Multi-point cooperation simulation beam synthesis and AI auxiliary acceleration calculation method thereof

Abstract

The invention discloses a multi-point cooperation simulation beam synthesis and AI auxiliary acceleration calculation method thereof, and belongs to the field of cellular-free MIMO wireless communication. According to the method, user angle estimation and interference angle estimation are carried out by utilizing the angle estimation capability of the system, and an uplink communication beam comprehensive problem model of the multi-point cooperation cellular-free MIMO system is constructed; a two-stage solving method is provided based on an ADMM framework, analog beam forming can be independently solved at all distributed APs through the depth expansion technology, digital beam forming solving can be carried out at a CPU after analog beam forming is determined, and final beam forming is obtained; and simulating the uplink communication beam comprehensive problem model based on optimal beam forming, and calculating and recording performance indexes. Through the intelligent method based on the model, the simulation beam comprehensive effect of the multi-point cooperation cellular-free MIMO system is improved, and the method has the advantages of being high in real-time performance, low in link overhead, low in calculation complexity and the like.

Description

Technical Field

[0001] The present invention relates to the field of non-cellular MIMO wireless communication technology, and in particular to a multi-point collaborative simulated beam synthesis and an AI-assisted accelerated calculation method thereof. Background Art

[0002] In recent years, non-cellular systems have attracted considerable attention due to their advantages in transmission reliability and speed. However, interference mitigation remains a significant challenge for these systems. Especially in the millimeter wave frequency band, signal propagation quality depends on the signal propagation conditions along the LoS path. Therefore, maintaining the signal quality along the LoS path ensures stronger interference mitigation performance. However, an unavoidable issue is that in the presence of interference, it is impossible to accurately estimate the target and interference angles, thus affecting the performance of subsequent signal processing algorithms. Therefore, subsequent beamforming designs require more targeted design to combat the misalignment of the beam direction caused by angle estimation errors.

[0003] Drawing on the theory of beam pattern synthesis in radar, wide main lobes and wide nulls can also be designed in the beamforming design of millimeter wave communications, making the beam more robust to angular errors. In actual multi-point coordinated multi-point coordinated cell-free MIMO (Multiple-Input, Multiple-Output) systems, in order to distinguish long-distance targets from interference, APs (Access Points) can be placed according to a certain pattern to form a virtual large MIMO array. In addition, through the hybrid architecture of the millimeter wave MIMO system, the beam can achieve performance while avoiding excessive fronthaul and hardware overhead. Therefore, considering the characteristics of the actual cell-free MIMO system and the propagation characteristics of the millimeter wave frequency band, a targeted and efficient beamforming design is needed to meet the corresponding wide beam requirements and have low computational complexity. Summary of the Invention

[0004] The present invention provides a multi-point coordinated simulated beam synthesis and its AI-assisted accelerated calculation method to achieve efficient and real-time anti-interference effect of a multi-point coordinated multi-point coordinated cellular-free MIMO system.

[0005] An embodiment of the present invention provides a multi-point coordinated simulated beam synthesis and AI-assisted accelerated computing method thereof, comprising the following steps:

[0006] S1, using the angle estimation capability of the multi-point coordinated cellular-free MIMO system to perform user angle estimation and interference angle estimation, and constructing an uplink communication beam synthesis problem model for the multi-point coordinated cellular-free MIMO system;

[0007] S2, performing a two-stage solution to the uplink communication beam synthesis problem model based on the ADMM (cross-directional method of multipliers) framework. First, analog beamforming is solved at each distributed AP using a deep expansion technique. After the analog beamforming is determined, digital beamforming is solved at the CPU to obtain the optimal beamforming.

[0008] S3, simulating the uplink communication beam synthesis problem model based on the optimal beamforming, and calculating and recording performance indicators.

