A method and system for directional network adaptive radiation control

Abstract

The application provides a directional network adaptive radiation control method and system, relates to the technical field of wireless network control, and comprises the following steps: acquiring real-time high-frequency motion data, relative distance data and environmental interference intensity data of a wireless network; continuously quantifying a motion state according to the real-time high-frequency motion data to construct motion characteristic quantities; performing multi-step trajectory prediction processing according to the motion characteristic quantities to output a trajectory prediction sequence; performing prediction reliability evaluation according to the trajectory prediction sequence to generate a trajectory uncertainty index; generating a radiation control signal according to the motion characteristic quantities and the trajectory uncertainty index to obtain the radiation control signal; and performing beamforming and power adaptive processing according to the radiation control signal and the real-time relative distance data to obtain a final beam shape and power parameter combination. The application significantly improves the communication quality and anti-interference ability of the directional network in a complex environment.

Description

Technical Field

[0001] This invention relates to the field of wireless network control technology, and more specifically, to a method and system for adaptive radiation control of directional networks. Background Technology

[0002] With the rapid development of wireless network collaborative operation technology, the demand for highly reliable and low-latency communication is increasing. Directional self-organizing networks (DSMs), due to their high directivity, anti-interference capabilities, and spectral efficiency, have become an important foundation for supporting swarm intelligence. However, in complex dynamic environments, the high-speed movement of network nodes, frequent topology changes, and strong uncertainty in electromagnetic interference pose significant challenges to traditional radiation control methods. Existing technologies typically employ beamforming and power control strategies based on fixed rules or static optimization algorithms. While these methods can achieve certain performance under stable conditions, they struggle to adapt to the highly dynamic and uncertain nature of network node movement and the electromagnetic environment. Their main drawbacks include: a lack of deep perception and fusion capabilities for the overall motion state and trajectory prediction reliability of nodes; an inability to achieve coordinated optimization of radiation parameters and three-dimensional spatial situation; and low computational efficiency and delayed adaptive adjustment in scenarios with strong interference and high-speed movement, leading to a significant decrease in communication link reliability.

[0003] Based on the shortcomings of the existing technologies, there is an urgent need for a directional network adaptive radiation control method and system. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive radiation control method for directional networks to improve the aforementioned problems. To achieve this objective, the technical solution adopted by this invention is as follows:

[0005] Firstly, this application provides a method for adaptive radiation control of directional networks, including:

[0006] Acquire real-time high-frequency motion data of each node in the wireless network under a cooperative motion environment, real-time relative distance data between each node, and environmental interference intensity data;

[0007] Based on the real-time high-frequency motion data, the motion state is continuously quantized to construct motion feature quantities that characterize the overall dynamic properties of the node.

[0008] Multi-step trajectory prediction is performed based on the motion features. A spatiotemporal graph convolutional network is used to fuse the attention mechanism, and the future spatiotemporal trajectory of each node is predicted by combining the historical trajectory sequence, and the trajectory prediction sequence is output.

[0009] The reliability of the prediction is evaluated based on the predicted trajectory sequence. By quantifying the trajectory prediction results, a trajectory uncertainty index that changes over time is generated.

[0010] Radiation control signals are generated based on the motion characteristic quantities and the trajectory uncertainty index, resulting in a radiation control signal jointly modulated in the frequency and spatial domains.

[0011] Beamforming and power adaptive processing are performed based on the radiation control signal and the real-time relative distance data. The spatial channel response is reconstructed using sparse representation theory. Combined with the environmental interference intensity data, the beam pointing and transmit power parameters are dynamically optimized through reinforcement learning strategy to obtain the final beam shape and power parameter combination.

[0012] Secondly, this application also provides a directional network adaptive radiation control system, comprising:

[0013] The acquisition module is used to acquire real-time high-frequency motion data of each node in the wireless network under a cooperative motion environment, real-time relative distance data between each node, and environmental interference intensity data.

[0014] The quantization module is used to continuously quantize the motion state based on the real-time high-frequency motion data and construct motion feature quantities that characterize the overall dynamic properties of the node.

[0015] The prediction module is used to perform multi-step trajectory prediction processing based on the motion features. It adopts a spatiotemporal graph convolutional network with attention mechanism and combines historical trajectory sequences to predict the future spatiotemporal trajectory of each node, and outputs a trajectory prediction sequence.

[0016] The evaluation module is used to evaluate the reliability of the prediction based on the trajectory prediction sequence and generate a trajectory uncertainty index that changes over time by quantifying the trajectory prediction results.

[0017] The generation module is used to generate a radiation control signal based on the motion characteristic quantity and the trajectory uncertainty index, so as to obtain a radiation control signal jointly modulated in the frequency domain and the spatial domain.

[0018] The output module is used to perform beamforming and power adaptive processing based on the radiation control signal and the real-time relative distance data, reconstruct the spatial channel response using sparse representation theory, and dynamically optimize the beam pointing and transmit power parameters through reinforcement learning strategy in combination with the environmental interference intensity data to obtain the final beam shape and power parameter combination.

