Radio frequency power control method and system of 5G mainboard

Abstract

The invention relates to the field of radio frequency power control, and discloses a radio frequency power control method and system for a 5G mainboard, and the method comprises the steps: 1, capturing the phase fluctuation of a millimeter wave frequency band through employing a superconductive quantum bit array, building a virtual channel response based on quantum entanglement state measurement, generating a channel topological graph, calculating and predicting a channel state based on the channel topological graph, and carrying out the calculation of a predicted channel state; calculating a prediction error according to the measured actual channel state, and adjusting the radio frequency power; 2, fusing the quantum measurement matrix and the deep residual network, introducing a residual attention mechanism to reconstruct an abnormal channel, carrying out local optimization, solving asymmetric Nash equilibrium, reducing algorithm complexity and calculation complexity, dynamically controlling radio frequency power, suppressing interference and improving power amplifier performance; and step 3, analyzing the influence of environmental factors on system errors, jointly calibrating and correcting the errors, constructing a learning model and expanding a dynamic range, determining the actual maximum transmitting power and distributing radio frequency power.

Description

Technical Field

[0001] The present invention relates to the field of radio frequency power control technology, and in particular to a radio frequency power control method and system for a 5G motherboard. Background Art

[0002] The RF power control technology of the 5G motherboard is one of the core technologies that supports the high-performance and low-power operation of the 5G network. To achieve higher data transmission rates than 4G, 5G requires more complex modulation and coding methods and higher transmission power to ensure signal quality. It can transmit more data within limited spectrum resources, not only to meet the communication needs of a large number of mobile phone users, but also to support large-scale connections of IoT devices. In order to accommodate more users and devices communicating simultaneously, power control is needed to optimize resource allocation, reduce interference between users, and increase system capacity. 5G introduces millimeter wave spectrum resources. High-frequency signals have large propagation losses, weak penetration capabilities, and are easily affected by obstacles. More sophisticated power control is required to compensate for path loss and shadow fading to ensure signal quality during transmission. 5G networks need to coexist with 2G, 3G, and 4G networks. 5G motherboards must support the transmission and reception of RF signals in multiple frequency bands. Signals in different frequency bands have different propagation characteristics and interference conditions, requiring RF power control to adapt to the complex environment of multiple frequency bands and achieve efficient operation in each band.

[0003] In existing technologies, in high-speed mobile scenarios, traditional closed-loop control algorithms are slow to respond and struggle to cope with rapidly changing channel conditions, resulting in a significant increase in bit error rates. In the dense multipath environment of the millimeter wave frequency band, the existing channel model is inaccurate, causing overcompensation in the power allocation algorithm, impacting communication quality. In large-scale MIMO systems, the power allocation algorithm has high computational complexity, with computational delays exceeding the frame period, failing to meet real-time requirements. The cross-cell interference management mechanism has a long coordination cycle and struggles to adapt to burst traffic, resulting in a significant drop in throughput for users at the cell edge. When broadband GaN power amplifiers operate in high frequency bands, traditional digital pre-distortion algorithms consume large resources, reducing the energy efficiency of the power amplifier. The dynamic range of existing power control architectures is limited, unable to simultaneously meet the requirements of high transmit power and low receive sensitivity. The training scenarios for machine learning-based power control systems are incomplete, and their ability to recognize burst interference patterns is poor. The modeling errors of the RF link digital twin are large, making simulation results difficult to guide practical applications. The system has weak suppression capabilities for out-of-band blocking interference and is susceptible to interference in specific frequency bands, leading to a sharp increase in bit error rates.

[0004] Based on this, it is necessary to provide a 5G motherboard radio frequency power control method and system to solve the interference problem of power control. Summary of the Invention

[0005] The object of the present invention is to provide a 5G motherboard radio frequency power control method and system. To solve the above-mentioned prior art problems, the present invention is implemented through the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a 5G motherboard radio frequency power control method, which specifically includes the following steps:

[0007] Step 1: Use a superconducting qubit array to capture millimeter-wave frequency phase fluctuations, construct a virtual channel response based on quantum entangled state measurements, and generate a channel topology map. Calculate and predict the channel state based on the channel topology map, calculate the prediction error based on the actual channel state, and adjust the RF power.