[0009] Optionally, in one embodiment of the present invention, step S1 specifically includes:

[0010] S11, receiving user angle estimation information without a cellular uplink and receiving interference angle estimation information based on a millimeter wave LoS single-path communication link between the user, the AP, and the jammer;

[0011] S12, constructing an uplink communication beam synthesis problem model of a multi-point coordinated non-cellular MIMO system according to the received user angle estimation information and interference angle estimation information.

[0012] Optionally, in one embodiment of the present invention, step S2 specifically includes:

[0013] S21, based on the characteristics of multi-point coordinated non-cellular MIMO systems, separates analog beamforming and digital beamforming optimization into a two-stage optimization problem, solving analog beamforming at each AP and digital beamforming at the CPU.

[0014] S22, obtain the respective optimization sub-problems at each AP, solve them using the ADMM framework, and perform simulated beamforming solutions at the AP using the deep unfolding technique;

[0015] In step S23, the AP transmits the analog beamforming back to the CPU, where a digital beamforming optimization subproblem is constructed and the digital beamforming is solved using the ADMM framework.

[0016] Optionally, in one embodiment of the present invention, step S3 specifically includes:

[0017] S31, determining transmission parameters of a coordinated multi-point non-cellular MIMO system according to the optimal beamforming, including analog beamforming and digital beamforming at a receiving end and minimum mainlobe jitter;

[0018] S32, inputting the transmission parameters into the uplink communication beam synthesis problem model, and simulating and calculating the beam pattern and gain in the multi-point coordinated non-cellular MIMO system, including the beam pattern of a single AP and the beam pattern of multiple APs;

[0019] S33, draw a chart based on the performance indicators to perform performance analysis.

[0020] Optionally, in one embodiment of the present invention, in step S1, the uplink communication beam synthesis problem model takes minimizing the main lobe jitter of the beam as the optimization goal and satisfies the main lobe width, null width and depth, side lobe level and analog beamforming unit modulus value constraints. The uplink communication beam synthesis problem model is:

[0021]

[0022] Where ε is the main lobe jitter, w BB For digital beamforming, To simulate beamforming, a(θ m ) is the array guidance vector, θ m is the main lobe discrete angle parameter calculated based on the user angle estimation information, θ n is the discrete angle parameter of the null sink calculated based on the interference angle estimation information, η Z represents the null depth limit, θ s is the sidelobe discrete angle parameter, η SL Indicates the sidelobe level limit, are the discrete angle sets of the main lobe, null, and side lobe, respectively. for

[0023] A collection of elements that simulates a beam.

[0024] Optionally, in one embodiment of the present invention, in step S22, with the goal of maximizing the simulated beam gain, an optimization sub-problem is established at each AP, specifically:

[0025]

[0026] Among them, ε l is the beam gain, and α is the given mainlobe jitter reference range.

[0027] Optionally, in one embodiment of the present invention, in step S22, the ADMM framework is used for solving, and the depth expansion technology is used to perform analog beamforming solution at the AP, including:

[0028] S221, establish an optimization problem solving framework based on the ADMM framework, first write the augmented Lagrangian function of the optimization problem and determine the constraints;

[0029] S222, performing alternate solutions for different variable groups, and stopping iteration when the target value converges or the maximum number of iterations is reached, thereby obtaining the optimal simulated beamforming;

[0030] S223, the least squares problem of analog beamforming is solved with the aid of deep expansion technology to obtain the optimal step size of the Riemann gradient descent method and accelerate the convergence speed.

[0031] Optionally, in one embodiment of the present invention, in step S23, the CPU constructs a digital beamforming optimization sub-problem in the same form as the original optimization problem, wherein the analog beamforming is known.

[0032] The multi-point collaborative simulated beam synthesis and its AI-assisted accelerated calculation method in the embodiments of the present invention comprehensively consider the anti-interference transmission problem of the non-cellular millimeter wave MIMO system in the scenario where interference exists, increase the system's ability to resist interference, and ensure the low overhead requirements required for calculation while ensuring anti-interference transmission, meeting the needs of anti-interference transmission in complex interference environments.