[0019] The beneficial effects of this invention are as follows:

[0020] This invention acquires node motion, relative distance, and interference data in real time, and generates spatiotemporal situational awareness results through continuous quantization of motion state, multi-step trajectory prediction, and reliability assessment. Ultimately, it drives beamforming and power control based on sparse channel reconstruction and reinforcement learning optimization, achieving adaptive and collaborative optimization of radiation parameters for three-dimensional spatial dynamics and communication reliability. This significantly improves the communication quality and anti-interference capability of directional networks in complex environments. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of a directional network adaptive radiation control method as described in an embodiment of the present invention;

[0023] Figure 2 This is a schematic diagram of a directional network adaptive radiation control system structure as described in an embodiment of the present invention;

[0024] Figure 3 This is a schematic diagram of the structure of a directional network adaptive radiation control device as described in an embodiment of the present invention.

[0025] The diagram is labeled as follows: 800, a directional network adaptive radiation control device; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component; 901, acquisition module; 902, quantization module; 903, prediction module; 904, evaluation module; 905, generation module; 906, output module. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] Example 1:

[0029] This embodiment provides a method for adaptive radiation control of directional networks.

[0030] See Figure 1 The figure shows that the method includes steps S100 to S600.

[0031] Step S100: Obtain real-time high-frequency motion data of each node in the wireless network under the cooperative motion environment, real-time relative distance data between each node, and environmental interference intensity data.

[0032] The core of step S100 lies in providing a comprehensive and real-time data foundation for the entire adaptive control process. The acquired high-frequency motion parameters, relative distances, and environmental interference intensity of the wireless network nodes accurately correspond to the dynamic topology changes and complex electromagnetic environment challenges faced by the nodes during cooperative motion in the three-dimensional spatial domain. These multi-dimensional data collectively constitute the raw input for sensing the system's situation. Specifically, the real-time high-frequency motion data covers the three-dimensional spatial position, velocity vector, acceleration vector, and angular velocity vector of each node. This data accurately describes the instantaneous motion attitude and dynamic characteristics of a single network node in the air. The real-time relative distance data between nodes reflects the geometric topology of the network and is a key geometric input for analyzing spatial relationships between nodes, determining link availability, and calculating beam pointing. The environmental interference intensity data characterizes the noise and interference levels of the electromagnetic environment within the operating frequency band and is an important basis for evaluating channel quality and achieving anti-interference communication. In practice, this data is continuously collected by multi-source sensors (such as GNSS / RTK modules, IMU inertial measurement units, UWB radio frequency sensing modules, etc.) mounted on various wireless network nodes, and distributed and shared in real time through a high-speed self-organizing communication data link built by the network, thereby providing a unified, synchronous and comprehensive three-dimensional spatial situational awareness foundation for all subsequent processing steps.

[0033] Step S200: Perform continuous quantization of motion state based on real-time high-frequency motion data to construct motion feature quantities that characterize the overall dynamic properties of the node;

[0034] Step S200 aims to extract features that characterize the essence of the collaborative behavior of network nodes from the underlying high-frequency motion data. Through the continuous quantification process of motion state, the independent motion information of a single node is transformed into feature quantities that reflect the overall motion consistency and dynamic evolution law of network nodes, providing a deep state representation for subsequent prediction and decision-making.

[0035] Step S300: Perform multi-step trajectory prediction processing based on motion features, use a spatiotemporal graph convolutional network to fuse attention mechanism, and combine historical trajectory sequences to predict the future spatiotemporal trajectory of each node, and output trajectory prediction sequence.

[0036] Step S300 focuses on making forward-looking inferences about future motion patterns. Based on the extracted motion features, it uses a spatiotemporal modeling method with an attention fusion mechanism to process the spatiotemporal correlations and geometric constraints between nodes, thereby deducing future trajectory sequences. This is a prerequisite for achieving forward-looking radiation control.

[0037] Step S400: Perform a prediction reliability assessment based on the trajectory prediction sequence. Quantify the trajectory prediction results to generate a trajectory uncertainty index that changes over time.

[0038] The core of step S400 is to quantitatively evaluate the credibility of the prediction result itself. By calculating the uncertainty index of trajectory prediction, the system is provided with the ability to perceive its own prediction reliability, thereby reasonably balancing prediction information and real-time status in decision-making.

[0039] Step S500: Generate a radiation control signal based on motion characteristic quantities and trajectory uncertainty index to obtain a radiation control signal jointly modulated in the frequency domain and spatial domain;

[0040] Step S500 serves as a bridge between "perception" and "prediction" information and actual "control" commands. By fusing characteristic quantities reflecting the current motion state with uncertainty indices characterizing prediction reliability, it generates radiation control signals that simultaneously modulate frequency and spatial domain characteristics, laying the theoretical foundation for adaptive beamforming.

[0041] Step S600: Based on the radiation control signal and real-time relative distance data, beamforming and power adaptive processing are performed. The spatial channel response is reconstructed using sparse representation theory. Combined with environmental interference intensity data, the beam pointing and transmit power parameters are dynamically optimized through reinforcement learning strategy to obtain the final beam shape and power parameter combination.

[0042] Step S600 finally completes the mapping from control signals to physical parameters. Based on the reconstructed spatial channel response, and combined with real-time environmental interference and radiation control requirements, it dynamically optimizes beam shape and transmit power through an autonomous learning strategy, thereby achieving precise adaptation of radiation parameters to three-dimensional spatial dynamics and communication reliability.