[0008] Step 2: Integrate the quantum measurement matrix with the deep residual network, introduce the residual attention mechanism to reconstruct abnormal channels, perform local optimization, solve the asymmetric Nash equilibrium, and reduce the algorithmic and computational complexity, dynamically control the RF power, suppress interference, and improve the performance of the power amplifier;

[0009] Step 3: Analyze the impact of environmental factors on system errors, perform joint calibration to correct errors, build a learning model and expand the dynamic range, determine the actual maximum transmit power, and allocate RF power.

[0010] Furthermore, the method for constructing the virtual channel response is:

[0011] Based on quantum entangled state measurement, through quantum measurement matrix Jointly construct virtual channel responses with deep generative adversarial networks (GANs);

[0012] The virtual channel response is compressed and observed through the quantum measurement matrix, and noise n is introduced. The compressed observation value is calculated, where and is the generator of GAN, z is the latent space noise vector;

[0013] Furthermore, the method for generating the channel topology map is:

[0014] Based on the generation of virtual channel responses, the channel multipath scatterers are mapped to dynamic graph nodes, and the edge weights are set as the scatterer mutual coupling coefficients , generate a channel topology map;

[0015] Furthermore, the method for calculating and predicting the channel state is:

[0016] Based on the generated channel topology graph, it is processed through the graph convolution network and combined with the changes in multipath fading and Doppler frequency shift, according to the formula Calculate the predicted channel state ,in, is the channel topology at time t, and the adjacency matrix dimension is N×N, are the ST-GCN network parameters, Indicates the online update cycle. represents the k-th path fading coefficient, It represents the Doppler shift of the kth path;

[0017] Furthermore, the method for calculating the prediction error is:

[0018] By inserting a known pilot signal at the signal transmitting end, the receiving end measures the actual channel state based on the minimum mean square error formula according to the received pilot signal and the transmitted pilot signal. ;

[0019] Based on the calculated predicted channel state , the actual channel state and the predicted channel state are calculated using the root mean square error formula to obtain the prediction error;

[0020] Furthermore, the method for reconstructing the abnormal channel is:

[0021] Based on the obtained quantum measurement matrix Fusion with the deep residual network ResNet;

[0022] By formula Fusion obtains mixed measurement matrix ,in, is the measurement weight;

[0023] Based on the obtained mixed measurement matrix, the residual attention mechanism is introduced to reconstruct the abnormal channel, and the main path component is reconstructed first. Get the sparse reconstruction estimate of the channel state, where It represents the predicted channel state attention weighted L1 norm, It means The square of the L2 norm, It represents the sparsity adjustment factor;

[0024] Furthermore, the local optimization method is:

[0025] The 256-antenna system of the adjacent cell is evenly divided into 32 dynamic hypergraph units. Based on the asymmetric Nash equilibrium optimization objective, the power is dynamically adjusted according to the interference between cells. Get the maximum transmission power of the dynamic hypergraph unit , the constraints are ,in, It represents the transmission power of the mth unit. It represents the signal-to-interference-and-noise ratio of the mth unit;

[0026] The gradient projection method is used to solve the asymmetric Nash equilibrium, and the gradient calculation is parallelized on the silicon-based photonic chip. Reduce the algorithm complexity, where represents the number of iterations, L is the Lagrangian function, is the step size factor, represents the gradient operator;

[0027] The polynomial order is compressed from 11 to 5 by the formula, where: It represents the optical frequency response correction matrix. It represents the predistortion coefficient;

[0028] The RF power is controlled by a dual-loop controller;

[0029] Furthermore, the method for extending the dynamic range is:

[0030] Based on the obtained bias vector of the embedded ambient temperature compensation coefficient , construct a 128×128 memristor array to store the nonlinear transfer function, and use the formula Get output voltage data ,in, It represents the input voltage data. represents the memristor conductance matrix;

[0031] Furthermore, the method for obtaining the actual maximum transmit power is:

[0032] Based on the obtained ambient temperature and signal operating frequency, a three-dimensional lookup table is constructed, and the formula Get the calibration error ,in, It represents the calibration factor;

[0033] The obtained calibration error is compared with the preset maximum calibration allowable error to obtain a calibration relative value;

[0034] Multiply the obtained calibration relative value by the maximum transmit power to obtain the transmit calibration power difference;

[0035] Perform subtraction processing on the maximum transmit power and the transmit calibration power difference to obtain the actual maximum transmit power;

[0036] Radio frequency power is allocated to channels within the dynamic hypergraph unit based on the obtained actual maximum transmit power.

[0037] In a second aspect, an embodiment of the present invention provides a 5G motherboard radio frequency power control system, which specifically includes the following modules:

[0038] Channel state perception module: This module uses a superconducting quantum bit array to capture millimeter-wave frequency phase fluctuations, constructs a virtual channel response based on quantum entanglement state measurements, and generates a channel topology map.

[0039] RF power adjustment module: This module calculates and predicts the channel state based on the channel topology map, calculates the prediction error based on the actual channel state, and adjusts the RF power.

[0040] Measurement matrix fusion module: It integrates the quantum measurement matrix with the deep residual network, introduces the residual attention mechanism to reconstruct abnormal channels, and performs local optimization;

[0041] Channel processing module: solves asymmetric Nash equilibrium and reduces algorithmic and computational complexity, dynamically controls RF power, suppresses interference, and improves power amplifier performance;

[0042] Environmental factor processing module: analyzes the impact of environmental factors on system errors and corrects errors through joint calibration;

[0043] Power allocation module: Builds a learning model and expands the dynamic range to determine the actual maximum transmit power and allocate RF power.

[0044] Beneficial effects of the present invention:

[0045] 1. Capturing millimeter-wave frequency band phase fluctuations through a superconducting quantum bit array and combining it with a graph convolutional network to calculate and predict channel states, tracking multipath fading and Doppler shift changes in the channel in real time, and adjusting RF power based on the error between the actual and predicted channel states to ensure communication links remain stable in complex and changing channel environments, reducing signal fading and packet loss rates. Using hypergraph partitioning and asymmetric Nash equilibrium optimization, inter-cell interference management and resource allocation are achieved, dynamically adjusting power based on interference, reducing the impact of inter-cell interference on communication quality, and improving spectrum utilization and system capacity.

[0046] 2. Integrate the quantum measurement matrix with the deep residual network to improve the compression performance of the measurement matrix, combine the residual attention mechanism to reconstruct abnormal channels, give priority to the main path component, reduce the number of channel reconstruction iterations, and increase the speed of channel reconstruction; reduce the computational complexity of traditional digital pre-distortion, significantly reduce the computing resources and time required for the system to process data, analyze the impact of ambient temperature and signal operating frequency on system errors, use joint calibration to correct errors in real time, and build a neural network learning model and memristor array to expand the system dynamic range, so that the system can adapt to different temperature environments and signal operating frequencies, and reduce the interference of environmental factors on system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1 This is a flowchart of the steps of a 5G motherboard radio frequency power control method provided in Example 1 of the present invention;

[0049] Figure 2 This is a structural diagram of a 5G motherboard radio frequency power control system provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0051] Example 1

[0052] like Figure 1 As shown, an embodiment of the present invention provides a 5G motherboard radio frequency power control method, which specifically includes the following steps:

[0053] Step 1: Use a superconducting qubit array to capture millimeter-wave frequency phase fluctuations, construct a virtual channel response based on quantum entangled state measurements, and generate a channel topology map. Calculate and predict the channel state based on the channel topology map, calculate the prediction error based on the actual channel state, and adjust the RF power.