[0033] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0035] Figure 1 This is a flowchart of a multi-point coordinated simulated beam synthesis and AI-assisted accelerated computing method thereof according to an embodiment of the present invention;

[0036] Figure 2 This is a flowchart of step S1 of an embodiment of the present invention;

[0037] Figure 3 This is a flowchart of step S2 of an embodiment of the present invention;

[0038] Figure 4 This is a flowchart of step S3 of an embodiment of the present invention;

[0039] Figure 5 This is a flow chart of an algorithm for designing efficient beamforming for multi-point coordinated cellular-free MIMO based on deep expansion according to an embodiment of the present invention;

[0040] Figure 6 Schematic diagram of the architecture of a hyperparameter network according to an embodiment of the present invention;

[0041] Figure 7 A schematic diagram of a deep expansion architecture according to an embodiment of the present invention;

[0042] Figure 8 Schematic diagram of the step size determination process of the traditional Riemann gradient method according to an embodiment of the present invention;

[0043] Figure 9 This is a beam gain diagram of a single AP simulated beam according to an embodiment of the present invention;

[0044] Figure 10 This is a comparison diagram of the gains of a single-AP simulated beam and a multi-AP hybrid beam according to an embodiment of the present invention;

[0045] Figure 11 This is a comparison chart of the time required to solve beamforming using the traditional Riemannian gradient descent method and the deep unfolding technology in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0047] Figure 1 This is a flowchart of a multi-point collaborative simulation beam synthesis and AI-assisted accelerated computing method provided according to an embodiment of the present invention.

[0048] like Figure 1 As shown, the multi-point collaborative simulation beam synthesis and its AI-assisted accelerated calculation method include the following steps:

[0049] S1, using the angle estimation capability of the multi-point coordinated cell-free MIMO system to perform user angle estimation and interference angle estimation, constructing an uplink communication beam synthesis problem model for the multi-point coordinated cell-free MIMO system.

[0050] In step S1, the uplink communication beam synthesis problem model takes minimizing the main lobe jitter of the beam as the optimization goal and satisfies the main lobe width, null width and depth, side lobe level and analog beamforming unit modulus value constraints. The uplink communication beam synthesis problem model is:

[0051]

[0052] Where ε is the main lobe jitter, w BB For digital beamforming, To simulate beamforming, a(θ m ) is the array guidance vector, θ m is the main lobe discrete angle parameter calculated based on the user angle estimation information, θ n is the discrete angle parameter of the null sink calculated based on the interference angle estimation information, η Z represents the null depth limit, θ s is the sidelobe discrete angle parameter, η SL Indicates the sidelobe level limit, are the discrete angle sets of the main lobe, null, and side lobe, respectively. is a set of elements that simulate the beam. Based on the user angle estimation information and the received interference angle estimation information, an angle interval is taken within a certain range and discretely sampled to obtain the corresponding discrete angle parameter.

[0053] The present invention proposes a multi-point coordinated simulated beam synthesis method and its AI-assisted accelerated calculation method, establishing an optimization model. The goal is to minimize the mainlobe jitter of the beam while satisfying the constraints of the mainlobe width, null width and depth, sidelobe level, and analog beamforming unit modulus.

[0054] The goal of minimizing mainlobe jitter is to ensure excellent signal quality in the mainlobe direction of the received beam. By designing a well-designed beamforming scheme, the system's overall interference resistance is optimized, thereby improving communication quality and reliability during uplink transmission. By adjusting the null width and depth, the system can eliminate the impact of interfering signals on the communication system, enabling it to tolerate a certain level of interference power.