[0043] Further, step S200 includes steps S210 to S230.

[0044] Step S210: Perform motion state encoding processing based on real-time high-frequency motion data, and perform joint dimensionality reduction and feature extraction on multi-dimensional motion data through nonlinear manifold learning to obtain the motion encoding vector of each node;

[0045] Step S220: Perform node motion coupling analysis based on motion encoding vectors. Using Lie group representation theory, map the independent motion encoding of each node to the rigid body motion constraint space of the node to obtain the motion constraint factor characterizing the overall motion consistency of the node.

[0046] Step S230: Based on the motion coding vector and motion constraint factor, perform dynamic feature fusion. By constructing the continuous coherence features of node motion, analyze the spatiotemporal correlation of wireless network nodes in three-dimensional space and the topological invariance of motion patterns to obtain motion feature quantities that characterize the overall dynamic characteristics of the nodes.

[0047] Specifically, step S210 first utilizes nonlinear manifold learning technology to process real-time high-frequency motion data. This technology can reveal the inherent low-dimensional essential structure of high-dimensional motion parameters (such as position, velocity, acceleration, etc.). By capturing their nonlinear distribution characteristics, it encodes the complex motion patterns of each node into compact and information-rich motion encoding vectors, thereby achieving deep extraction and dimensionality reduction representation of individual motion features. The motion encoding formula is:

[0048] h i =F Θ (x i );

[0049] Among them, h i x represents the motion encoding vector of node i; i F represents the original high-dimensional motion state vector of node i, containing the set of the node's three-dimensional position, velocity vector, acceleration vector, and angular velocity vector at a specific moment, comprehensively describing the instantaneous kinematic state of a single network node in the three-dimensional spatial domain; Θ This represents a nonlinear manifold learning mapping function with parameter Θ, where parameter Θ learns the intrinsic low-dimensional manifold structure in high-dimensional motion data through training.

[0050] Building upon this, step S220 focuses on resolving the coordination issue between individuals and the whole during motion. It utilizes Lie group representation theory to characterize the mathematical properties of rigid body motion in three-dimensional space, mapping the independent motion codes of each node to a unified group representation space. This allows for the analysis of the consistency between individual motion and group motion constraints. The generated motion constraint factor effectively reflects the structural stability in cooperative motion tasks such as formation maintenance and formation changes. The formula for the motion constraint factor is:

[0051]

[0052] Where c represents the cluster motion constraint factor, a scalar value that is the average of the motion deviations of all nodes. The smaller the value, the more consistent the motion of each node with the overall coordinated motion, and the more stable the network structure; the larger the value, the more dispersed or uncoordinated the node motions. i represents the node index; N represents the total number of nodes; G is the Lie group representation mapping function, which encodes the motion vector h of each node. i Mapped to its corresponding Lie group element, used to represent the rigid body motion of the node in three-dimensional space; ‖·‖ F denoted by Frobenius norm; T is the reference group element for the overall motion of the nodes, representing the expected or average rigid body motion state of the entire wireless network nodes in three-dimensional space at the current moment.

[0053] Finally, step S230 integrates the aforementioned motion encoding vectors and motion constraint factors, employing the continuous cohomology theory derived from computational topology for dynamic feature fusion. This theory effectively analyzes the birth and persistence of topological features (such as connectivity and ring structures) during the spatiotemporal evolution of complex motion trajectories, thereby revealing the spatiotemporal correlation patterns of wireless network nodes in the three-dimensional spatial domain and their inherent stability against local disturbances. Ultimately, it extracts motion feature quantities that comprehensively characterize the overall dynamic properties of the nodes (possessing both geometric relationships and topological robustness), providing a deep situational understanding for subsequent prediction and control. The dynamic feature fusion formula is:

[0054]

[0055] Wherein, P0 is the persistent graph, containing lifecycle information of point cloud topological features; PH k Let represent the k-dimensional continuous cohomology computation function, which performs topological analysis on the point cloud composed of the motion codes of all nodes within a time window, calculating the birth and death times of its k-dimensional topological features (connected components, loop structures); m represents the overall dynamic characteristics of the nodes, i.e., the final fused feature vector; M represents the feature fusion function, which receives the motion code vector h of all nodes. iUsing node motion constraint factor c and persistent graph P0 as inputs, a machine learning model fuses this information to capture features at the geometric, algebraic, and topological levels; {h i (t-τ:t)} represents the motion coding sequence of node i within the most recent time window τ.

[0056] Further, step S300 includes steps S310 to S330.

[0057] Step S310: Perform dynamic topological evolution processing based on motion feature quantities and historical trajectory sequences. Encode the geometric constraints of interactions between network nodes through spatiotemporal graph convolution to obtain evolutionary features that characterize spatial association and temporal dependence.

[0058] Step S320: Perform continuous dynamics deduction based on evolutionary characteristics, and use a neural network driven by differential equations to simulate the motion trajectory of wireless network nodes in the three-dimensional spatial domain under the influence of aerodynamics, and obtain multi-step state prediction tensors.

[0059] Step S330: Based on the evolutionary characteristics and multi-step state prediction tensor, perform trajectory optimization and generation processing, integrate the attention mechanism to explicitly model nodes in the network whose motion state change rate exceeds a preset threshold, and output the future spatiotemporal trajectory prediction sequence.