[0054] In step 1, a superconducting quantum bit array is used to capture sub-nanosecond phase fluctuations in the millimeter-wave frequency band;

[0055] It should be noted that superconducting qubits use superconducting devices such as superconducting Josephson junctions to realize the functions of qubits. Superconducting qubits are integrated into arrays. By increasing the number of qubits, the spatial coverage and measurement accuracy of millimeter-wave signals are improved. Multiple qubits simultaneously measure signals at different spatial locations in the millimeter-wave frequency band and capture phase fluctuation information. The millimeter-wave frequency band is an important component of 5G communications, providing 5G networks with ultra-high-speed data transmission capabilities to meet the needs of application scenarios with extremely high bandwidth requirements.

[0056] Based on quantum entangled state measurement, through quantum measurement matrix Jointly construct virtual channel responses with deep generative adversarial networks (GANs);

[0057] Specifically, the virtual channel response is compressed and observed through the quantum measurement matrix, and noise n is introduced, and the formula The compressed observation value is calculated, where , is the generator of GAN, z is the potential space noise vector, and the virtual channel response is generated by GAN , combined with the quantum measurement matrix, to achieve efficient compressed sensing and virtual construction of the channel state;

[0058] Based on the generation of virtual channel response, the channel multipath scatterers are mapped to dynamic graph nodes, and the number of nodes N is less than or equal to 500, and the edge weight is set to the scatterer mutual coupling coefficient , generate channel topology graph; transform the complex channel multipath structure into a computable graph structure;

[0059] It should be noted that the edge weight is set to the scatterer mutual coupling coefficient to represent the correlation between signals propagating through different scatterer paths. The scatterer mutual coupling coefficient indicates the degree of signal coupling between two scatterers and is related to the changes in amplitude and phase of the signals reaching the receiving end from the two scatterers.

[0060] Based on the generated channel topology graph, it is processed through the graph convolution network and combined with the changes in multipath fading and Doppler frequency shift, according to the formula Calculate the predicted channel state ,in, is the channel topology at time t, and the adjacency matrix dimension is N×N, are the ST-GCN network parameters, It represents the online update period. The online update period of ST-GCN network parameters is set to 0.5ms to ensure the adaptation to the rapidly changing channels. represents the k-th path fading coefficient, It represents the Doppler shift of the kth path;

[0061] By inserting a known pilot signal at the signal transmitting end, the receiving end measures the actual channel state based on the minimum mean square error formula according to the received pilot signal and the transmitted pilot signal. ;

[0062] Based on the calculated predicted channel state , the actual channel state and the predicted channel state are calculated using the root mean square error formula to obtain the prediction error;

[0063] A relative error value is obtained by calculating the difference between the predicted error and the preset standard error, a power adjustment value is obtained by multiplying the relative error value by the power adjustment factor, and the RF power is adjusted in real time according to the power adjustment value;

[0064] The RF power is adjusted in real time based on the predicted channel state. As the channel state changes, the receiver continuously measures the actual channel state, while the transmitter continuously calculates the predicted channel state and prediction error. The RF power is then adjusted in real time based on the power adjustment value to adapt to dynamic channel changes and ensure stable and reliable communication quality.

[0065] Step 2: Integrate the quantum measurement matrix with the deep residual network, introduce the residual attention mechanism to reconstruct abnormal channels, perform local optimization, solve the asymmetric Nash equilibrium, and reduce the algorithmic and computational complexity, dynamically control the RF power, suppress interference, and improve the performance of the power amplifier;

[0066] Based on the obtained quantum measurement matrix Fusion with the deep residual network ResNet;

[0067] By formula Fusion obtains mixed measurement matrix ,in, It is the measurement weight, which is used to balance the contribution of the quantum measurement matrix and the virtual matrix. By combining the high precision of the quantum measurement matrix and the learning ability of ResNet on data features, the compression performance of the measurement matrix is ​​improved.