[0055] In an embodiment of the present invention, Figure 2 As shown, step S1 specifically includes:

[0056] S11, based on the millimeter wave LoS single-path communication link between the user, AP and jammer, receives user angle estimation information without cellular uplink and receives interference angle estimation information, and both angle information have a certain estimation error;

[0057] S12, constructing an uplink communication beam synthesis problem model of a multi-point coordinated non-cellular MIMO system according to the received user angle estimation information and interference angle estimation information, including optimization of analog beamforming and digital beamforming.

[0058] A non-cellular millimeter-wave MIMO system consists of several multi-antenna APs and several multi-antenna users, with a limited number of multi-antenna jammers interfering with communications. Each AP is connected to a central processing unit (CPU) via a fronthaul link for coordinated transmission on behalf of the users, and all APs share system information. During uplink transmission, each user communicates on a different frequency band, eliminating inter-user interference. Furthermore, due to the presence of external jammers, each user receives interference signals from the jammers, further impacting normal communication quality.

[0059] In an embodiment of the present invention, step S12 is specifically as follows:

[0060] S12a: Modeling the antenna array steering vector in a non-cellular millimeter wave LoS scenario, and preparing relevant model data for the angles of the user and the jammer;

[0061] S12b, based on the array guidance model, uses beam synthesis theory to construct a related form of beamforming optimization problem, in which there is a fusion of two layers of beamforming: analog beamforming and digital beamforming.

[0062] By establishing a problem model based on non-cellular millimeter-wave hybrid beamforming, the real physical model corresponding to the problem architecture can be effectively obtained, which is helpful for designing the anti-interference communication strategy of multi-point coordinated non-cellular MIMO system.

[0063] S2, based on the ADMM framework, performs a two-stage solution to the uplink communication beam synthesis problem model. First, deep expansion technology is used to solve analog beamforming at each distributed AP. After determining the analog beamforming, digital beamforming is solved at the CPU to obtain the optimal beamforming.

[0064] In an embodiment of the present invention, Figure 3 As shown, step S2 specifically includes:

[0065] S21, based on the characteristics of non-cellular coordinated multi-point systems, separates analog beamforming and digital beamforming optimization into a two-stage optimization problem. The analog beamforming problem is solved at each AP, while the digital beamforming problem is solved at the CPU.

[0066] In step S22, each AP obtains its own optimization subproblem and solves it using the ADMM framework. To reduce the computational complexity of the analog beamforming solution, deep unfolding technology is introduced to perform analog beamforming solutions at the AP.

[0067] In step S23, the AP transmits the analog beamforming back to the CPU, where a digital beamforming optimization subproblem is constructed and the digital beamforming is solved using the ADMM framework.

[0068] In step S21, based on the model of the uplink communication beam synthesis problem in a multi-point coordinated, non-cellular MIMO system, the problem is decomposed into two stages: the first stage solves the analog beam at the AP, and the second stage solves the digital beam at the CPU. In the first stage, the unit modulus constraint of the analog beam is taken into account, and the subproblem at the AP becomes maximizing the gain of the analog beam. In the second stage, the subproblem at the CPU remains in its original form, but the analog beam is known.

[0069] To better utilize the distributed collaboration capabilities of multiple APs while reducing fronthaul link overhead, we decompose the analog and digital beam optimization problems and establish optimization subproblems at each AP with the goal of maximizing beam gain:

[0070]

[0071] Among them, ε l is the beam gain, and α is the given mainlobe jitter reference range.

[0072] Considering the unit modulus constraint of the analog beam, the subproblem objective at each AP is to maximize the beam gain. Where α is the given main lobe jitter reference range, ε l is the beam gain. After solving the analog beam at each AP, the AP sends the analog beam to the CPU via the fronthaul link, and the CPU solves the digital beam. The problem is:

[0073]

[0074] Among them, W RF This two-stage solution fully utilizes the distributed processing capabilities of the cellular system and can effectively reduce the CPU computation overhead.