[0060] Specifically, step S310 first models the dynamic interactions between network nodes using a spatiotemporal graph convolutional network based on motion feature quantities and historical trajectory data. This network simultaneously captures the geometric constraints between nodes that change over time and space, thereby encoding evolutionary features that simultaneously contain spatial correlation and temporal dependence. Step S320 further constructs and solves a continuous-time dynamic system using neural differential equations based on the above evolutionary features, thereby simulating the real motion process of wireless network nodes affected by aerodynamic effects in a three-dimensional spatial domain, deducing future multi-step motion states, and forming a state prediction tensor. Step S330 integrates the evolutionary features and the multi-step state prediction tensor, introduces an attention mechanism to enhance the modeling of nodes with drastic changes in motion state (such as nodes whose acceleration or angular velocity exceeds a threshold), coordinates the overall trajectory and local maneuvers through optimized generation methods, and finally outputs a future spatiotemporal trajectory prediction sequence that conforms to physical constraints and cooperative motion laws.

[0061] Further, step S400 includes steps S410 to S430.

[0062] Step S410: Perform instantaneous uncertainty quantification processing on the trajectory prediction sequence, calculate the probability distribution of the predicted trajectory in three-dimensional space using the nonparametric kernel density estimation method, and obtain the prediction confidence distribution at each time point;

[0063] Step S420: Perform time-series uncertainty propagation processing based on the predicted confidence distribution, analyze the correlation characteristics of the confidence distribution evolution over time using a time-varying autoregressive model, and obtain the uncertainty propagation function;

[0064] Step S430: Based on the prediction confidence distribution and uncertainty propagation function, perform comprehensive processing, and integrate instantaneous and temporal uncertainty measures through information entropy theory to generate a trajectory uncertainty index that changes with time.

[0065] Specifically, step S410 addresses the reliability issue of trajectory prediction in three-dimensional space by employing a non-parametric kernel density estimation method to process the predicted trajectory sequence. This method does not require prior assumptions about the probability distribution and can directly characterize the possible distribution of network nodes at future times based on the prediction data itself, thereby calculating the probability density distribution of predicted points at each time point and obtaining a confidence distribution reflecting the instantaneous prediction accuracy. Step S420 addresses the evolution of uncertainty over time by using a time-varying autoregressive model to analyze the correlation between the aforementioned confidence distribution at different times. This model can capture the temporal characteristics of the dynamic propagation of uncertainty, thereby establishing a mathematical relationship for inferring future uncertainty states from the current uncertainty state and obtaining an uncertainty propagation function characterizing the evolution of uncertainty. Step S430, to comprehensively evaluate the overall reliability of the prediction, integrates the instantaneous confidence distribution and the temporal propagation function based on information entropy theory. Information entropy can quantify the degree of uncertainty of the entire prediction system. By integrating instantaneous uncertainty and temporal propagation uncertainty, a comprehensive uncertainty index characterizing the reliability of trajectory prediction over time is finally generated, providing a key credibility basis for subsequent decision-making.

[0066] Further, step S500 includes steps S510 to S530.

[0067] Step S510: Perform three-dimensional spatial situation fusion processing based on motion feature quantities and trajectory uncertainty index. Learn the nonlinear mapping relationship between the motion mode of wireless network nodes and communication reliability in the three-dimensional spatial domain through a generative flow model to obtain the spatial situation awareness vector.

[0068] Step S520: Based on the spatial situational awareness vector, frequency domain anti-interference signal generation processing is performed. A baseband signal with time-varying spectral characteristics is constructed through a nonlinear dynamic system, and the situational awareness vector is encoded into anti-interference frequency domain modulation coefficients to obtain an adaptive frequency domain modulation signal.

[0069] Step S530: Perform spatial beamforming processing based on the spatial situation awareness vector and the adaptive frequency domain modulation signal. Map the three-dimensional spatial situation onto the complex spherical space using Klein geometry to generate a frequency domain and spatial domain joint modulation radiation control signal.

[0070] Specifically, step S510 addresses the complexity of the relationship between the dynamic motion of wireless network nodes in three-dimensional space and communication reliability by employing a generative flow model to deeply fuse motion characteristic quantities and trajectory uncertainty indices. This model learns high-dimensional nonlinear mapping relationships through invertible transformations, encoding the node's motion patterns and implicit channel reliability features into a unified spatial situational awareness vector, thereby achieving a joint representation of the potential relationship between node spatial location, motion trends, and communication quality. Step S520, based on the real-time environmental state characterized by the spatial situational awareness vector, utilizes a nonlinear dynamic system to generate a baseband signal with time-varying chaotic characteristics to cope with complex electromagnetic interference. Furthermore, the vector representing the real-time situation is encoded into frequency domain modulation coefficients, dynamically adjusting the signal's spectral structure to form an adaptive frequency domain modulation signal with environmental awareness capabilities. Step S530, in order to achieve accurate directivity and coverage adaptability of the radiated signal in three-dimensional space, introduces Klein geometry theory to map the spatial situational awareness vector onto a complex spherical space to characterize the geometric relationship and signal propagation constraints of wireless network nodes in the three-dimensional spatial domain. By combining this geometric mapping relationship with the adaptive frequency domain modulation signal, the direction and shape of the beam in space are optimized in a coordinated manner, and finally a joint modulation radiation control signal with both frequency domain anti-interference characteristics and spatial domain directional capability is generated.