[0068] It should be noted that the measurement weight is used to balance the contribution of the quantum measurement matrix and the virtual matrix. ResNet learns and generates data features from historical data, which is stable and reliable.

[0069] Based on the obtained mixed measurement matrix, the residual attention mechanism is introduced to reconstruct the abnormal channel, and the main path component is reconstructed first. Get the sparse reconstruction estimate of the channel state, where It represents the weighted L1 norm of the predicted channel state attention. By assigning different weights to different paths, the algorithm pays more attention to the main path component. It represents the sparsity adjustment factor, which is used to adjust the degree of sparsity constraint during the reconstruction process, reduce the number of iterations, and increase the speed of channel reconstruction. Channel reconstruction helps to more accurately control the RF power and reduce the bit error rate.

[0070] Based on channel reconstruction, in a massive MIMO system, 256 antennas are evenly divided into 32 hypergraph units. Local optimization is performed followed by global coordination. Through hypergraph partitioning and distributed optimization, computational complexity and latency are reduced.

[0071] Specifically, the 256-antenna system of the adjacent cell is evenly divided into 32 dynamic hypergraph units. Based on the asymmetric Nash equilibrium optimization goal, the optimization of inter-cell interference management and resource allocation is achieved. The power is dynamically adjusted according to the inter-cell interference situation to reduce the impact of interference on communication quality. Get the maximum transmission power of the dynamic hypergraph unit , the constraints are ,in, It represents the transmission power of the mth unit. It represents the signal-to-interference-and-noise ratio of the mth unit;

[0072] Dynamically partitioning hypergraph units decomposes complex multi-cell interference problems into multiple, relatively simple sub-problems for distributed processing, reducing computational complexity and coordination delays. The asymmetric Nash equilibrium optimization objective comprehensively considers the SINR and transmit power of each unit. By optimizing transmit power, effective inter-cell interference management and rational resource allocation are achieved while meeting total power constraints.

[0073] The gradient projection method is used to solve the asymmetric Nash equilibrium, and the gradient calculation is parallelized on the silicon-based photonic chip. Reduce the algorithm complexity, where represents the number of iterations, L is the Lagrangian function, It is the step size factor, which is used to adjust the step size of updating the solution vector P along the gradient direction. It uses the Armijo rule for adaptive adjustment to ensure the convergence and stability of the algorithm. It represents the gradient operator. When the gradient operator acts on the function L, a vector is obtained. Each component of the vector is the partial derivative of L with respect to each component of P.

[0074] After reducing the algorithm complexity, the Mach-Zehnder modulator (MZM) is used to separate the nonlinear components of the power amplifier, reducing the computational complexity of traditional digital predistortion (DPD).

[0075] Specifically, through the formula The polynomial order is compressed from 11 to 5, where It represents the optical domain frequency response correction matrix with a dimension of 128×128. It corrects the signal operating frequency response in the optical domain to compensate for the distortion of the signal operating frequency characteristics of the power amplifier. It represents the predistortion coefficient, which is used to adjust the degree of nonlinear processing of the input signal. It is solved by QR decomposition to ensure the accuracy and effectiveness of the coefficient.

[0076] Through the dual-loop controller, intelligent and dynamic control of RF power is achieved to balance communication quality, transmission power and interference suppression;

[0077] Adopting a dual-loop meta-reinforcement learning architecture, the outer meta-policy network uses Transformer-XL to model cross-scenario temporal features, while the inner execution network uses a distributed PPO algorithm to generate power policies in real time based on a reward function to suppress sudden interference. Combined with quantum-photonic joint digital twins, the model's ability to identify sudden interference patterns is enhanced, enabling it to respond to sudden interference and adjust power to ensure communication system reliability.