[0075] In step S22, the ADMM framework is used to solve the problem, and the depth expansion technology is used to perform simulated beamforming solution at the AP, including:

[0076] S221, establish an optimization problem solving framework based on the ADMM framework, first write the augmented Lagrangian function of the optimization problem and determine the constraints;

[0077] S222, performing alternate solutions for different variable groups, and stopping iteration when the target value converges or the maximum number of iterations is reached, thereby obtaining the optimal simulated beamforming;

[0078] S223, the least squares problem of analog beamforming is solved with the aid of deep expansion technology to obtain the optimal step size of the Riemann gradient descent method and accelerate the convergence speed.

[0079] S3, simulates the optimal beamforming and uplink communication beam synthesis problem model, calculates and records the performance indicators.

[0080] In an embodiment of the present invention, Figure 4 As shown, step S3 specifically includes:

[0081] S31, determining transmission parameters of the multi-point coordinated non-cellular MIMO system based on the optimal beamforming, including analog beamforming and digital beamforming at the receiving end and minimum main lobe jitter;

[0082] S32, inputting the transmission parameters into the uplink communication beam synthesis problem model, and simulating and calculating the beam pattern and gain in the multi-point coordinated non-cellular MIMO system, including the beam pattern of a single AP and the beam pattern of multiple APs;

[0083] S33, draws charts based on performance indicators, analyzes computational complexity, and performs performance analysis on a multi-point collaborative simulated beam synthesis and its AI-assisted accelerated computing method.

[0084] In a further embodiment, based on known system deployments, a model for the uplink communication beam synthesis problem in a multi-point coordinated cell-free MIMO system is constructed, taking into account the mainlobe width, null width and depth, sidelobe level, and analog beamforming unit modulus constraints. Based on the distributed processing characteristics of cell-free systems and the features of analog and digital beams, the uplink communication beam synthesis problem is decomposed into a two-stage solution: solving the analog beam at the AP and incorporating deep expansion techniques, and solving the digital beam at the CPU, ultimately achieving the optimal solution to the original problem. Simulation results demonstrate that the proposed method can rationally design system beamforming with relatively low computational complexity, achieving reliable anti-interference performance.

[0085] In an embodiment of the present invention, during the modeling of the multi-point collaborative simulation beam synthesis problem, a model of a single array steering vector is constructed, and taking into account the characteristics of a non-cellular system, an array steering vector model of multiple arrays is constructed, and the problem is modeled based on the structure of hybrid beamforming. At the same time, based on the requirements and drawing on the beam synthesis theory, a constraint with a wide main lobe, a wide null, and a low sidelobe level is constructed to minimize the main lobe level jitter.

[0086] In an embodiment of the present invention, the performance analysis of multi-point coordinated simulated beam synthesis and its AI-assisted accelerated computing method includes:

[0087] Analyze the beam pattern and gain of a single AP beamforming.

[0088] Analyze the beam patterns and gains of hybrid beamforming for multiple APs.

[0089] Analyze the computational complexity and real-time indicators of deep expansion technology and traditional methods.

[0090] In an embodiment of the present invention, a multi-point coordinated simulated beam synthesis and AI-assisted accelerated computing method thereof include:

[0091] Define variables and parameters in a multi-point coordinated cell-free MIMO system. Variables include receive analog and digital beamforming and mainlobe jitter levels. Parameters include mainlobe width, null width and depth, sidelobe levels, and the number and shape of antennas per AP array.

[0092] Based on the variables and parameters, as well as the performance indicators of the multi-point cooperative MIMO system without a cell, the objective function for beamforming in the multi-point cooperative MIMO system without a cell is constructed to minimize the mainlobe level jitter. The constraints of the multi-point cooperative MIMO system without a cell are defined, including the upper and lower limits of the mainlobe jitter, the null depth and width limits, the sidelobe level constraints, and the constant mode constraints of the analog beamforming.

[0093] According to the objective function and constraints, a multi-point collaborative simulation beam synthesis and its AI-assisted accelerated calculation method are constructed to obtain a non-convex optimization problem.

