Power allocation matrix and noise power Set temporary variables Used to store the latest analog precoding matrix during the iteration process, where The operation means assigning the value of the variable on the right to the value on the left. The initial simulation precoding matrix is ​​generated randomly; Begin iterative loops of the gradient ascent algorithm until the convergence condition is met; at the start of each iteration, Assigned to the analog precoding matrix ,Right now ; Calculate spectral efficiency about Euclidean gradient : ， ； in, Representative calculation about Euclidean gradient, auxiliary matrix Defined as ; Iterate through and process each non-zero element in the analog precoding matrix. Each non-zero element that needs optimization , corresponding to The Line 1 Column elements; Calculate the Riemann gradient, and project the calculated Euclidean gradient onto the tangent space of the constant modulus constrained manifold to obtain the corresponding Riemann gradient: ； ； in, Representative calculation about The Riemann gradient, Representing the first of the matrix Line 1 Column elements, Represents the operation of taking the real part, superscript To obtain conjugate; Update element values ​​along the Riemann gradient direction with a preset step size parameter. Update the current element value: ， ； in, The element values ​​are updated along the Riemann gradient direction. The preset step size parameter controls the magnitude of each update; The manifold shrinking operation reprojects the updated element values ​​back onto the constant modulus constrained manifold, ensuring that the modulus constraint is satisfied. ； ； in, Represents the updated element value. Represents the phase taking a complex number. The base is the natural number; Then, after updating all elements in the current gradient ascent algorithm iteration, we obtain (14) ; Check the convergence condition of the gradient ascent algorithm and calculate... Corresponding spectral efficiency value And compared with the spectral efficiency value of the previous iteration Comparison, if the system's spectral efficiency The gain change is less than a preset minimum positive threshold. If the preset maximum number of iterations is reached, the gradient ascent algorithm is considered to have converged, and the iteration loop of the gradient ascent algorithm is exited. The value is assigned to the analog precoding matrix As output, i.e. Otherwise, repeat the gradient ascent algorithm iteration for the next round. Step S303, Joint optimization of digital precoding matrix design and power allocation matrix: Update the analog precoding matrix in step S302. After that, fix The current latest value is used; subsequently, the digital precoding matrix is ​​generated using the principles of eigenvalue decomposition and water-filling algorithms. and power allocation matrix Joint optimization design; Received channel matrix Noise power and the system's total transmit power constraint Analog precoding matrix ; Perform eigenvalue decomposition on the equivalent channel matrix. Perform eigenvalue decomposition: ， ； in, It is a unitary matrix, and its column vectors are eigenvectors; For a diagonal matrix, its diagonal elements The eigenvalues ​​are sorted in descending order; then... Assign to ,Right now This yields the digital precoding matrix output. Optimize the power allocation matrix based on the water injection algorithm principle; initialize parameters and set the iteration counter. ; Calculate the water injection level, for the first The next iteration calculates the current water injection level. : ； ； For the first The number of sub-channels considered in each iteration, where the first iteration considers all Sub-channels, i.e. ; It is the first eigenvalues ​​arranged in descending order; Simultaneously, based on the current water injection level, the test power value for each sub-channel is calculated. : ； ； Check if the minimum test power satisfies the positive constraint. ,like Then set Remove the first active sub-channel from the currently active sub-channel set. Sub-channels, update , Return to formula (16) for the next iteration; if If so, then perform the following operations; remember The optimal power allocation scheme is as follows: ， ； in The final water level for the iteration. The final number of activated sub-channels, i.e. The number of sub-channels; at this point, the power allocation matrix is ​​set as follows: ; for The values ​​of the elements on the diagonal; Step S304, completing steps S302 and S303 constitutes a complete iteration of the alternating optimization algorithm; the updated digital precoding matrix from step S303 is then... and power allocation matrix As the initial value for the next iteration, steps S302 and S303 are repeated; this alternating optimization process continues until the system's spectral efficiency is achieved. The gain change is less than a preset minimum positive threshold. If the preset maximum number of iterations is reached, the alternating optimization algorithm will declare convergence and terminate the iteration. Finally, the alternating optimization algorithm outputs the converged simulated precoding matrix. Digital precoding matrix and power allocation matrix Together they constitute the given A hybrid precoder that maximizes the system's spectral efficiency.

8. The method for dynamic radio frequency allocation in ultra-large-scale MIMO according to claim 7, characterized in that, Step S4 includes: Step S401: Based on the effective degrees of freedom of water-filled drive defined in step S1 and the alternating optimization algorithm proposed in step S3, design the outer loop of the dynamic radio frequency allocation scheme. Step S402, system initialization, according to Calculate its effective degrees of freedom driven by water filling. The effective degree of freedom value driven by the water filling is used as the initial value for the number of activated radio frequency chains: ； ； Simultaneously set , as well as ; Step S403: Begin the adaptive adjustment loop. In each iteration, first execute the alternating optimization algorithm to update the corresponding analog precoding matrix. Digital precoding matrix and power allocation matrix Then utilize the current analog precoding matrix and digital precoding matrix , and the channel matrix Multiply to construct the equivalent channel matrix ;; ， ； This equivalent channel matrix reflects the end-to-end channel characteristics after processing by the current precoder; then, the water-filled effective degrees of freedom of the equivalent channel matrix are calculated. According to the definition of effective degrees of freedom for water filling, we have ， ； Step S404: Determine if the RF chain configuration is optimal by comparing the current configuration. With equivalent channel effective degrees of freedom ,like Adjust the current RF chain configuration to At the same time, return to step S403; like This indicates that the current number of RF chains matches the number of data streams supported by the channel, and the configuration has reached its optimal state. The loop exits, and the current value is returned as the final optimization result: the optimal number of RF chains. Optimal analog precoding matrix Optimal digital precoding matrix Optimal power allocation matrix .

Images