[0078] Entangled photon pairs are used to generate reverse noise waveforms, improve the adjacent channel leakage ratio, suppress out-of-band noise generated by the power amplifier, reduce interference to adjacent channels, improve spectrum utilization, and reduce signal distortion caused by noise, thereby improving the performance and efficiency of the power amplifier.

[0079] Step 3: Analyze the impact of environmental factors on system errors, perform joint calibration to correct errors, build a learning model and expand the dynamic range, determine the actual maximum transmit power, and allocate RF power;

[0080] Based on the ambient temperature obtained by the temperature sensor And the signal operating frequency obtained by spectrum analyzer , analyze the ambient temperature Sum signal operating frequency Impact on systematic errors;

[0081] Through joint calibration, errors caused by ambient temperature changes and different signal operating frequencies can be corrected in real time;

[0082] Specifically, pairs of input voltage data and output voltage data at different ambient temperatures are collected as training data sets;

[0083] Construct a neural network learning model, using input voltage data, ambient temperature, and memristor conductance matrix as input features, and embedding the bias vector of the ambient temperature compensation coefficient As an output target;

[0084] Based on the obtained bias vector of the embedded ambient temperature compensation coefficient , construct a 128×128 memristor array to store the nonlinear transfer function, and use the formula Get output voltage data , to achieve dynamic range expansion; among them, It represents the input voltage data. represents the memristor conductance matrix;

[0085] It should be noted that the memristor array stores and processes nonlinear functions, performs nonlinear compensation on the input signal, and expands the system's dynamic range to meet the requirements of high-power transmission and low-noise reception. The ambient temperature compensation coefficient in the bias vector reduces the impact of ambient temperature changes on system performance and improves system stability. The output voltage data, after compensation by the memristor array, has a wider dynamic range. The input voltage data represents the original signal that needs dynamic range expansion.

[0086] The memristor conductance matrix determines the nonlinear processing capability of the memristor array to the input signal, and the signal compensation is achieved by adjusting the conductance value; the bias vector Embedded ambient temperature compensation coefficient is used to adjust the DC bias of the output signal and compensate for the impact of ambient temperature changes on system performance;

[0087] Based on the obtained ambient temperature and signal operating frequency, a three-dimensional lookup table is constructed, and the formula Get the calibration error ,in, represents the calibration coefficient, which is updated online by the least squares method;

[0088] The obtained calibration error is compared with the preset maximum calibration allowable error to obtain a calibration relative value;

[0089] Multiply the obtained calibration relative value by the maximum transmit power to obtain the transmit calibration power difference;

[0090] Perform subtraction processing on the maximum transmit power and the transmit calibration power difference to obtain the actual maximum transmit power;

[0091] Allocate radio frequency power to channels within the dynamic hypergraph unit based on the obtained actual maximum transmit power;

[0092] The technical solution of the embodiments of the present invention is as follows: using a superconducting quantum bit array to capture phase fluctuations in the millimeter wave frequency band, and combining it with a graph convolutional network to calculate and predict the channel state, tracking the multipath fading and Doppler shift changes of the channel in real time, adjusting the radio frequency power according to the error between the actual and predicted channel states, ensuring the stability of the communication link in complex and changing channel environments, reducing signal fading and packet loss rate, and using hypergraph partitioning and asymmetric Nash equilibrium optimization to achieve inter-cell interference management and resource allocation. Dynamically adjust the power according to interference, reduce the impact of inter-cell interference on communication quality, and simultaneously improve spectrum utilization and system capacity; integrating the quantum measurement matrix with the deep residual network to improve the compression performance of the measurement matrix, combining the residual attention mechanism to reconstruct abnormal channels, prioritizing the main path component, reducing the number of channel reconstruction iterations, and improving the channel reconstruction speed; reducing the computational complexity of traditional digital pre-distortion, significantly reducing the computing resources and time required for system data processing, analyzing the impact of ambient temperature and signal operating frequency on system error, using joint calibration to correct the error in real time, and constructing a neural network learning model and memristor array to expand the system dynamic range, so that the system can adapt to different temperature environments and signal operating frequencies, and reducing the interference of environmental factors on system performance.