[0094] A model of the uplink communication beam synthesis problem of a multi-point cooperative non-cellular MIMO system is constructed, and it is decomposed into a two-stage optimization problem based on analog beams and digital beams. The ADMM framework is used to decompose the non-convex optimization problem into sub-problems including analog and digital beamforming at the receiving end. The augmented Lagrange multiplier method is combined with the penalty function method to solve the two sub-problems separately. The deep expansion technology is applied when solving the analog beam to obtain the optimal solution of the analog beam and accelerate the convergence speed. The analog beam is transmitted back to the CPU and the digital beam is solved to obtain all the optimal solutions of the non-convex optimization problem.

[0095] In an embodiment of the present invention, the simulation parameter setting of the multi-point coordinated simulated beam synthesis and its AI-assisted accelerated computing method includes:

[0096] According to the performance indicators of the multi-point cooperative MIMO system without cell, set the simulation parameters of the multi-point cooperative MIMO system without cell, including the number, location and effective area of APs, users and jammers;

[0097] According to the beamforming algorithm of the multi-point cooperative MIMO system without cell, set the optimization parameters of the multi-point cooperative MIMO system without cell, including the objective function, constraints, basic framework of the optimization algorithm, convergence conditions, etc.

[0098] According to the application scenario of the multi-point coordinated non-cellular MIMO system beamforming algorithm, set the application parameters of the multi-point coordinated non-cellular MIMO system transmission, including the application type.

[0099] In an embodiment of the present invention, the analysis of simulation results of multi-point coordinated simulated beam synthesis and its AI-assisted accelerated computing method includes:

[0100] Run the anti-interference beamforming design simulation algorithm for the multi-point cooperative MIMO system without cell, and record the simulation results of the multi-point cooperative MIMO system without cell, including beam pattern and gain.

[0101] Furthermore, the specific algorithm flow of the design method for efficient beamforming of multi-point cooperative non-cellular MIMO based on deep expansion is as follows: Figure 5 As shown, specifically:

[0102] Step a: Decompose the original optimization problem into analog beam design on the AP side and digital beam design on the CPU side.

[0103] Step b: For the analog beam design on the AP side, solve the sub-problem:

[0104]

[0105] According to the ADMM framework, the problem is first transformed into:

[0106]

[0107] in, Write the augmented Lagrangian function:

[0108]

[0109] Solve subproblem 1 alternately:

[0110]

[0111] And sub-question 2:

[0112]

[0113] Then update the penalty parameter:

[0114]

[0115] The three problems are solved alternately until the target value converges. When solving sub-problem 1, a classic method is the Riemann gradient descent method. In order to accelerate the algorithm operation by using AI deep expansion technology, a hyperparameter network is used to optimize the iterative step size, thereby accelerating the convergence speed of the traditional method and reducing the corresponding computational complexity to achieve high efficiency and real-time performance. For the AI-assisted part, a basic schematic diagram of the hyperparameter neural network structure in the deep expansion of this algorithm is given, as shown in the figure. Figure 6 As shown in Figure 1. The entire network consists of 5 complex linear layers, and the output activation function of the first to fourth layers is the CReLU function. The output of the last layer is the sum of the real and imaginary parts, and in order to ensure the non-negativity of the output step size parameter, the activation function of the last layer is set to the absolute value Abs function. The input of the entire network is the least squares solution to the problem, the corresponding Euclidean gradient and the corresponding Riemann gradient, and the output is the step size parameter in all iterative processes. The expression of CReLU is:

[0116] CReLU(z)=max{0,real(z)}+1i*max{0,imag(z)}

[0117] Figure 7 and Figure 8 The block diagrams of the Riemannian gradient descent method assisted by AI deep expansion and the traditional Riemannian gradient descent method are shown. The difference between the two is that the hyperparameter network of AI deep expansion can predict the step size parameters for all iterative steps at once, while the traditional step size determination method based on the Armijo backtracking line search method requires a large amount of calculation and judgment for each iteration. The computational complexity of this step is difficult to accurately measure, making it unacceptable in practical applications. The method based on AI deep expansion can accelerate the CPU parallel calculation of the neural network in actual hardware calculations, thus having a considerable advantage in reducing the algorithm's computational complexity.