[0071] Further, step S600 includes steps S610 to S630.

[0072] Step S610: Perform sparse spatial channel reconstruction processing based on real-time relative distance data and environmental interference intensity data. By reconstructing the sparse channel response under the dynamic topology of the wireless network, high-dimensional spatial channel state information is obtained.

[0073] Step S620: Perform distributed beamforming processing based on the high-dimensional space channel state information and radiation control signal. Calculate the beamforming vector that satisfies the three-dimensional spatial geometric constraints and radiation control requirements using a distributed optimization algorithm to obtain the initial control scheme.

[0074] Step S630: Based on the initial control scheme and environmental interference intensity data, anti-interference adaptive optimization processing is performed. The beam main lobe direction and null depth are jointly optimized in the continuous action space through deep reinforcement learning algorithm to generate the beam shape and power parameter combination that drives the physical radio frequency link.

[0075] Specifically, step S610 addresses the high-dimensionality and sparsity of channel response under dynamic topology in wireless networks by utilizing compressed sensing theory to process real-time relative distance and environmental interference intensity data. A sparse reconstruction algorithm is then used to recover the spatial channel response contaminated by noise and interference, thereby accurately reconstructing high-dimensional channel state information reflecting the true propagation characteristics between nodes in three-dimensional space. Step S620, based on the reconstructed high-dimensional channel state information and the requirements of radiation control signals, employs a distributed optimization algorithm to solve for the beamforming vector. This algorithm can simultaneously satisfy overall three-dimensional spatial geometric constraints (such as beam pointing accuracy and sidelobe suppression) and radiation control indicators while each node processes information locally, thus obtaining a preliminary beam weight and power allocation scheme that conforms to spatial and signal constraints. Step S630 addresses the problem of strong interference sources dynamically appearing in complex electromagnetic environments. It explores strategies in the continuous action space through a deep reinforcement learning algorithm. This algorithm can autonomously learn how to jointly adjust the beam main lobe direction to align with the desired node, while forming a deep null in the direction of interference to suppress the interference. Finally, it generates a beam shape and transmit power parameter combination that can directly drive the actual radio frequency link and has both optimal spatial directivity and anti-interference capability.

[0076] Further, step S630 includes steps S631 to S633.

[0077] Step S631: Based on the initial control scheme and environmental interference intensity data, perform main lobe direction optimization processing. Search for beamforming parameters that maximize main lobe gain in the three-dimensional continuous action space using the near-end strategy optimization algorithm to obtain the optimal main lobe pointing parameters.

[0078] Step S632: Perform null depth optimization processing based on the optimal main lobe pointing parameters and environmental interference intensity data. Generate adaptive nulls in the interference direction based on the preset adversarial learning framework to obtain the depth null configuration parameters for targeted interference sources.

[0079] Step S633: Perform joint optimization of beam parameters based on the optimal main lobe pointing parameters and depth null configuration parameters, and use a collaborative filtering algorithm to coordinate the adjustment of the main lobe and null to generate a combination of beam shape and power parameters.

[0080] Step S631 first uses a near-end policy optimization algorithm to intelligently search for beamforming parameters in a three-dimensional continuous action space based on the initial control scheme and environmental interference intensity data. This algorithm efficiently explores the parameter space through a policy gradient method to directly maximize the gain of the main lobe in the desired direction, thereby accurately obtaining the optimal main lobe pointing parameters. Step S632 then uses an adversarial learning framework to simulate the dynamic game between the interference environment and the beamformer based on the determined optimal main lobe pointing parameters and real-time interference data. This framework adaptively generates deep nulls in known strong interference directions through adversarial training, effectively suppressing interference signals and ultimately obtaining deep null configuration parameters for specific interference sources. Step S633 finally integrates the optimal main lobe pointing parameters and deep null configuration parameters, using a collaborative filtering algorithm to analyze the inherent correlation and potential conflicts between the main lobe and null control. Through collaborative adjustment, it achieves the optimal balance between the main lobe gain and the null depth, ultimately generating a beamform and power parameter combination that simultaneously ensures a high-quality communication link and strong anti-interference capability.

[0081] The joint cost function in the joint optimization process is expressed as:

[0082]

[0083] Among them, J(w,P) t ) represents the joint cost function; w represents the beamforming weight vector; P t Indicates the transmit power; θ d ,φ d ,r represents the desired three-dimensional spherical coordinates of the target node, where ,a(θ) represents the azimuth, elevation, and distance of the target node, respectively; d ,φ d ,r) represents the three-dimensional spatial channel steering vector; θ I,k ,φ I,k ,r I,k Let P represent the three-dimensional spherical coordinates of the k-th interference source; K represents the total number of detected interference sources; α, β, and γ are weights representing the importance of main lobe performance, interference suppression, and power efficiency in the current optimization objective, respectively; max This indicates the maximum permissible transmit power of the radio frequency link of a wireless network node; H represents the conjugate transpose.