[0093] Example 2

[0094] like Figure 2 As shown, an embodiment of the present invention provides a 5G motherboard radio frequency power control system, which specifically includes the following modules:

[0095] Channel state perception module: This module uses a superconducting quantum bit array to capture millimeter-wave frequency phase fluctuations, constructs a virtual channel response based on quantum entanglement state measurements, and generates a channel topology map.

[0096] RF power adjustment module: This module calculates and predicts the channel state based on the channel topology map, calculates the prediction error based on the actual channel state, and adjusts the RF power.

[0097] Measurement matrix fusion module: It integrates the quantum measurement matrix with the deep residual network, introduces the residual attention mechanism to reconstruct abnormal channels, and performs local optimization;

[0098] Channel processing module: solves asymmetric Nash equilibrium and reduces algorithmic and computational complexity, dynamically controls RF power, suppresses interference, and improves power amplifier performance;

[0099] Environmental factor processing module: analyzes the impact of environmental factors on system errors and corrects errors through joint calibration;

[0100] Power allocation module: Builds a learning model and expands the dynamic range to determine the actual maximum transmit power and allocate RF power.

[0101] An embodiment of the present invention is described in detail above, but the content described is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention; the above formulas are all dimensionless and numerical calculations, and the formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field based on actual conditions and historical experience, and can be adjusted according to actual conditions; the above description is only a preferred embodiment of the present invention and is not used to limit the present invention. All equal changes and improvements made according to the scope of application of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A 5G motherboard radio frequency power control method, characterized in that: The following steps are involved: Step 1: Use a superconducting qubit array to capture millimeter-wave frequency phase fluctuations, construct a virtual channel response based on quantum entangled state measurements, and generate a channel topology map. Calculate and predict the channel state based on the channel topology map, calculate the prediction error based on the actual channel state, and adjust the RF power. Step 2: Integrate the quantum measurement matrix with the deep residual network, introduce the residual attention mechanism to reconstruct abnormal channels, perform local optimization, solve the asymmetric Nash equilibrium, and reduce the algorithmic and computational complexity, dynamically control the RF power, suppress interference, and improve the performance of the power amplifier; Step 3: Analyze the impact of environmental factors on system errors, perform joint calibration to correct errors, build a learning model and expand the dynamic range, determine the actual maximum transmit power, and allocate RF power.

2. The RF power control method of a 5G motherboard according to claim 1, characterized in that: The method for constructing the virtual channel response is: Based on quantum entangled state measurement, through quantum measurement matrix Jointly construct virtual channel responses with deep generative adversarial networks (GANs); The virtual channel response is compressed and observed through the quantum measurement matrix, and noise n is introduced. The compressed observation value is calculated, where and is the generator of GAN, and z is the latent space noise vector.

3. The RF power control method of a 5G motherboard according to claim 1, characterized in that: The method for generating the channel topology map is: Based on the generation of virtual channel responses, the channel multipath scatterers are mapped to dynamic graph nodes, and the edge weights are set as the scatterer mutual coupling coefficients , generate a channel topology map.

4. The RF power control method of a 5G motherboard according to claim 1, wherein: The method for calculating and predicting the channel state is: Based on the generated channel topology graph, it is processed through the graph convolution network and combined with the changes in multipath fading and Doppler frequency shift, according to the formula Calculate the predicted channel state ,in, is the channel topology at time t, and the adjacency matrix dimension is N×N, are the ST-GCN network parameters, Indicates the online update cycle. represents the k-th path fading coefficient, It represents the Doppler shift of the kth path.