[0118] Step c: For digital beam design on the CPU side, solve the subproblem:

[0119]

[0120] According to the ADMM framework, the problem is transformed into:

[0121]

[0122] in, a cpu (θ i )=W RF,H a(θ i ). Write the augmented Lagrangian function:

[0123]

[0124] Solve subproblem 3 alternately:

[0125]

[0126] And sub-question 4:

[0127]

[0128] st

[0129]

[0130] Then update the penalty parameter:

[0131]

[0132] Until the target value converges. At this point, all the variables to be optimized have been obtained.

[0133] In order to verify the anti-interference transmission performance of the present invention, the following simulation experiments were carried out:

[0134] Five users, 10 APs, and two jammers are deployed within a large cube. The APs are evenly distributed along one edge of the cube's upper base, with fixed and close spacing between them. The users and jammers are evenly distributed within the space, with users concentrated but at a distance from the APs. This ensures that all AP arrays meet far-field conditions. Each AP is equipped with a single radio chain phased array with 64 antennas, and both the users and jammers are equipped with a single antenna.

[0135] First, the single AP analog beamforming algorithm is simulated under single interference angle and single target angle. The simulation results are as follows: Figure 9 As shown in Figure 1, the horizontal axis represents the angle range, and the vertical axis represents the beam gain. It can be seen that the proposed beamforming design algorithm can achieve the requirements of wide mainlobe and wide null while ensuring the level of sidelobe.

[0136] Then we compared the gain comparison diagram of single AP simulated beam and multi-AP beam. The simulation results are as follows: Figure 10 The horizontal axis represents the angular range, and the vertical axis represents the beam gain. It can be seen that under the same interference and target conditions, the hybrid beamforming effect of multiple APs significantly outperforms the simulated beamforming effect of a single AP. Furthermore, after implementing multi-point coordination, the jitter of the main lobe is significantly reduced.

[0137] In order to reflect the low complexity of the algorithm, the present invention gives the number of iterations and running time required for each of the algorithm after introducing the depth expansion technology and the traditional algorithm to solve the problem once, such as Figure 11 As shown in Figure 2, the number of iterations required by the depth expansion method is significantly less than that of the Riemann gradient descent method, which can effectively reduce the computational complexity.

[0138] According to the multi-point collaborative simulated beam synthesis and its AI-assisted accelerated calculation method proposed in the embodiment of the present invention, the system's angle estimation capability is used to perform user angle estimation and interference angle estimation, and an uplink communication beam synthesis problem model of the multi-point collaborative non-cellular MIMO system is constructed; based on the ADMM framework, a two-stage solution method is proposed, which can independently solve the simulated beamforming at each distributed AP using the deep expansion technology. After determining the simulated beamforming, the digital beamforming solution can be performed at the CPU to obtain the final beamforming; based on the optimal beamforming, the uplink communication beam synthesis problem model is simulated, and the performance indicators are calculated and recorded. The present invention improves the simulated beam synthesis effect of the multi-point collaborative non-cellular MIMO system through a model-based intelligent method, and has the advantages of high real-time performance, low link overhead, and low computational complexity.

[0139] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0140] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "N" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0141] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or N executable instructions for implementing a custom logical function or step of a process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

Claims

1. A multi-point collaborative simulated beam synthesis and AI-assisted accelerated computing method, characterized in that: The following steps are involved: S1, using the angle estimation capability of the multi-point coordinated cellular-free MIMO system to perform user angle estimation and interference angle estimation, and constructing an uplink communication beam synthesis problem model for the multi-point coordinated cellular-free MIMO system; S2, performing a two-stage solution to the uplink communication beam synthesis problem model based on the ADMM framework, first solving analog beamforming at each distributed AP using a deep unfolding technique, and then performing digital beamforming solution at the CPU after determining the analog beamforming to obtain the optimal beamforming; S3, simulating the uplink communication beam synthesis problem model based on the optimal beamforming, and calculating and recording performance indicators.