[0084] The dynamic weight adjustment function in the coordinated adjustment process is expressed as:

[0085]

[0086] Where, β (n) This represents the weight coefficient of the zero-trap depth term at the nth optimization time. I represents the intensity of the k-th interference source perceived at time n; thThis represents the interference intensity threshold, used to determine whether the current electromagnetic environment is severe; β min β max η represents the minimum and maximum values ​​of the weight β; η represents the slope factor of the adjustment curve, used to control the sensitivity of the weight to changes in interference intensity; δ represents the offset of the adjustment curve, used to set the critical point of interference intensity at which the weight begins to increase significantly.

[0087] Example 2:

[0088] like Figure 2 As shown, this embodiment provides a directional network adaptive radiation control system, the system including:

[0089] The acquisition module 901 is used to acquire real-time high-frequency motion data of each node in the wireless network under the cooperative motion environment, real-time relative distance data between each node, and environmental interference intensity data.

[0090] The quantization module 902 is used to continuously quantize the motion state based on real-time high-frequency motion data and construct motion feature quantities that characterize the overall dynamic properties of the node.

[0091] The prediction module 903 is used to perform multi-step trajectory prediction processing based on motion features. It adopts a spatiotemporal graph convolutional network with attention mechanism and combines historical trajectory sequences to predict the future spatiotemporal trajectory of each node, and outputs a trajectory prediction sequence.

[0092] Evaluation module 904 is used to evaluate the reliability of the prediction based on the trajectory prediction sequence. By quantifying the trajectory prediction results, it generates a trajectory uncertainty index that changes over time.

[0093] The generation module 905 is used to generate a radiation control signal based on motion characteristic quantities and trajectory uncertainty index, so as to obtain a radiation control signal jointly modulated in the frequency domain and spatial domain.

[0094] The output module 906 is used to perform beamforming and power adaptive processing based on radiation control signals and real-time relative distance data. It reconstructs the spatial channel response using sparse representation theory, combines environmental interference intensity data, and dynamically optimizes beam pointing and transmit power parameters through reinforcement learning strategies to obtain the final beam shape and power parameter combination.

[0095] In one specific embodiment of this application, the quantization module 902 includes:

[0096] The first quantization unit is used to encode motion state based on real-time high-frequency motion data. It performs joint dimensionality reduction and feature extraction on multi-dimensional motion data through nonlinear manifold learning to obtain the motion encoding vector of each node.

[0097] The second quantization unit is used to perform node motion coupling analysis based on the motion encoding vector. By using Lie group representation theory, the independent motion encoding of each node is mapped to the rigid body motion constraint space of the node to obtain the motion constraint factor characterizing the overall motion consistency of the node.

[0098] The third quantization unit is used to perform dynamic feature fusion based on the motion coding vector and motion constraint factor. By constructing the continuous coherence features of node motion, it analyzes the spatiotemporal correlation of wireless network nodes in three-dimensional space and the topological invariance of motion patterns, and obtains motion feature quantities that characterize the overall dynamic properties of the nodes.

[0099] In one specific embodiment of this application, the prediction module 903 includes:

[0100] The first prediction unit is used to perform dynamic topological evolution processing based on motion feature quantities and historical trajectory sequences. It encodes the geometric constraints of the interaction between network nodes through spatiotemporal graph convolution to obtain evolutionary features that characterize spatial correlation and temporal dependence.

[0101] The second prediction unit is used to perform continuous dynamic deduction based on evolutionary characteristics. It uses a neural network driven by differential equations to simulate the motion trajectory of wireless network nodes in the three-dimensional spatial domain under the influence of aerodynamics, and obtains a multi-step state prediction tensor.

[0102] The third prediction unit is used to optimize and generate trajectories based on evolutionary features and multi-step state prediction tensors. It integrates an attention mechanism to explicitly model nodes in the network whose motion state change rate exceeds a preset threshold, and outputs a future spatiotemporal trajectory prediction sequence.

[0103] Example 3:

[0104] Corresponding to the above method embodiments, this embodiment also provides a directional network adaptive radiation control device. The directional network adaptive radiation control device described below and the directional network adaptive radiation control method described above can be referred to in correspondence.

[0105] Figure 3 This is a block diagram illustrating a directional network adaptive radiation control device 800 according to an exemplary embodiment. Figure 3 As shown, the directional network adaptive radiation control device 800 may include a processor 801 and a memory 802. The directional network adaptive radiation control device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0106] The processor 801 controls the overall operation of the directional network adaptive radiation control device 800 to complete all or part of the steps in the directional network adaptive radiation control method described above. The memory 802 stores various types of data to support the operation of the directional network adaptive radiation control device 800. This data may include, for example, instructions for any application or method operating on the directional network adaptive radiation control device 800, and application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the directional network adaptive radiation control device 800 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, or an NFC module.

[0107] In an exemplary embodiment, a directional network adaptive radiation control device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the aforementioned directional network adaptive radiation control method.

[0108] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the directional network adaptive radiation control method described above. For example, the computer-readable storage medium may be the memory 802 including program instructions described above, which may be executed by a processor 801 of a directional network adaptive radiation control device 800 to complete the directional network adaptive radiation control method described above.