5. The RF power control method of a 5G motherboard according to claim 1, characterized in that: The method for calculating the prediction error is: By inserting a known pilot signal at the signal transmitting end, the receiving end measures the actual channel state based on the minimum mean square error formula according to the received pilot signal and the transmitted pilot signal. ; Based on the calculated predicted channel state , the actual channel state and the predicted channel state are calculated using the root mean square error formula to obtain the prediction error.

6. The RF power control method of a 5G motherboard according to claim 1, characterized in that: The method for reconstructing the abnormal channel is: Based on the obtained quantum measurement matrix Fusion with the deep residual network ResNet; The mixed measurement matrix is ​​obtained by formula fusion ; Based on the obtained mixed measurement matrix, the residual attention mechanism is introduced to reconstruct the abnormal channel, and the main path component is reconstructed first. Get the sparse reconstruction estimate of the channel state, where It represents the predicted channel state attention weighted L1 norm, It means The square of the L2 norm, It represents the sparsity adjustment factor.

7. The RF power control method of a 5G motherboard according to claim 1, characterized in that: The local optimization method is: The 256-antenna system of the adjacent cell is evenly divided into 32 dynamic hypergraph units. Based on the asymmetric Nash equilibrium optimization objective, the power is dynamically adjusted according to the interference between cells. Get the maximum transmission power of the dynamic hypergraph unit , the constraints are ,in, It represents the transmission power of the mth unit. It represents the signal-to-interference-and-noise ratio of the mth unit; The gradient projection method is used to solve the asymmetric Nash equilibrium, and the gradient calculation is parallelized on the silicon-based photonic chip. Reduce the algorithm complexity, where represents the number of iterations, L is the Lagrangian function, is the step size factor, represents the gradient operator; The polynomial order is compressed from 11 to 5 by the formula, where: It represents the optical frequency response correction matrix. It represents the predistortion coefficient; The RF power is controlled by a dual-loop controller.

8. The RF power control method of a 5G motherboard according to claim 1, wherein: The method for extending the dynamic range is: Based on the obtained bias vector of the embedded ambient temperature compensation coefficient , construct a 128×128 memristor array to store the nonlinear transfer function, and use the formula Get output voltage data ,in, It represents the input voltage data. represents the memristor conductance matrix.

9. The RF power control method of a 5G motherboard according to claim 1, characterized in that: The method for obtaining the actual maximum transmit power is: Based on the obtained ambient temperature and signal operating frequency, a three-dimensional lookup table is constructed, and the formula Get the calibration error ,in, It represents the calibration factor; The obtained calibration error is compared with the preset maximum calibration allowable error to obtain a calibration relative value; Multiply the obtained calibration relative value by the maximum transmit power to obtain the transmit calibration power difference; Perform subtraction processing on the maximum transmit power and the transmit calibration power difference to obtain the actual maximum transmit power; Radio frequency power is allocated to channels within the dynamic hypergraph unit based on the obtained actual maximum transmit power.

10. A 5G motherboard radio frequency power control system, the system being used to execute the control method according to any one of claims 1 to 9, characterized in that: Includes the following modules: Channel state perception module: This module uses a superconducting quantum bit array to capture millimeter-wave frequency phase fluctuations, constructs a virtual channel response based on quantum entanglement state measurements, and generates a channel topology map. RF power adjustment module: This module calculates and predicts the channel state based on the channel topology map, calculates the prediction error based on the actual channel state, and adjusts the RF power. Measurement matrix fusion module: It integrates the quantum measurement matrix with the deep residual network, introduces the residual attention mechanism to reconstruct abnormal channels, and performs local optimization; Channel processing module: solves asymmetric Nash equilibrium and reduces algorithmic and computational complexity, dynamically controls RF power, suppresses interference, and improves power amplifier performance; Environmental factor processing module: analyzes the impact of environmental factors on system errors and corrects errors through joint calibration; Power allocation module: Builds a learning model and expands the dynamic range to determine the actual maximum transmit power and allocate RF power.

Images