2. The method according to claim 1, characterized in that Step S1 specifically includes: S11, receiving user angle estimation information without a cellular uplink and receiving interference angle estimation information based on a millimeter wave LoS single-path communication link between the user, the AP, and the jammer; S12, constructing an uplink communication beam synthesis problem model of a multi-point coordinated non-cellular MIMO system according to the received user angle estimation information and interference angle estimation information.

3. The method according to claim 1, characterized in that Step S2 specifically includes: S21, based on the characteristics of multi-point coordinated non-cellular MIMO systems, separates analog beamforming and digital beamforming optimization into a two-stage optimization problem, solving analog beamforming at each AP and digital beamforming at the CPU. S22, obtain the respective optimization sub-problems at each AP, solve them using the ADMM framework, and perform simulated beamforming solutions at the AP using the deep unfolding technique; In step S23, the AP transmits the analog beamforming back to the CPU, where a digital beamforming optimization subproblem is constructed and the digital beamforming is solved using the ADMM framework.

4. The method according to claim 1, wherein Step S3 specifically includes: S31, determining transmission parameters of a coordinated multi-point non-cellular MIMO system according to the optimal beamforming, including analog beamforming and digital beamforming at a receiving end and minimum mainlobe jitter; S32, inputting the transmission parameters into the uplink communication beam synthesis problem model, and simulating and calculating the beam pattern and gain in the multi-point coordinated non-cellular MIMO system, including the beam pattern of a single AP and the beam pattern of multiple APs; S33, drawing a chart based on the performance indicators to perform performance analysis.

5. The method according to claim 1, wherein In step S1, the uplink communication beam synthesis problem model takes minimizing the main lobe jitter of the beam as the optimization goal and satisfies the main lobe width, null width and depth, side lobe level and analog beamforming unit modulus value constraints. The uplink communication beam synthesis problem model is: Where ε is the main lobe jitter, w BB For digital beamforming, To simulate beamforming, a(θ m ) is the array guidance vector, θ m is the main lobe discrete angle parameter calculated based on the user angle estimation information, θ n is the discrete angle parameter of the null sink calculated based on the interference angle estimation information, η Z represents the null depth limit, θ s is the sidelobe discrete angle parameter, η SL Indicates the sidelobe level limit, are the discrete angle sets of main lobe, null and side lobe, is the set of elements that simulate the beam. Based on the estimated angle, an angle interval is taken within a certain range to the left and right of it and discretely sampled to obtain the corresponding discrete angle parameter.

6. The method according to claim 5, characterized in that In step S22, with the goal of maximizing the simulated beam gain, an optimization sub-problem is established at each AP, specifically: Among them, ε l is the beam gain, and α is the given mainlobe jitter reference range.

7. The method according to claim 6, characterized in that In step S22, the ADMM framework is used to solve the problem, and the depth expansion technology is used to perform simulated beamforming solution at the AP, including: S221, establish an optimization problem solving framework based on the ADMM framework, first write the augmented Lagrangian function of the optimization problem and determine the constraints; S222, performing alternate solutions for different variable groups, and stopping iteration when the target value converges or the maximum number of iterations is reached, thereby obtaining the optimal simulated beamforming; S223, the least squares problem of analog beamforming is solved with the aid of deep expansion technology to obtain the optimal step size of the Riemann gradient descent method and accelerate the convergence speed.

8. The method according to claim 6, characterized in that In step S23, the CPU constructs a digital beamforming optimization sub-problem in the same form as the original optimization problem, wherein the analog beamforming is known.

Images