[0109] 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 changes 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 adaptive radiation control of directional networks, characterized in that, include: Acquire real-time high-frequency motion data of each node in the wireless network under a cooperative motion environment, real-time relative distance data between each node, and environmental interference intensity data; Based on the real-time high-frequency motion data, the motion state is continuously quantized to construct motion feature quantities that characterize the overall dynamic properties of network nodes. Multi-step trajectory prediction is performed based on the motion features. A spatiotemporal graph convolutional network is used to fuse the attention mechanism, and the future spatiotemporal trajectory of each node is predicted by combining the historical trajectory sequence, and the trajectory prediction sequence is output. Based on the trajectory prediction sequence, a prediction reliability assessment is performed, and the trajectory prediction results are quantified to generate a trajectory uncertainty index that changes over time. Radiation control signals are generated based on the motion characteristic quantities and the trajectory uncertainty index, resulting in a radiation control signal jointly modulated in the frequency and spatial domains. Beamforming and power adaptive processing are performed based on the radiation control signal and the real-time relative distance data. The spatial channel response is reconstructed using sparse representation theory. Combined with the environmental interference intensity data, the beam pointing and transmit power parameters are dynamically optimized through reinforcement learning strategy to obtain the final beam shape and power parameter combination. The generation of radiation control signals based on the motion characteristic quantities and the trajectory uncertainty index includes: Based on the motion feature quantity and the trajectory uncertainty index, a three-dimensional spatial situation fusion process is performed. The nonlinear mapping relationship between the motion mode of wireless network nodes and communication reliability in the three-dimensional spatial domain is learned through a generative flow model to obtain a spatial situation awareness vector. Based on the spatial situational awareness vector, frequency domain anti-interference signal generation processing is performed. A baseband signal with time-varying spectral characteristics is constructed through a nonlinear dynamic system, and the situational awareness vector is encoded into anti-interference frequency domain modulation coefficients to obtain an adaptive frequency domain modulation signal. Spatial beamforming is performed based on the spatial situation awareness vector and the adaptive frequency domain modulation signal. The three-dimensional spatial situation is mapped to a complex spherical space through Klein geometry to generate a frequency domain and spatial domain joint modulation radiation control signal.

2. The directional network adaptive radiation control method according to claim 1, characterized in that, Continuous quantization of motion state based on the real-time high-frequency motion data includes: Based on the real-time high-frequency motion data, motion state encoding processing is performed, and multi-dimensional motion data is jointly reduced in dimensionality and extracted using nonlinear manifold learning to obtain the motion encoding vector of each node. Based on the motion encoding vector, node motion coupling analysis is performed. The independent motion encoding of each node is mapped to the rigid body motion constraint space of the node through Lie group representation theory, so as to obtain the motion constraint factor characterizing the overall motion consistency of the node. By performing dynamic feature fusion based on the motion coding vector and the motion constraint factor, and constructing continuous coherence features of node motion, the spatiotemporal correlation of wireless network nodes in three-dimensional space and the topological invariance of motion patterns are analyzed to obtain motion feature quantities that characterize the overall dynamic properties.

3. The directional network adaptive radiation control method according to claim 1, characterized in that, Multi-step trajectory prediction processing based on the motion feature quantities includes: Dynamic topological evolution processing is performed based on the motion feature quantities and historical trajectory sequences. Geometric constraints of interactions between network nodes are encoded through spatiotemporal graph convolution to obtain evolutionary features that characterize spatial correlation and temporal dependence. Based on the evolutionary characteristics, continuous dynamics deduction is performed, and a neural network driven by differential equations is used to simulate the motion trajectory of a wireless network node in a three-dimensional spatial domain under the influence of aerodynamics, so as to obtain a multi-step state prediction tensor. Trajectory optimization and generation are performed based on the evolutionary features and the multi-step state prediction tensor. An attention mechanism is integrated to explicitly model nodes in the network whose motion state change rate exceeds a preset threshold, and a future spatiotemporal trajectory prediction sequence is output.

4. The directional network adaptive radiation control method according to claim 1, characterized in that, The prediction reliability assessment based on the predicted trajectory sequence includes: The instantaneous uncertainty quantification process is performed on the trajectory prediction sequence, and the probability distribution of the predicted trajectory in three-dimensional space is calculated by the non-parametric kernel density estimation method to obtain the prediction confidence distribution at each time point. Based on the predicted confidence distribution, time-series uncertainty propagation processing is performed, and the correlation characteristics of the confidence distribution evolving over time are analyzed using a time-varying autoregressive model to obtain the uncertainty propagation function; Based on the predicted confidence distribution and the uncertainty propagation function, the instantaneous and temporal uncertainty measures are integrated through information entropy theory to generate a trajectory uncertainty index that varies with time.

5. The directional network adaptive radiation control method according to claim 1, characterized in that, Beamforming and power adaptive processing are performed based on the radiation control signal and the real-time relative distance data, including: Based on the real-time relative distance data and the environmental interference intensity data, sparse spatial channel reconstruction processing is performed. By reconstructing the sparse channel response under the dynamic topology of the wireless network, high-dimensional spatial channel state information is obtained. Distributed beamforming processing is performed based on the high-dimensional spatial channel state information and the radiation control signal. A distributed optimization algorithm is used to calculate the beamforming vector that meets the three-dimensional spatial geometric constraints and radiation control requirements to obtain the initial control scheme. Based on the initial control scheme and the environmental interference intensity data, anti-interference adaptive optimization processing is performed. The beam main lobe direction and null depth are jointly optimized in the continuous action space through a deep reinforcement learning algorithm to generate a combination of beam shape and power parameters to drive the physical radio frequency link.

6. The directional network adaptive radiation control method according to claim 5, characterized in that, Based on the initial control scheme and the environmental interference intensity data, anti-interference adaptive optimization processing is performed. A deep reinforcement learning algorithm is used to jointly optimize the beam main lobe direction and null depth in the continuous action space, generating a beamform and power parameter combination to drive the physical RF link, including: Based on the initial control scheme and the environmental interference intensity data, the main lobe direction is optimized. The beamforming parameters that maximize the main lobe gain are searched in the three-dimensional continuous motion space by the near-end strategy optimization algorithm to obtain the optimal main lobe pointing parameters. Null depth optimization is performed based on the optimal main lobe pointing parameters and the environmental interference intensity data. An adaptive null is generated in the interference direction based on a preset adversarial learning framework to obtain the depth null configuration parameters for targeted interference sources. The beam parameters are jointly optimized based on the optimal main lobe pointing parameters and the depth null configuration parameters. The main lobe and null are then coordinated and adjusted using a collaborative filtering algorithm to generate a combination of beam shape and power parameters.

7. A directional network adaptive radiation control system, characterized in that, include: The acquisition module is used to acquire real-time high-frequency motion data of each node in the wireless network under a cooperative motion environment, real-time relative distance data between each node, and environmental interference intensity data. The quantization module is used to continuously quantize the motion state based on the real-time high-frequency motion data and construct motion feature quantities that characterize the overall dynamic properties of the node. The prediction module is used to perform multi-step trajectory prediction processing based on the motion features. It adopts a spatiotemporal graph convolutional network with attention mechanism and combines historical trajectory sequences to predict the future spatiotemporal trajectory of each node, and outputs a trajectory prediction sequence. The evaluation module is used to evaluate the reliability of the prediction based on the trajectory prediction sequence and generate a trajectory uncertainty index that changes over time by quantifying the trajectory prediction results. The generation module is used to generate a radiation control signal based on the motion characteristic quantity and the trajectory uncertainty index, so as to obtain a radiation control signal jointly modulated in the frequency domain and the spatial domain. The output module is used to perform beamforming and power adaptive processing based on the radiation control signal and the real-time relative distance data, reconstruct the spatial channel response using sparse representation theory, and dynamically optimize the beam pointing and transmit power parameters through reinforcement learning strategy in combination with the environmental interference intensity data to obtain the final beam shape and power parameter combination. The generation of radiation control signals based on the motion characteristic quantities and the trajectory uncertainty index includes: Based on the motion feature quantity and the trajectory uncertainty index, a three-dimensional spatial situation fusion process is performed. The nonlinear mapping relationship between the motion mode of wireless network nodes and communication reliability in the three-dimensional spatial domain is learned through a generative flow model to obtain a spatial situation awareness vector. Based on the spatial situational awareness vector, frequency domain anti-interference signal generation processing is performed. A baseband signal with time-varying spectral characteristics is constructed through a nonlinear dynamic system, and the situational awareness vector is encoded into anti-interference frequency domain modulation coefficients to obtain an adaptive frequency domain modulation signal. Spatial beamforming is performed based on the spatial situation awareness vector and the adaptive frequency domain modulation signal. The three-dimensional spatial situation is mapped to a complex spherical space through Klein geometry to generate a frequency domain and spatial domain joint modulation radiation control signal.

8. The directional network adaptive radiation control system according to claim 7, characterized in that, The quantization module includes: The first quantization unit is used to perform motion state encoding processing based on the real-time high-frequency motion data, and to perform joint dimensionality reduction and feature extraction on multi-dimensional motion data through nonlinear manifold learning to obtain the motion encoding vector of each node. The second quantization unit is used to perform node motion coupling analysis based on the motion encoding vector. By using Lie group representation theory, the independent motion encoding of each node is mapped to the rigid body motion constraint space of the node to obtain the motion constraint factor characterizing the overall motion consistency of the node. The third quantization unit is used to perform dynamic feature fusion based on the motion coding vector and the motion constraint factor. By constructing the continuous coherence features of node motion, it analyzes the spatiotemporal correlation of wireless network nodes in three-dimensional space and the topological invariance of motion patterns, and obtains motion feature quantities that characterize the overall dynamic properties of the nodes.

9. The directional network adaptive radiation control system according to claim 7, characterized in that, The prediction module includes: The first prediction unit is used to perform dynamic topological evolution processing based on the motion feature quantities and historical trajectory sequences, and to encode the geometric constraints of the interaction between network nodes through spatiotemporal graph convolution to obtain evolutionary features that characterize spatial correlation and temporal dependence. The second prediction unit is used to perform continuous dynamic deduction processing based on the evolution characteristics, and to simulate the motion trajectory of network nodes in the three-dimensional space under the influence of aerodynamics by using a neural network driven by differential equations to obtain a multi-step state prediction tensor. The third prediction unit is used to perform trajectory optimization and generation processing based on the evolutionary features and the multi-step state prediction tensor, and to explicitly model nodes in the network whose motion state change rate exceeds a preset threshold by integrating the attention mechanism, and output the future spatiotemporal trajectory prediction sequence.

Images